[llvm] ab111e9 - Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation""
Usman Nadeem via llvm-commits
llvm-commits at lists.llvm.org
Fri Sep 10 18:23:49 PDT 2021
Author: Usman Nadeem
Date: 2021-09-10T18:11:24-07:00
New Revision: ab111e982fd183ba084605ef4219810f159b4198
URL: https://github.com/llvm/llvm-project/commit/ab111e982fd183ba084605ef4219810f159b4198
DIFF: https://github.com/llvm/llvm-project/commit/ab111e982fd183ba084605ef4219810f159b4198.diff
LOG: Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation""
This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b.
Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67
after fixing the failing tests.
Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
Added:
Modified:
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_abd.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acge.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acgt.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acle.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_aclt.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_add.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_and.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_asr.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfdot.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalb.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalt.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bic.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpeq.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpge.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpgt.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmple.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmplt.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpne.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpuo.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_div.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_divr.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dot.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup-bfloat.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dupq.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_eor.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsl.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsr.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mad.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_max.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_maxnm.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_min.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_minnm.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mla.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mls.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_msb.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mul.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulh.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulx.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmad.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmla.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmls.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmsb.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_orr.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qadd.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qsub.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_scale.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sub.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_subr.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sudot.c
clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_usdot.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_aba.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlbt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bcax.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bdep.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bext.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bgrp.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl1n.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl2n.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eor3.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eorbt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eortb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hadd.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsub.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsubr.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_nbsl.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmul.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb_128.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt_128.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qadd.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalbt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslbt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmulh.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlah.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlsh.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmulh.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrshl.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qshl.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsub.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsubr.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rhadd.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rshl.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sqadd.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublbt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subltb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwb.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwt.c
clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_uqadd.c
llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll
llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-dup.ll
Removed:
################################################################################
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_abd.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_abd.c
index 5b5b34d2edca2..25903e049070b 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_abd.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_abd.c
@@ -428,14 +428,16 @@ svuint64_t test_svabd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svabd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svabd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svabd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svabd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svabd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svabd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svabd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svabd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svabd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svabd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svabd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svabd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svabd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svabd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svabd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svabd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svabd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svabd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svabd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svabd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svabd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svabd_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svabd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svabd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svabd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svabd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svabd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svabd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svabd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svabd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svabd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svabd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svabd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svabd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svabd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svabd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svabd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svabd_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svabd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svabd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svabd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svabd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svabd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svabd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svabd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svabd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svabd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svabd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svabd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svabd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svabd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svabd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svabd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1011,6 +1059,7 @@ svfloat16_t test_svabd_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2)
return SVE_ACLE_FUNC(svabd,_f16,_x,)(pg, op1, op2);
}
+//
// CHECK-LABEL: @test_svabd_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
@@ -1048,7 +1097,8 @@ svfloat64_t test_svabd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svabd_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1106,8 @@ svfloat64_t test_svabd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1120,8 @@ svfloat16_t test_svabd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svabd_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1129,8 @@ svfloat16_t test_svabd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1143,8 @@ svfloat32_t test_svabd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svabd_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1152,8 @@ svfloat32_t test_svabd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1166,16 @@ svfloat64_t test_svabd_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svabd_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1187,16 @@ svfloat16_t test_svabd_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svabd_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1208,16 @@ svfloat32_t test_svabd_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svabd_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1229,16 @@ svfloat64_t test_svabd_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svabd_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1250,16 @@ svfloat16_t test_svabd_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svabd_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1271,16 @@ svfloat32_t test_svabd_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svabd_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acge.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acge.c
index 791599bd74a37..4342520f9fd7a 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acge.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acge.c
@@ -74,7 +74,8 @@ svbool_t test_svacge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svacge_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -82,7 +83,8 @@ svbool_t test_svacge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z17test_svacge_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -95,7 +97,8 @@ svbool_t test_svacge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svacge_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -103,7 +106,8 @@ svbool_t test_svacge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z17test_svacge_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acgt.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acgt.c
index b1297fe1cfc27..b2e0b16793546 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acgt.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acgt.c
@@ -74,7 +74,8 @@ svbool_t test_svacgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svacgt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -82,7 +83,8 @@ svbool_t test_svacgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z17test_svacgt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -95,7 +97,8 @@ svbool_t test_svacgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svacgt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -103,7 +106,8 @@ svbool_t test_svacgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z17test_svacgt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acle.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acle.c
index b1af88e1c225f..f432984e86645 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acle.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acle.c
@@ -74,7 +74,8 @@ svbool_t test_svacle_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svacle_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -82,7 +83,8 @@ svbool_t test_svacle_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z17test_svacle_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -95,7 +97,8 @@ svbool_t test_svacle_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svacle_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -103,7 +106,8 @@ svbool_t test_svacle_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z17test_svacle_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_aclt.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_aclt.c
index 4e5740b498d51..ddeb983f6e462 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_aclt.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_aclt.c
@@ -74,7 +74,8 @@ svbool_t test_svaclt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svaclt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -82,7 +83,8 @@ svbool_t test_svaclt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z17test_svaclt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -95,7 +97,8 @@ svbool_t test_svaclt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svaclt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -103,7 +106,8 @@ svbool_t test_svaclt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z17test_svaclt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_add.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_add.c
index da95476e2d15c..675260e513132 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_add.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_add.c
@@ -428,14 +428,16 @@ svuint64_t test_svadd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svadd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svadd_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svadd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svadd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svadd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svadd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svadd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svadd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svadd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svadd_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svadd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svadd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svadd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svadd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svadd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svadd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svadd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svadd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svadd_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svadd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svadd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svadd_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svadd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svadd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svadd_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svadd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svadd_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svadd_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svadd_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svadd_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svadd_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svadd_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svadd_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svadd_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svadd_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svadd_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svadd_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svadd_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_and.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_and.c
index 33d39cfc2be1f..90320f7ba5a02 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_and.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_and.c
@@ -428,14 +428,16 @@ svuint64_t test_svand_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svand_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svand_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svand_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svand_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svand_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svand_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svand_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svand_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svand_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svand_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svand_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svand_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svand_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svand_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svand_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svand_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svand_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svand_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svand_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svand_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svand_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svand_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svand_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svand_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svand_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svand_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svand_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svand_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svand_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svand_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svand_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svand_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svand_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svand_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svand_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svand_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svand_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svand_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svand_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svand_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svand_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svand_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svand_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svand_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svand_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svand_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svand_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svand_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svand_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svand_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svand_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svand_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svand_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svand_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svand_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svand_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svand_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svand_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svand_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svand_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svand_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_asr.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_asr.c
index b4bb06a44c39f..3d0a6a2e61043 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_asr.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_asr.c
@@ -223,7 +223,8 @@ svint64_t test_svasr_s64_x(svbool_t pg, svint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svasr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -231,7 +232,8 @@ svint64_t test_svasr_s64_x(svbool_t pg, svint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_zu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -244,14 +246,16 @@ svint64_t test_svasr_n_s64_z(svbool_t pg, svint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_mu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -263,14 +267,16 @@ svint64_t test_svasr_n_s64_m(svbool_t pg, svint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_xu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -434,14 +440,16 @@ svint32_t test_svasr_wide_s32_x(svbool_t pg, svint32_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svasr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_zu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -454,7 +462,8 @@ svint8_t test_svasr_n_s8_z(svbool_t pg, svint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svasr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -462,7 +471,8 @@ svint8_t test_svasr_n_s8_z(svbool_t pg, svint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_zu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -475,7 +485,8 @@ svint16_t test_svasr_n_s16_z(svbool_t pg, svint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svasr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -483,7 +494,8 @@ svint16_t test_svasr_n_s16_z(svbool_t pg, svint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_zu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -495,13 +507,15 @@ svint32_t test_svasr_n_s32_z(svbool_t pg, svint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svasr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_mu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -513,14 +527,16 @@ svint8_t test_svasr_n_s8_m(svbool_t pg, svint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svasr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_mu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -532,14 +548,16 @@ svint16_t test_svasr_n_s16_m(svbool_t pg, svint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svasr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_mu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -550,13 +568,15 @@ svint32_t test_svasr_n_s32_m(svbool_t pg, svint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svasr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_xu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -568,14 +588,16 @@ svint8_t test_svasr_n_s8_x(svbool_t pg, svint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svasr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_xu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -587,14 +609,16 @@ svint16_t test_svasr_n_s16_x(svbool_t pg, svint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svasr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_xu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -605,13 +629,15 @@ svint32_t test_svasr_n_s32_x(svbool_t pg, svint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_mu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -623,14 +649,16 @@ svint8_t test_svasr_wide_n_s8_m(svbool_t pg, svint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_mu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -642,14 +670,16 @@ svint16_t test_svasr_wide_n_s16_m(svbool_t pg, svint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_mu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -660,14 +690,16 @@ svint32_t test_svasr_wide_n_s32_m(svbool_t pg, svint32_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_zu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -680,7 +712,8 @@ svint8_t test_svasr_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -688,7 +721,8 @@ svint8_t test_svasr_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_zu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -701,7 +735,8 @@ svint16_t test_svasr_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -709,7 +744,8 @@ svint16_t test_svasr_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_zu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -721,13 +757,15 @@ svint32_t test_svasr_wide_n_s32_z(svbool_t pg, svint32_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_xu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -739,14 +777,16 @@ svint8_t test_svasr_wide_n_s8_x(svbool_t pg, svint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_xu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -758,14 +798,16 @@ svint16_t test_svasr_wide_n_s16_x(svbool_t pg, svint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svasr_wide_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_xu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfdot.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfdot.c
index deab191390c00..5d49890f56b2c 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfdot.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfdot.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -57,13 +59,15 @@ svfloat32_t test_bfdot_lane_3_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t z)
// CHECK-LABEL: @test_bfdot_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_bfdot_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalb.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalb.c
index c5733eb812e7f..502190ec4d965 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalb.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -57,13 +59,15 @@ svfloat32_t test_bfmlalb_lane_7_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t
// CHECK-LABEL: @test_bfmlalb_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_bfmlalb_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalt.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalt.c
index 6fdd3206fee6b..428985c79d184 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalt.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +bf16 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -57,13 +59,15 @@ svfloat32_t test_bfmlalt_lane_7_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t
// CHECK-LABEL: @test_bfmlalt_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_bfmlalt_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bic.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bic.c
index e391d41fe795b..e60921b1d688e 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bic.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bic.c
@@ -428,14 +428,16 @@ svuint64_t test_svbic_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svbic_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svbic_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svbic_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svbic_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svbic_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svbic_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svbic_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svbic_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svbic_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svbic_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svbic_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svbic_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svbic_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svbic_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svbic_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svbic_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svbic_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svbic_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svbic_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svbic_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svbic_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svbic_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svbic_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svbic_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svbic_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svbic_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svbic_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svbic_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svbic_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svbic_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svbic_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svbic_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svbic_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svbic_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svbic_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svbic_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svbic_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svbic_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svbic_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svbic_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svbic_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svbic_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svbic_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svbic_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svbic_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svbic_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svbic_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svbic_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svbic_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svbic_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svbic_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svbic_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svbic_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpeq.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpeq.c
index cedd2738b7ef7..5821ba47fbe25 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpeq.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpeq.c
@@ -161,7 +161,8 @@ svbool_t test_svcmpeq_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmpeq_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -169,7 +170,8 @@ svbool_t test_svcmpeq_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -182,7 +184,8 @@ svbool_t test_svcmpeq_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpeq_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -190,7 +193,8 @@ svbool_t test_svcmpeq_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -255,13 +259,15 @@ svbool_t test_svcmpeq_wide_s32(svbool_t pg, svint32_t op1, svint64_t op2)
// CHECK-LABEL: @test_svcmpeq_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpeq_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -273,7 +279,8 @@ svbool_t test_svcmpeq_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svcmpeq_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -281,7 +288,8 @@ svbool_t test_svcmpeq_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -294,7 +302,8 @@ svbool_t test_svcmpeq_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svcmpeq_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -302,7 +311,8 @@ svbool_t test_svcmpeq_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -314,13 +324,15 @@ svbool_t test_svcmpeq_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svcmpeq_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpeq_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -332,7 +344,8 @@ svbool_t test_svcmpeq_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svcmpeq_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -340,7 +353,8 @@ svbool_t test_svcmpeq_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -353,7 +367,8 @@ svbool_t test_svcmpeq_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svcmpeq_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -361,7 +376,8 @@ svbool_t test_svcmpeq_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -431,7 +447,8 @@ svbool_t test_svcmpeq_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svcmpeq_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -439,7 +456,8 @@ svbool_t test_svcmpeq_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -452,7 +470,8 @@ svbool_t test_svcmpeq_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svcmpeq_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -460,7 +479,8 @@ svbool_t test_svcmpeq_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -473,7 +493,8 @@ svbool_t test_svcmpeq_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svcmpeq_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -481,7 +502,8 @@ svbool_t test_svcmpeq_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -493,13 +515,15 @@ svbool_t test_svcmpeq_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svcmpeq_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpeq_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -511,7 +535,8 @@ svbool_t test_svcmpeq_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpeq_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -519,7 +544,8 @@ svbool_t test_svcmpeq_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpeq_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -532,7 +558,8 @@ svbool_t test_svcmpeq_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpeq_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -540,7 +567,8 @@ svbool_t test_svcmpeq_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpeq_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpge.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpge.c
index 26ead4897ea17..56d020a30fe6c 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpge.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpge.c
@@ -161,7 +161,8 @@ svbool_t test_svcmpge_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmpge_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -169,7 +170,8 @@ svbool_t test_svcmpge_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -182,7 +184,8 @@ svbool_t test_svcmpge_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpge_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -190,7 +193,8 @@ svbool_t test_svcmpge_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -308,13 +312,15 @@ svbool_t test_svcmpge_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmpge_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpge_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -326,7 +332,8 @@ svbool_t test_svcmpge_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svcmpge_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -334,7 +341,8 @@ svbool_t test_svcmpge_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -347,7 +355,8 @@ svbool_t test_svcmpge_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svcmpge_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -355,7 +364,8 @@ svbool_t test_svcmpge_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -367,13 +377,15 @@ svbool_t test_svcmpge_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svcmpge_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpge_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -385,7 +397,8 @@ svbool_t test_svcmpge_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svcmpge_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -393,7 +406,8 @@ svbool_t test_svcmpge_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -406,7 +420,8 @@ svbool_t test_svcmpge_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svcmpge_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -414,7 +429,8 @@ svbool_t test_svcmpge_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -484,7 +500,8 @@ svbool_t test_svcmpge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svcmpge_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -492,7 +509,8 @@ svbool_t test_svcmpge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -505,7 +523,8 @@ svbool_t test_svcmpge_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svcmpge_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -513,7 +532,8 @@ svbool_t test_svcmpge_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -526,7 +546,8 @@ svbool_t test_svcmpge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svcmpge_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -534,7 +555,8 @@ svbool_t test_svcmpge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -546,13 +568,15 @@ svbool_t test_svcmpge_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svcmpge_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpge_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -564,7 +588,8 @@ svbool_t test_svcmpge_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpge_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -572,7 +597,8 @@ svbool_t test_svcmpge_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -585,7 +611,8 @@ svbool_t test_svcmpge_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpge_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -593,7 +620,8 @@ svbool_t test_svcmpge_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -605,13 +633,15 @@ svbool_t test_svcmpge_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpge_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpge_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -623,7 +653,8 @@ svbool_t test_svcmpge_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmpge_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -631,7 +662,8 @@ svbool_t test_svcmpge_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -644,7 +676,8 @@ svbool_t test_svcmpge_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmpge_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -652,7 +685,8 @@ svbool_t test_svcmpge_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpgt.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpgt.c
index a5881cb129ab1..91abb5750ba20 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpgt.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpgt.c
@@ -161,7 +161,8 @@ svbool_t test_svcmpgt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmpgt_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -169,7 +170,8 @@ svbool_t test_svcmpgt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -182,7 +184,8 @@ svbool_t test_svcmpgt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpgt_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -190,7 +193,8 @@ svbool_t test_svcmpgt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -308,13 +312,15 @@ svbool_t test_svcmpgt_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmpgt_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpgt_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -326,7 +332,8 @@ svbool_t test_svcmpgt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svcmpgt_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -334,7 +341,8 @@ svbool_t test_svcmpgt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -347,7 +355,8 @@ svbool_t test_svcmpgt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svcmpgt_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -355,7 +364,8 @@ svbool_t test_svcmpgt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -367,13 +377,15 @@ svbool_t test_svcmpgt_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svcmpgt_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpgt_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -385,7 +397,8 @@ svbool_t test_svcmpgt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svcmpgt_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -393,7 +406,8 @@ svbool_t test_svcmpgt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -406,7 +420,8 @@ svbool_t test_svcmpgt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svcmpgt_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -414,7 +429,8 @@ svbool_t test_svcmpgt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -484,7 +500,8 @@ svbool_t test_svcmpgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svcmpgt_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -492,7 +509,8 @@ svbool_t test_svcmpgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -505,7 +523,8 @@ svbool_t test_svcmpgt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svcmpgt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -513,7 +532,8 @@ svbool_t test_svcmpgt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -526,7 +546,8 @@ svbool_t test_svcmpgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svcmpgt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -534,7 +555,8 @@ svbool_t test_svcmpgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -546,13 +568,15 @@ svbool_t test_svcmpgt_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svcmpgt_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpgt_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -564,7 +588,8 @@ svbool_t test_svcmpgt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpgt_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -572,7 +597,8 @@ svbool_t test_svcmpgt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -585,7 +611,8 @@ svbool_t test_svcmpgt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpgt_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -593,7 +620,8 @@ svbool_t test_svcmpgt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -605,13 +633,15 @@ svbool_t test_svcmpgt_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpgt_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpgt_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -623,7 +653,8 @@ svbool_t test_svcmpgt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmpgt_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -631,7 +662,8 @@ svbool_t test_svcmpgt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -644,7 +676,8 @@ svbool_t test_svcmpgt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmpgt_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -652,7 +685,8 @@ svbool_t test_svcmpgt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmple.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmple.c
index 694b7c7a93afc..20d27b227fd68 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmple.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmple.c
@@ -161,7 +161,8 @@ svbool_t test_svcmple_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmple_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -169,7 +170,8 @@ svbool_t test_svcmple_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -182,7 +184,8 @@ svbool_t test_svcmple_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmple_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -190,7 +193,8 @@ svbool_t test_svcmple_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -308,13 +312,15 @@ svbool_t test_svcmple_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmple_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmple_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -326,7 +332,8 @@ svbool_t test_svcmple_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svcmple_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -334,7 +341,8 @@ svbool_t test_svcmple_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -347,7 +355,8 @@ svbool_t test_svcmple_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svcmple_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -355,7 +364,8 @@ svbool_t test_svcmple_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -367,13 +377,15 @@ svbool_t test_svcmple_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svcmple_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmple_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -385,7 +397,8 @@ svbool_t test_svcmple_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svcmple_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -393,7 +406,8 @@ svbool_t test_svcmple_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -406,7 +420,8 @@ svbool_t test_svcmple_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svcmple_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -414,7 +429,8 @@ svbool_t test_svcmple_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -484,7 +500,8 @@ svbool_t test_svcmple_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svcmple_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -492,7 +509,8 @@ svbool_t test_svcmple_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -505,7 +523,8 @@ svbool_t test_svcmple_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svcmple_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -513,7 +532,8 @@ svbool_t test_svcmple_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -526,7 +546,8 @@ svbool_t test_svcmple_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svcmple_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -534,7 +555,8 @@ svbool_t test_svcmple_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmple_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -546,13 +568,15 @@ svbool_t test_svcmple_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svcmple_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmple.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmple_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmple.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -564,7 +588,8 @@ svbool_t test_svcmple_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmple_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmple.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -572,7 +597,8 @@ svbool_t test_svcmple_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmple.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -585,7 +611,8 @@ svbool_t test_svcmple_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmple_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmple.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -593,7 +620,8 @@ svbool_t test_svcmple_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmple.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -605,13 +633,15 @@ svbool_t test_svcmple_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmple_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpls.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmple_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpls.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -623,7 +653,8 @@ svbool_t test_svcmple_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmple_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpls.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -631,7 +662,8 @@ svbool_t test_svcmple_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpls.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -644,7 +676,8 @@ svbool_t test_svcmple_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmple_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpls.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -652,7 +685,8 @@ svbool_t test_svcmple_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpls.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmplt.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmplt.c
index 1ef15579ade27..f534235634ba0 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmplt.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmplt.c
@@ -161,7 +161,8 @@ svbool_t test_svcmplt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmplt_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -169,7 +170,8 @@ svbool_t test_svcmplt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -182,7 +184,8 @@ svbool_t test_svcmplt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmplt_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -190,7 +193,8 @@ svbool_t test_svcmplt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -308,13 +312,15 @@ svbool_t test_svcmplt_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmplt_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmplt_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -326,7 +332,8 @@ svbool_t test_svcmplt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svcmplt_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -334,7 +341,8 @@ svbool_t test_svcmplt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -347,7 +355,8 @@ svbool_t test_svcmplt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svcmplt_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -355,7 +364,8 @@ svbool_t test_svcmplt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -367,13 +377,15 @@ svbool_t test_svcmplt_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svcmplt_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmplt_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -385,7 +397,8 @@ svbool_t test_svcmplt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svcmplt_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -393,7 +406,8 @@ svbool_t test_svcmplt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -406,7 +420,8 @@ svbool_t test_svcmplt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svcmplt_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -414,7 +429,8 @@ svbool_t test_svcmplt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -484,7 +500,8 @@ svbool_t test_svcmplt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svcmplt_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -492,7 +509,8 @@ svbool_t test_svcmplt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -505,7 +523,8 @@ svbool_t test_svcmplt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svcmplt_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -513,7 +532,8 @@ svbool_t test_svcmplt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -526,7 +546,8 @@ svbool_t test_svcmplt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svcmplt_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -534,7 +555,8 @@ svbool_t test_svcmplt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -546,13 +568,15 @@ svbool_t test_svcmplt_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svcmplt_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmplt_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -564,7 +588,8 @@ svbool_t test_svcmplt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmplt_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -572,7 +597,8 @@ svbool_t test_svcmplt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -585,7 +611,8 @@ svbool_t test_svcmplt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmplt_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -593,7 +620,8 @@ svbool_t test_svcmplt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -605,13 +633,15 @@ svbool_t test_svcmplt_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmplt_wide_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplo.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmplt_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplo.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -623,7 +653,8 @@ svbool_t test_svcmplt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmplt_wide_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplo.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -631,7 +662,8 @@ svbool_t test_svcmplt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplo.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -644,7 +676,8 @@ svbool_t test_svcmplt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svcmplt_wide_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplo.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -652,7 +685,8 @@ svbool_t test_svcmplt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplo.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpne.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpne.c
index 42bad6790261d..bcc2f16373539 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpne.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpne.c
@@ -161,7 +161,8 @@ svbool_t test_svcmpne_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svcmpne_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -169,7 +170,8 @@ svbool_t test_svcmpne_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -182,7 +184,8 @@ svbool_t test_svcmpne_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpne_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -190,7 +193,8 @@ svbool_t test_svcmpne_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -255,13 +259,15 @@ svbool_t test_svcmpne_wide_s32(svbool_t pg, svint32_t op1, svint64_t op2)
// CHECK-LABEL: @test_svcmpne_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpne_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -273,7 +279,8 @@ svbool_t test_svcmpne_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svcmpne_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -281,7 +288,8 @@ svbool_t test_svcmpne_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -294,7 +302,8 @@ svbool_t test_svcmpne_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svcmpne_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -302,7 +311,8 @@ svbool_t test_svcmpne_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -314,13 +324,15 @@ svbool_t test_svcmpne_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svcmpne_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpne_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -332,7 +344,8 @@ svbool_t test_svcmpne_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svcmpne_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -340,7 +353,8 @@ svbool_t test_svcmpne_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -353,7 +367,8 @@ svbool_t test_svcmpne_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svcmpne_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -361,7 +376,8 @@ svbool_t test_svcmpne_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -431,7 +447,8 @@ svbool_t test_svcmpne_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svcmpne_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -439,7 +456,8 @@ svbool_t test_svcmpne_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -452,7 +470,8 @@ svbool_t test_svcmpne_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svcmpne_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -460,7 +479,8 @@ svbool_t test_svcmpne_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -473,7 +493,8 @@ svbool_t test_svcmpne_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svcmpne_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -481,7 +502,8 @@ svbool_t test_svcmpne_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -493,13 +515,15 @@ svbool_t test_svcmpne_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svcmpne_wide_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpne_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -511,7 +535,8 @@ svbool_t test_svcmpne_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpne_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -519,7 +544,8 @@ svbool_t test_svcmpne_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -532,7 +558,8 @@ svbool_t test_svcmpne_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CHECK-LABEL: @test_svcmpne_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -540,7 +567,8 @@ svbool_t test_svcmpne_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpuo.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpuo.c
index e8742c168fb0d..2593ff07bb557 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpuo.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpuo.c
@@ -74,7 +74,8 @@ svbool_t test_svcmpuo_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svcmpuo_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -82,7 +83,8 @@ svbool_t test_svcmpuo_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -95,7 +97,8 @@ svbool_t test_svcmpuo_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svcmpuo_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -103,7 +106,8 @@ svbool_t test_svcmpuo_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -116,7 +120,8 @@ svbool_t test_svcmpuo_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svcmpuo_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
@@ -124,7 +129,8 @@ svbool_t test_svcmpuo_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_div.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_div.c
index 4f384678c2431..ab28142b91e17 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_div.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_div.c
@@ -229,7 +229,8 @@ svuint64_t test_svdiv_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svdiv_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -237,7 +238,8 @@ svuint64_t test_svdiv_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -250,7 +252,8 @@ svint32_t test_svdiv_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svdiv_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -258,7 +261,8 @@ svint32_t test_svdiv_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -271,7 +275,8 @@ svint64_t test_svdiv_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svdiv_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -279,7 +284,8 @@ svint64_t test_svdiv_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -292,7 +298,8 @@ svuint32_t test_svdiv_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svdiv_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -300,7 +307,8 @@ svuint32_t test_svdiv_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -313,14 +321,16 @@ svuint64_t test_svdiv_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svdiv_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -332,14 +342,16 @@ svint32_t test_svdiv_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svdiv_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -351,14 +363,16 @@ svint64_t test_svdiv_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svdiv_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -370,14 +384,16 @@ svuint32_t test_svdiv_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svdiv_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -389,14 +405,16 @@ svuint64_t test_svdiv_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svdiv_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -408,14 +426,16 @@ svint32_t test_svdiv_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svdiv_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -427,14 +447,16 @@ svint64_t test_svdiv_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svdiv_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -446,14 +468,16 @@ svuint32_t test_svdiv_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svdiv_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -624,7 +648,8 @@ svfloat64_t test_svdiv_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svdiv_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -632,7 +657,8 @@ svfloat64_t test_svdiv_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -645,7 +671,8 @@ svfloat16_t test_svdiv_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svdiv_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -653,7 +680,8 @@ svfloat16_t test_svdiv_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -666,7 +694,8 @@ svfloat32_t test_svdiv_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svdiv_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -674,7 +703,8 @@ svfloat32_t test_svdiv_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -687,14 +717,16 @@ svfloat64_t test_svdiv_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svdiv_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -706,14 +738,16 @@ svfloat16_t test_svdiv_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svdiv_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -725,14 +759,16 @@ svfloat32_t test_svdiv_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svdiv_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -744,14 +780,16 @@ svfloat64_t test_svdiv_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svdiv_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -763,14 +801,16 @@ svfloat16_t test_svdiv_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svdiv_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -782,14 +822,16 @@ svfloat32_t test_svdiv_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svdiv_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_divr.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_divr.c
index b703082f0b0ef..451fa06e8a569 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_divr.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_divr.c
@@ -229,7 +229,8 @@ svuint64_t test_svdivr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svdivr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -237,7 +238,8 @@ svuint64_t test_svdivr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -250,7 +252,8 @@ svint32_t test_svdivr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svdivr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -258,7 +261,8 @@ svint32_t test_svdivr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -271,7 +275,8 @@ svint64_t test_svdivr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svdivr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -279,7 +284,8 @@ svint64_t test_svdivr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -292,7 +298,8 @@ svuint32_t test_svdivr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svdivr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -300,7 +307,8 @@ svuint32_t test_svdivr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -313,14 +321,16 @@ svuint64_t test_svdivr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svdivr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -332,14 +342,16 @@ svint32_t test_svdivr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svdivr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -351,14 +363,16 @@ svint64_t test_svdivr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svdivr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -370,14 +384,16 @@ svuint32_t test_svdivr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svdivr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -389,14 +405,16 @@ svuint64_t test_svdivr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svdivr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -408,14 +426,16 @@ svint32_t test_svdivr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svdivr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -427,14 +447,16 @@ svint64_t test_svdivr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svdivr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -446,14 +468,16 @@ svuint32_t test_svdivr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svdivr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -624,7 +648,8 @@ svfloat64_t test_svdivr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svdivr_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -632,7 +657,8 @@ svfloat64_t test_svdivr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -645,7 +671,8 @@ svfloat16_t test_svdivr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svdivr_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -653,7 +680,8 @@ svfloat16_t test_svdivr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -666,7 +694,8 @@ svfloat32_t test_svdivr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svdivr_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -674,7 +703,8 @@ svfloat32_t test_svdivr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -687,14 +717,16 @@ svfloat64_t test_svdivr_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svdivr_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -706,14 +738,16 @@ svfloat16_t test_svdivr_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svdivr_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -725,14 +759,16 @@ svfloat32_t test_svdivr_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svdivr_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -744,14 +780,16 @@ svfloat64_t test_svdivr_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svdivr_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -763,14 +801,16 @@ svfloat16_t test_svdivr_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svdivr_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -782,14 +822,16 @@ svfloat32_t test_svdivr_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svdivr_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dot.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dot.c
index b08c6be1529a1..9600d006aa41c 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dot.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dot.c
@@ -76,13 +76,15 @@ svuint64_t test_svdot_u64(svuint64_t op1, svuint16_t op2, svuint16_t op3)
// CHECK-LABEL: @test_svdot_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_s32u11__SVInt32_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -93,13 +95,15 @@ svint32_t test_svdot_n_s32(svint32_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svdot_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_s64u11__SVInt64_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -110,13 +114,15 @@ svint64_t test_svdot_n_s64(svint64_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svdot_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_u32u12__SVUint32_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -127,13 +133,15 @@ svuint32_t test_svdot_n_u32(svuint32_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svdot_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdot_n_u64u12__SVUint64_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup-bfloat.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup-bfloat.c
index b914fb6fc2cd9..146718d278149 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup-bfloat.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup-bfloat.c
@@ -17,12 +17,14 @@
// CHECK-LABEL: @test_svdup_n_bf16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z17test_svdup_n_bf16u6__bf16(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP0]]
//
svbfloat16_t test_svdup_n_bf16(bfloat16_t op) {
@@ -83,13 +85,15 @@ svbfloat16_t test_svdup_n_bf16_x(svbool_t pg, bfloat16_t op) {
// CHECK-LABEL: @test_svdup_lane_bf16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.tbl.nxv8bf16(<vscale x 8 x bfloat> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svdup_lane_bf16u14__SVBFloat16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.tbl.nxv8bf16(<vscale x 8 x bfloat> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x bfloat> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup.c
index a09e952625701..e8bdb3f0a3273 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup.c
@@ -16,12 +16,14 @@
// CHECK-LABEL: @test_svdup_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svdup_n_s8a(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
svint8_t test_svdup_n_s8(int8_t op)
@@ -31,12 +33,14 @@ svint8_t test_svdup_n_s8(int8_t op)
// CHECK-LABEL: @test_svdup_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_s16s(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
svint16_t test_svdup_n_s16(int16_t op)
@@ -46,12 +50,14 @@ svint16_t test_svdup_n_s16(int16_t op)
// CHECK-LABEL: @test_svdup_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_s32i(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svint32_t test_svdup_n_s32(int32_t op)
@@ -61,12 +67,14 @@ svint32_t test_svdup_n_s32(int32_t op)
// CHECK-LABEL: @test_svdup_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_s64l(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
svint64_t test_svdup_n_s64(int64_t op)
@@ -76,12 +84,14 @@ svint64_t test_svdup_n_s64(int64_t op)
// CHECK-LABEL: @test_svdup_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svdup_n_u8h(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP0]]
//
svuint8_t test_svdup_n_u8(uint8_t op)
@@ -91,12 +101,14 @@ svuint8_t test_svdup_n_u8(uint8_t op)
// CHECK-LABEL: @test_svdup_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_u16t(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP0]]
//
svuint16_t test_svdup_n_u16(uint16_t op)
@@ -106,12 +118,14 @@ svuint16_t test_svdup_n_u16(uint16_t op)
// CHECK-LABEL: @test_svdup_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_u32j(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svuint32_t test_svdup_n_u32(uint32_t op)
@@ -121,12 +135,14 @@ svuint32_t test_svdup_n_u32(uint32_t op)
// CHECK-LABEL: @test_svdup_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_u64m(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP0]]
//
svuint64_t test_svdup_n_u64(uint64_t op)
@@ -136,12 +152,14 @@ svuint64_t test_svdup_n_u64(uint64_t op)
// CHECK-LABEL: @test_svdup_n_f16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_f16Dh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP0]]
//
svfloat16_t test_svdup_n_f16(float16_t op)
@@ -151,12 +169,14 @@ svfloat16_t test_svdup_n_f16(float16_t op)
// CHECK-LABEL: @test_svdup_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_f32f(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
svfloat32_t test_svdup_n_f32(float32_t op)
@@ -166,12 +186,14 @@ svfloat32_t test_svdup_n_f32(float32_t op)
// CHECK-LABEL: @test_svdup_n_f64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_f64d(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP0]]
//
svfloat64_t test_svdup_n_f64(float64_t op)
@@ -730,13 +752,15 @@ svfloat64_t test_svdup_n_f64_x(svbool_t pg, float64_t op)
// CHECK-LABEL: @test_svdup_lane_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svdup_lane_s8u10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -747,13 +771,15 @@ svint8_t test_svdup_lane_s8(svint8_t data, uint8_t index)
// CHECK-LABEL: @test_svdup_lane_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_s16u11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -764,13 +790,15 @@ svint16_t test_svdup_lane_s16(svint16_t data, uint16_t index)
// CHECK-LABEL: @test_svdup_lane_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_s32u11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -781,13 +809,15 @@ svint32_t test_svdup_lane_s32(svint32_t data, uint32_t index)
// CHECK-LABEL: @test_svdup_lane_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_s64u11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -798,13 +828,15 @@ svint64_t test_svdup_lane_s64(svint64_t data, uint64_t index)
// CHECK-LABEL: @test_svdup_lane_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svdup_lane_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -815,13 +847,15 @@ svuint8_t test_svdup_lane_u8(svuint8_t data, uint8_t index)
// CHECK-LABEL: @test_svdup_lane_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -832,13 +866,15 @@ svuint16_t test_svdup_lane_u16(svuint16_t data, uint16_t index)
// CHECK-LABEL: @test_svdup_lane_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -849,13 +885,15 @@ svuint32_t test_svdup_lane_u32(svuint32_t data, uint32_t index)
// CHECK-LABEL: @test_svdup_lane_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -866,13 +904,15 @@ svuint64_t test_svdup_lane_u64(svuint64_t data, uint64_t index)
// CHECK-LABEL: @test_svdup_lane_f16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16(<vscale x 8 x half> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_f16u13__SVFloat16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16(<vscale x 8 x half> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
//
@@ -883,13 +923,15 @@ svfloat16_t test_svdup_lane_f16(svfloat16_t data, uint16_t index)
// CHECK-LABEL: @test_svdup_lane_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32(<vscale x 4 x float> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_f32u13__SVFloat32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32(<vscale x 4 x float> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
@@ -900,13 +942,15 @@ svfloat32_t test_svdup_lane_f32(svfloat32_t data, uint32_t index)
// CHECK-LABEL: @test_svdup_lane_f64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64(<vscale x 2 x double> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svdup_lane_f64u13__SVFloat64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64(<vscale x 2 x double> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP1]]
//
@@ -917,12 +961,14 @@ svfloat64_t test_svdup_lane_f64(svfloat64_t data, uint64_t index)
// CHECK-LABEL: @test_svdup_n_b8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.dup.x.nxv16i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i1> [[DOTSPLATINSERT]], <vscale x 16 x i1> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svdup_n_b8b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.dup.x.nxv16i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i1> [[DOTSPLATINSERT]], <vscale x 16 x i1> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
svbool_t test_svdup_n_b8(bool op)
@@ -932,13 +978,15 @@ svbool_t test_svdup_n_b8(bool op)
// CHECK-LABEL: @test_svdup_n_b16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.dup.x.nxv8i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i1> [[DOTSPLATINSERT]], <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_b16b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.dup.x.nxv8i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i1> [[DOTSPLATINSERT]], <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -949,13 +997,15 @@ svbool_t test_svdup_n_b16(bool op)
// CHECK-LABEL: @test_svdup_n_b32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.dup.x.nxv4i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i1> [[DOTSPLATINSERT]], <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_b32b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.dup.x.nxv4i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i1> [[DOTSPLATINSERT]], <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
@@ -966,13 +1016,15 @@ svbool_t test_svdup_n_b32(bool op)
// CHECK-LABEL: @test_svdup_n_b64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.dup.x.nxv2i1(i1 [[OP:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i1> [[DOTSPLATINSERT]], <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svdup_n_b64b(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.dup.x.nxv2i1(i1 [[OP:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i1> [[DOTSPLATINSERT]], <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dupq.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dupq.c
index effd3345303fe..3c13080e14f70 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dupq.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dupq.c
@@ -568,9 +568,8 @@ svfloat64_t test_svdupq_n_f64(float64_t x0, float64_t x1)
// CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
// CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vector.insert.nxv16i8.v16i8(<vscale x 16 x i8> undef, <16 x i8> [[TMP15]], i64 0)
// CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dupq.lane.nxv16i8(<vscale x 16 x i8> [[TMP17]], i64 0)
-// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
+// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP19]]
//
// CPP-CHECK-LABEL: @_Z16test_svdupq_n_b8bbbbbbbbbbbbbbbb(
// CPP-CHECK-NEXT: entry:
@@ -609,9 +608,8 @@ svfloat64_t test_svdupq_n_f64(float64_t x0, float64_t x1)
// CPP-CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
// CPP-CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vector.insert.nxv16i8.v16i8(<vscale x 16 x i8> undef, <16 x i8> [[TMP15]], i64 0)
// CPP-CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dupq.lane.nxv16i8(<vscale x 16 x i8> [[TMP17]], i64 0)
-// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
+// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP19]]
//
svbool_t test_svdupq_n_b8(bool x0, bool x1, bool x2, bool x3,
bool x4, bool x5, bool x6, bool x7,
@@ -643,10 +641,9 @@ svbool_t test_svdupq_n_b8(bool x0, bool x1, bool x2, bool x3,
// CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
// CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.v8i16(<vscale x 8 x i16> undef, <8 x i16> [[TMP15]], i64 0)
// CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dupq.lane.nxv8i16(<vscale x 8 x i16> [[TMP17]], i64 0)
-// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CHECK-NEXT: [[TMP21:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP20]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP21]]
+// CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP19]])
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
//
// CPP-CHECK-LABEL: @_Z17test_svdupq_n_b16bbbbbbbb(
// CPP-CHECK-NEXT: entry:
@@ -669,10 +666,9 @@ svbool_t test_svdupq_n_b8(bool x0, bool x1, bool x2, bool x3,
// CPP-CHECK-NEXT: [[TMP16:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
// CPP-CHECK-NEXT: [[TMP17:%.*]] = call <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.v8i16(<vscale x 8 x i16> undef, <8 x i16> [[TMP15]], i64 0)
// CPP-CHECK-NEXT: [[TMP18:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dupq.lane.nxv8i16(<vscale x 8 x i16> [[TMP17]], i64 0)
-// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CPP-CHECK-NEXT: [[TMP21:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP20]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP21]]
+// CPP-CHECK-NEXT: [[TMP19:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP19]])
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP20]]
//
svbool_t test_svdupq_n_b16(bool x0, bool x1, bool x2, bool x3,
bool x4, bool x5, bool x6, bool x7)
@@ -694,10 +690,9 @@ svbool_t test_svdupq_n_b16(bool x0, bool x1, bool x2, bool x3,
// CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
// CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v4i32(<vscale x 4 x i32> undef, <4 x i32> [[TMP7]], i64 0)
// CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dupq.lane.nxv4i32(<vscale x 4 x i32> [[TMP9]], i64 0)
-// CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> [[TMP11]])
-// CHECK-NEXT: [[TMP13:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP12]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP13]]
+// CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP11]])
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP12]]
//
// CPP-CHECK-LABEL: @_Z17test_svdupq_n_b32bbbb(
// CPP-CHECK-NEXT: entry:
@@ -712,10 +707,9 @@ svbool_t test_svdupq_n_b16(bool x0, bool x1, bool x2, bool x3,
// CPP-CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
// CPP-CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v4i32(<vscale x 4 x i32> undef, <4 x i32> [[TMP7]], i64 0)
// CPP-CHECK-NEXT: [[TMP10:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dupq.lane.nxv4i32(<vscale x 4 x i32> [[TMP9]], i64 0)
-// CPP-CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> [[TMP11]])
-// CPP-CHECK-NEXT: [[TMP13:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP12]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP13]]
+// CPP-CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP11]])
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP12]]
//
svbool_t test_svdupq_n_b32(bool x0, bool x1, bool x2, bool x3)
{
@@ -732,10 +726,9 @@ svbool_t test_svdupq_n_b32(bool x0, bool x1, bool x2, bool x3)
// CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
// CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 2 x i64> @llvm.experimental.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP3]], i64 0)
// CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dupq.lane.nxv2i64(<vscale x 2 x i64> [[TMP5]], i64 0)
-// CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]])
-// CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP8]])
-// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP9]]
+// CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP7]])
+// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP8]]
//
// CPP-CHECK-LABEL: @_Z17test_svdupq_n_b64bb(
// CPP-CHECK-NEXT: entry:
@@ -746,10 +739,9 @@ svbool_t test_svdupq_n_b32(bool x0, bool x1, bool x2, bool x3)
// CPP-CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
// CPP-CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 2 x i64> @llvm.experimental.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP3]], i64 0)
// CPP-CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dupq.lane.nxv2i64(<vscale x 2 x i64> [[TMP5]], i64 0)
-// CPP-CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]])
-// CPP-CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP8]])
-// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP9]]
+// CPP-CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP7]])
+// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP8]]
//
svbool_t test_svdupq_n_b64(bool x0, bool x1)
{
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_eor.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_eor.c
index 66704fbdcec3c..01aacd10839e6 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_eor.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_eor.c
@@ -428,14 +428,16 @@ svuint64_t test_sveor_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_sveor_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_sveor_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_sveor_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_sveor_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_sveor_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_sveor_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_sveor_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_sveor_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_sveor_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_sveor_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_sveor_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_sveor_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_sveor_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_sveor_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_sveor_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_sveor_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_sveor_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_sveor_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_sveor_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_sveor_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_sveor_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_sveor_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_sveor_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_sveor_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_sveor_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_sveor_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_sveor_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_sveor_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_sveor_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_sveor_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_sveor_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_sveor_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_sveor_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_sveor_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_sveor_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_sveor_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_sveor_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_sveor_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_sveor_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_sveor_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_sveor_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_sveor_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_sveor_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_sveor_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_sveor_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_sveor_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_sveor_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_sveor_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_sveor_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_sveor_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_sveor_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_sveor_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_sveor_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsl.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsl.c
index cb212faa0c03a..5b76a039d52d4 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsl.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsl.c
@@ -734,13 +734,15 @@ svuint32_t test_svlsl_wide_u32_x(svbool_t pg, svuint32_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_mu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -752,14 +754,16 @@ svint8_t test_svlsl_wide_n_s8_m(svbool_t pg, svint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_mu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -771,14 +775,16 @@ svint16_t test_svlsl_wide_n_s16_m(svbool_t pg, svint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_mu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -789,14 +795,16 @@ svint32_t test_svlsl_wide_n_s32_m(svbool_t pg, svint32_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_zu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -809,7 +817,8 @@ svint8_t test_svlsl_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -817,7 +826,8 @@ svint8_t test_svlsl_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_zu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -830,7 +840,8 @@ svint16_t test_svlsl_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -838,7 +849,8 @@ svint16_t test_svlsl_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_zu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -850,13 +862,15 @@ svint32_t test_svlsl_wide_n_s32_z(svbool_t pg, svint32_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_xu10__SVBool_tu10__SVInt8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -868,14 +882,16 @@ svint8_t test_svlsl_wide_n_s8_x(svbool_t pg, svint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_xu10__SVBool_tu11__SVInt16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -887,14 +903,16 @@ svint16_t test_svlsl_wide_n_s16_x(svbool_t pg, svint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsl_wide_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_xu10__SVBool_tu11__SVInt32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsr.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsr.c
index 3dd34eabf008d..c512f43ddbed3 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsr.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsr.c
@@ -375,13 +375,15 @@ svuint32_t test_svlsr_wide_u32_x(svbool_t pg, svuint32_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_mu10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -393,14 +395,16 @@ svuint8_t test_svlsr_wide_n_u8_m(svbool_t pg, svuint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_mu10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -412,14 +416,16 @@ svuint16_t test_svlsr_wide_n_u16_m(svbool_t pg, svuint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_mu10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -430,14 +436,16 @@ svuint32_t test_svlsr_wide_n_u32_m(svbool_t pg, svuint32_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_zu10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -450,7 +458,8 @@ svuint8_t test_svlsr_wide_n_u8_z(svbool_t pg, svuint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -458,7 +467,8 @@ svuint8_t test_svlsr_wide_n_u8_z(svbool_t pg, svuint8_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_zu10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -471,7 +481,8 @@ svuint16_t test_svlsr_wide_n_u16_z(svbool_t pg, svuint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -479,7 +490,8 @@ svuint16_t test_svlsr_wide_n_u16_z(svbool_t pg, svuint16_t op1, uint64_t op2)
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_zu10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -491,13 +503,15 @@ svuint32_t test_svlsr_wide_n_u32_z(svbool_t pg, svuint32_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_xu10__SVBool_tu11__SVUint8_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -509,14 +523,16 @@ svuint8_t test_svlsr_wide_n_u8_x(svbool_t pg, svuint8_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_xu10__SVBool_tu12__SVUint16_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -528,14 +544,16 @@ svuint16_t test_svlsr_wide_n_u16_x(svbool_t pg, svuint16_t op1, uint64_t op2)
// CHECK-LABEL: @test_svlsr_wide_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_xu10__SVBool_tu12__SVUint32_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mad.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mad.c
index 269d3918d0266..01a29a16e098b 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mad.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mad.c
@@ -428,14 +428,16 @@ svuint64_t test_svmad_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
// CHECK-LABEL: @test_svmad_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmad_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmad_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmad_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmad_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmad_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmad_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmad_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmad_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmad_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmad_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmad_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmad_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmad_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmad_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmad_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmad_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmad_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmad_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmad_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmad_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmad_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmad_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmad_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmad_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmad_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmad_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmad_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmad_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmad_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmad_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmad_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmad_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmad_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmad_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmad_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmad_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmad_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmad_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmad_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmad_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmad_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmad_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmad_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmad_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmad_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmad_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmad_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmad_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmad_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmad_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmad_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmad_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CHECK-LABEL: @test_svmad_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmad_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmad_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svmad_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmad_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svmad_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmad_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svmad_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmad_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svmad_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmad_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svmad_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmad_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svmad_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmad_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_max.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_max.c
index 3ba13569c1789..14cb35afe04f0 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_max.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_max.c
@@ -428,14 +428,16 @@ svuint64_t test_svmax_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svmax_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmax_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmax_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmax_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmax_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmax_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmax_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmax_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmax_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmax_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmax_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmax_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmax_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmax_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmax_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmax_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmax_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmax_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmax_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmax_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmax_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmax_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmax_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmax_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmax_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmax_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmax_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmax_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmax_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmax_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmax_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmax_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmax_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmax_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmax_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmax_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmax_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmax_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmax_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmax_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmax_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmax_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmax_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmax_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmax_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmax_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmax_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmax_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmax_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmax_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmax_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmax_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmax_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svmax_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svmax_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svmax_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svmax_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmax_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svmax_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svmax_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmax_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svmax_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svmax_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmax_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svmax_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmax_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svmax_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmax_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svmax_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmax_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svmax_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmax_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svmax_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmax_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_maxnm.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_maxnm.c
index 3ff6c42b20e23..4759a69b78631 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_maxnm.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_maxnm.c
@@ -176,7 +176,8 @@ svfloat64_t test_svmaxnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svmaxnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svmaxnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svmaxnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svmaxnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svmaxnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svmaxnm_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svmaxnm_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svmaxnm_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svmaxnm_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svmaxnm_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svmaxnm_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmaxnm_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_min.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_min.c
index d4c15435ba1ba..bf4657194b884 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_min.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_min.c
@@ -428,14 +428,16 @@ svuint64_t test_svmin_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svmin_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmin_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmin_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmin_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmin_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmin_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmin_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmin_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmin_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmin_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmin_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmin_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmin_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmin_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmin_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmin_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmin_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmin_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmin_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmin_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmin_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmin_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmin_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmin_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmin_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmin_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmin_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmin_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmin_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmin_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmin_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmin_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmin_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmin_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmin_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmin_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmin_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmin_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmin_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmin_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmin_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmin_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmin_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmin_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmin_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmin_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmin_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmin_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmin_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmin_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmin_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmin_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmin_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svmin_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svmin_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svmin_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svmin_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmin_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svmin_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svmin_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmin_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svmin_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svmin_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmin_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svmin_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmin_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svmin_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmin_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svmin_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmin_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svmin_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmin_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svmin_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmin_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_minnm.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_minnm.c
index e1a353ea427e5..6807d56a122e7 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_minnm.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_minnm.c
@@ -176,7 +176,8 @@ svfloat64_t test_svminnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svminnm_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svminnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svminnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svminnm_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svminnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svminnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svminnm_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svminnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svminnm_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svminnm_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svminnm_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svminnm_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svminnm_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svminnm_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svminnm_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svminnm_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svminnm_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svminnm_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svminnm_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svminnm_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mla.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mla.c
index 9d16676a7bb29..a7c8165ad8bc2 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mla.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mla.c
@@ -428,14 +428,16 @@ svuint64_t test_svmla_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
// CHECK-LABEL: @test_svmla_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmla_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmla_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmla_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmla_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmla_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmla_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmla_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmla_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmla_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmla_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmla_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmla_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmla_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmla_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmla_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmla_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmla_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmla_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmla_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmla_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmla_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmla_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmla_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmla_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmla_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmla_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmla_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmla_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmla_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmla_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmla_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmla_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmla_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmla_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmla_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmla_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmla_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmla_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmla_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmla_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmla_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmla_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmla_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmla_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmla_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmla_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmla_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmla_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmla_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmla_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmla_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmla_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CHECK-LABEL: @test_svmla_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmla_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmla_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svmla_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmla_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svmla_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmla_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svmla_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmla_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svmla_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmla_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svmla_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmla_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svmla_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmla_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mls.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mls.c
index badd6eca36600..0744c50c029fc 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mls.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mls.c
@@ -428,14 +428,16 @@ svuint64_t test_svmls_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
// CHECK-LABEL: @test_svmls_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmls_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmls_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmls_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmls_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmls_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmls_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmls_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmls_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmls_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmls_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmls_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmls_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmls_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmls_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmls_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmls_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmls_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmls_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmls_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmls_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmls_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmls_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmls_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmls_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmls_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmls_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmls_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmls_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmls_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmls_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmls_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmls_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmls_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmls_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmls_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmls_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmls_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmls_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmls_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmls_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmls_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmls_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmls_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmls_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmls_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmls_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmls_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmls_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmls_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmls_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmls_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmls_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CHECK-LABEL: @test_svmls_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmls_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmls_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svmls_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmls_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svmls_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmls_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svmls_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmls_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svmls_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmls_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svmls_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmls_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svmls_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmls_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_msb.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_msb.c
index 3d90b3703bbfa..fc5f6f4141c9d 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_msb.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_msb.c
@@ -428,14 +428,16 @@ svuint64_t test_svmsb_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
// CHECK-LABEL: @test_svmsb_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmsb_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmsb_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmsb_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmsb_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmsb_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmsb_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmsb_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmsb_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmsb_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmsb_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmsb_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmsb_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmsb_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmsb_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmsb_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmsb_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmsb_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmsb_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmsb_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmsb_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmsb_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmsb_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmsb_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmsb_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmsb_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmsb_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmsb_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmsb_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmsb_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmsb_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmsb_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmsb_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmsb_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmsb_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmsb_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmsb_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmsb_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
// CHECK-LABEL: @test_svmsb_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmsb_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmsb_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmsb_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
// CHECK-LABEL: @test_svmsb_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmsb_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
// CHECK-LABEL: @test_svmsb_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmsb_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
// CHECK-LABEL: @test_svmsb_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmsb_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
// CHECK-LABEL: @test_svmsb_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmsb_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
// CHECK-LABEL: @test_svmsb_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmsb_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
// CHECK-LABEL: @test_svmsb_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CHECK-LABEL: @test_svmsb_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svmsb_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svmsb_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svmsb_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svmsb_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svmsb_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svmsb_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
// CHECK-LABEL: @test_svmsb_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mul.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mul.c
index 23d1a87891737..040dfea9bda2e 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mul.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mul.c
@@ -428,14 +428,16 @@ svuint64_t test_svmul_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svmul_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmul_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmul_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmul_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmul_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmul_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmul_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmul_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmul_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmul_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmul_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmul_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmul_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmul_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmul_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmul_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmul_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmul_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmul_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmul_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmul_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmul_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmul_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmul_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmul_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmul_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmul_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmul_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmul_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmul_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmul_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmul_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmul_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmul_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmul_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmul_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmul_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmul_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmul_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmul_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmul_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmul_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmul_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmul_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmul_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmul_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmul_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmul_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmul_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmul_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmul_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmul_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmul_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svmul_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svmul_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svmul_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svmul_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmul_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svmul_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svmul_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmul_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svmul_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svmul_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmul_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svmul_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmul_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svmul_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmul_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svmul_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmul_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svmul_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmul_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svmul_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmul_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulh.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulh.c
index 68827598b367f..366169ac563af 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulh.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulh.c
@@ -428,14 +428,16 @@ svuint64_t test_svmulh_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svmulh_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svmulh_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmulh_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svmulh_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svmulh_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmulh_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svmulh_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svmulh_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmulh_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svmulh_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svmulh_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmulh_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svmulh_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmulh_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svmulh_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svmulh_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmulh_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svmulh_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svmulh_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmulh_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svmulh_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svmulh_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmulh_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svmulh_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmulh_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svmulh_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmulh_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svmulh_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmulh_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svmulh_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmulh_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svmulh_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmulh_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svmulh_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmulh_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svmulh_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmulh_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svmulh_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svmulh_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svmulh_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmulh_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svmulh_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmulh_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svmulh_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmulh_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svmulh_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svmulh_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svmulh_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmulh_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svmulh_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmulh_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svmulh_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svmulh_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulx.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulx.c
index 86a972b2505e8..23879919d93e1 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulx.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulx.c
@@ -176,7 +176,8 @@ svfloat64_t test_svmulx_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svmulx_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svmulx_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svmulx_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmulx_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svmulx_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svmulx_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmulx_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svmulx_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svmulx_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmulx_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svmulx_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmulx_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svmulx_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmulx_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svmulx_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svmulx_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svmulx_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svmulx_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svmulx_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svmulx_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmad.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmad.c
index 5e708bdc2604d..efe6a72d0aca9 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmad.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmad.c
@@ -176,7 +176,8 @@ svfloat64_t test_svnmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CHECK-LABEL: @test_svnmad_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svnmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svnmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmad_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svnmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svnmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmad_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svnmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svnmad_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmad_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svnmad_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmad_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svnmad_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmad_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svnmad_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmad_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svnmad_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmad_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svnmad_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmad_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmla.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmla.c
index a2ed49a4633b8..6fe517250af86 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmla.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmla.c
@@ -176,7 +176,8 @@ svfloat64_t test_svnmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CHECK-LABEL: @test_svnmla_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svnmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svnmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmla_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svnmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svnmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmla_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svnmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svnmla_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmla_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svnmla_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmla_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svnmla_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmla_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svnmla_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmla_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svnmla_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmla_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svnmla_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmla_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmls.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmls.c
index c4698f529bdb2..17716efe8ccef 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmls.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmls.c
@@ -176,7 +176,8 @@ svfloat64_t test_svnmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CHECK-LABEL: @test_svnmls_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svnmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svnmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmls_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svnmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svnmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmls_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svnmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svnmls_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmls_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svnmls_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmls_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svnmls_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmls_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svnmls_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmls_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svnmls_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmls_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svnmls_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmls_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmsb.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmsb.c
index c16c0cfe16b82..e0a047612b8f3 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmsb.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmsb.c
@@ -176,7 +176,8 @@ svfloat64_t test_svnmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CHECK-LABEL: @test_svnmsb_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svnmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svnmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svnmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svnmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svnmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svnmsb_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svnmsb_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svnmsb_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svnmsb_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svnmsb_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svnmsb_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
// CHECK-LABEL: @test_svnmsb_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_orr.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_orr.c
index 2a8d656a78303..d4a63eed7f872 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_orr.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_orr.c
@@ -428,14 +428,16 @@ svuint64_t test_svorr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svorr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svorr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svorr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svorr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svorr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svorr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svorr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svorr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svorr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svorr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svorr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svorr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svorr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svorr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svorr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svorr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svorr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svorr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svorr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svorr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svorr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svorr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svorr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svorr_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svorr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svorr_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svorr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svorr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svorr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svorr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svorr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svorr_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svorr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svorr_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svorr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svorr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svorr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svorr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svorr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svorr_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svorr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svorr_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svorr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svorr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svorr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svorr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svorr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svorr_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svorr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svorr_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svorr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svorr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svorr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qadd.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qadd.c
index 08e0c071952a6..d60eb754077a8 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qadd.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qadd.c
@@ -136,13 +136,15 @@ svuint64_t test_svqadd_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svqadd_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqadd_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -153,13 +155,15 @@ svint8_t test_svqadd_n_s8(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqadd_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -170,13 +174,15 @@ svint16_t test_svqadd_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqadd_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -187,13 +193,15 @@ svint32_t test_svqadd_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqadd_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -204,13 +212,15 @@ svint64_t test_svqadd_n_s64(svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqadd_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqadd_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -221,13 +231,15 @@ svuint8_t test_svqadd_n_u8(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqadd_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -238,13 +250,15 @@ svuint16_t test_svqadd_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqadd_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -255,13 +269,15 @@ svuint32_t test_svqadd_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqadd_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqadd_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qsub.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qsub.c
index 8ec7f0f377dd2..d0d477a0febe5 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qsub.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qsub.c
@@ -136,13 +136,15 @@ svuint64_t test_svqsub_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svqsub_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqsub_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -153,13 +155,15 @@ svint8_t test_svqsub_n_s8(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqsub_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -170,13 +174,15 @@ svint16_t test_svqsub_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqsub_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -187,13 +193,15 @@ svint32_t test_svqsub_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqsub_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -204,13 +212,15 @@ svint64_t test_svqsub_n_s64(svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqsub_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svqsub_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -221,13 +231,15 @@ svuint8_t test_svqsub_n_u8(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqsub_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -238,13 +250,15 @@ svuint16_t test_svqsub_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqsub_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -255,13 +269,15 @@ svuint32_t test_svqsub_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqsub_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svqsub_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_scale.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_scale.c
index 035adedd821ed..f0e8060513dd5 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_scale.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_scale.c
@@ -176,7 +176,8 @@ svfloat64_t test_svscale_f64_x(svbool_t pg, svfloat64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svscale_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -184,7 +185,8 @@ svfloat64_t test_svscale_f64_x(svbool_t pg, svfloat64_t op1, svint64_t op2)
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_zu10__SVBool_tu13__SVFloat16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -197,7 +199,8 @@ svfloat16_t test_svscale_n_f16_z(svbool_t pg, svfloat16_t op1, int16_t op2)
// CHECK-LABEL: @test_svscale_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -205,7 +208,8 @@ svfloat16_t test_svscale_n_f16_z(svbool_t pg, svfloat16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_zu10__SVBool_tu13__SVFloat32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -218,7 +222,8 @@ svfloat32_t test_svscale_n_f32_z(svbool_t pg, svfloat32_t op1, int32_t op2)
// CHECK-LABEL: @test_svscale_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -226,7 +231,8 @@ svfloat32_t test_svscale_n_f32_z(svbool_t pg, svfloat32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_zu10__SVBool_tu13__SVFloat64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -239,14 +245,16 @@ svfloat64_t test_svscale_n_f64_z(svbool_t pg, svfloat64_t op1, int64_t op2)
// CHECK-LABEL: @test_svscale_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_mu10__SVBool_tu13__SVFloat16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -258,14 +266,16 @@ svfloat16_t test_svscale_n_f16_m(svbool_t pg, svfloat16_t op1, int16_t op2)
// CHECK-LABEL: @test_svscale_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_mu10__SVBool_tu13__SVFloat32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -277,14 +287,16 @@ svfloat32_t test_svscale_n_f32_m(svbool_t pg, svfloat32_t op1, int32_t op2)
// CHECK-LABEL: @test_svscale_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_mu10__SVBool_tu13__SVFloat64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -296,14 +308,16 @@ svfloat64_t test_svscale_n_f64_m(svbool_t pg, svfloat64_t op1, int64_t op2)
// CHECK-LABEL: @test_svscale_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_xu10__SVBool_tu13__SVFloat16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -315,14 +329,16 @@ svfloat16_t test_svscale_n_f16_x(svbool_t pg, svfloat16_t op1, int16_t op2)
// CHECK-LABEL: @test_svscale_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_xu10__SVBool_tu13__SVFloat32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -334,14 +350,16 @@ svfloat32_t test_svscale_n_f32_x(svbool_t pg, svfloat32_t op1, int32_t op2)
// CHECK-LABEL: @test_svscale_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_xu10__SVBool_tu13__SVFloat64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sub.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sub.c
index a6e1c7016ac61..3418a7e5523c0 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sub.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sub.c
@@ -428,14 +428,16 @@ svuint64_t test_svsub_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svsub_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsub_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsub_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsub_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svsub_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svsub_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsub_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsub_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svsub_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svsub_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svsub_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svsub_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsub_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svsub_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsub_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svsub_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsub_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svsub_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svsub_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svsub_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsub_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svsub_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsub_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svsub_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svsub_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svsub_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svsub_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svsub_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsub_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svsub_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsub_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svsub_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsub_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svsub_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svsub_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svsub_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsub_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svsub_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsub_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svsub_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svsub_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svsub_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svsub_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svsub_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svsub_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svsub_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svsub_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svsub_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svsub_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svsub_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svsub_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svsub_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svsub_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svsub_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svsub_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svsub_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svsub_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svsub_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svsub_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svsub_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svsub_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svsub_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_subr.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_subr.c
index 218d43dcedf3d..69d11b8d9990d 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_subr.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_subr.c
@@ -428,14 +428,16 @@ svuint64_t test_svsubr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svsubr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -448,7 +450,8 @@ svint8_t test_svsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsubr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -456,7 +459,8 @@ svint8_t test_svsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -469,7 +473,8 @@ svint16_t test_svsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsubr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -477,7 +482,8 @@ svint16_t test_svsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -490,7 +496,8 @@ svint32_t test_svsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsubr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -498,7 +505,8 @@ svint32_t test_svsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -510,14 +518,16 @@ svint64_t test_svsubr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svsubr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -530,7 +540,8 @@ svuint8_t test_svsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsubr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -538,7 +549,8 @@ svuint8_t test_svsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -551,7 +563,8 @@ svuint16_t test_svsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsubr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -559,7 +572,8 @@ svuint16_t test_svsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -572,7 +586,8 @@ svuint32_t test_svsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svsubr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -580,7 +595,8 @@ svuint32_t test_svsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -592,13 +608,15 @@ svuint64_t test_svsubr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svsubr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -610,14 +628,16 @@ svint8_t test_svsubr_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsubr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -629,14 +649,16 @@ svint16_t test_svsubr_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsubr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -648,14 +670,16 @@ svint32_t test_svsubr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsubr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -666,13 +690,15 @@ svint64_t test_svsubr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svsubr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -684,14 +710,16 @@ svuint8_t test_svsubr_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsubr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -703,14 +731,16 @@ svuint16_t test_svsubr_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsubr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -722,14 +752,16 @@ svuint32_t test_svsubr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svsubr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -740,13 +772,15 @@ svuint64_t test_svsubr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svsubr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -758,14 +792,16 @@ svint8_t test_svsubr_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsubr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -777,14 +813,16 @@ svint16_t test_svsubr_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsubr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -796,14 +834,16 @@ svint32_t test_svsubr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsubr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -814,13 +854,15 @@ svint64_t test_svsubr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svsubr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -832,14 +874,16 @@ svuint8_t test_svsubr_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsubr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -851,14 +895,16 @@ svuint16_t test_svsubr_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsubr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -870,14 +916,16 @@ svuint32_t test_svsubr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svsubr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -1048,7 +1096,8 @@ svfloat64_t test_svsubr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CHECK-LABEL: @test_svsubr_n_f16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1056,7 +1105,8 @@ svfloat64_t test_svsubr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP3]]
@@ -1069,7 +1119,8 @@ svfloat16_t test_svsubr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svsubr_n_f32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1077,7 +1128,8 @@ svfloat16_t test_svsubr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP3]]
@@ -1090,7 +1142,8 @@ svfloat32_t test_svsubr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svsubr_n_f64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1098,7 +1151,8 @@ svfloat32_t test_svsubr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_zu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP3]]
@@ -1111,14 +1165,16 @@ svfloat64_t test_svsubr_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svsubr_n_f16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1130,14 +1186,16 @@ svfloat16_t test_svsubr_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svsubr_n_f32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1149,14 +1207,16 @@ svfloat32_t test_svsubr_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svsubr_n_f64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_mu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
@@ -1168,14 +1228,16 @@ svfloat64_t test_svsubr_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
// CHECK-LABEL: @test_svsubr_n_f16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
//
@@ -1187,14 +1249,16 @@ svfloat16_t test_svsubr_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
// CHECK-LABEL: @test_svsubr_n_f32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
//
@@ -1206,14 +1270,16 @@ svfloat32_t test_svsubr_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
// CHECK-LABEL: @test_svsubr_n_f64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_xu10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sudot.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sudot.c
index e23bf6ce10d4d..152709d888296 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sudot.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sudot.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -target-feature +i8mm -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -29,13 +31,15 @@ svint32_t test_svsudot_s32(svint32_t x, svint8_t y, svuint8_t z) {
// CHECK-LABEL: @test_svsudot_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[Y:%.*]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsudot_n_s32u11__SVInt32_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[Y:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_usdot.c b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_usdot.c
index 0eefd34b490b2..5ede96cf35ae1 100644
--- a/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_usdot.c
+++ b/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_usdot.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -target-feature +i8mm -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -target-feature +i8mm -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -29,13 +31,15 @@ svint32_t test_svusdot_s32(svint32_t x, svuint8_t y, svint8_t z) {
// CHECK-LABEL: @test_svusdot_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[Y:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svusdot_n_s32u11__SVInt32_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[Y:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_aba.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_aba.c
index 64b2a14346525..ee632edb455a6 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_aba.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_aba.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_svaba_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svaba_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.saba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svaba_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.saba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_svaba_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svaba_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_svaba_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svaba_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_svaba_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svaba_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_svaba_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svaba_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uaba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svaba_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uaba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_svaba_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svaba_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_svaba_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svaba_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_svaba_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svaba_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svaba_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalb.c
index fcb0eedd6be46..40e6329278c8e 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svabalb_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
// CHECK-LABEL: @test_svabalb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svabalb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svabalb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svabalb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svabalb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svabalb_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svabalb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svabalb_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svabalb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svabalb_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svabalb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalb_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalt.c
index 246f84b2273ba..4e2db22220369 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svabalt_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
// CHECK-LABEL: @test_svabalt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svabalt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svabalt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svabalt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svabalt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svabalt_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svabalt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svabalt_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svabalt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svabalt_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svabalt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabalt_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlb.c
index d91398c81deca..4ff81c2ec8c54 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svabdlb_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svabdlb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svabdlb_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svabdlb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svabdlb_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svabdlb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svabdlb_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svabdlb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svabdlb_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svabdlb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svabdlb_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svabdlb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlt.c
index 491ac415dde25..d0164065df7cc 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svabdlt_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svabdlt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svabdlt_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svabdlt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svabdlt_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svabdlt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svabdlt_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svabdlt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svabdlt_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svabdlt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svabdlt_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svabdlt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclb.c
index dc9bfbc58e503..b16c9b09523b9 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -51,13 +53,15 @@ svuint64_t test_svadclb_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svadclb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclb_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -70,13 +74,15 @@ svuint32_t test_svadclb_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svadclb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclb_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclt.c
index a42ac072e13e2..7ac6cadc8d204 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -51,13 +53,15 @@ svuint64_t test_svadclt_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svadclt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclt_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -70,13 +74,15 @@ svuint32_t test_svadclt_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svadclt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svadclt_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnb.c
index 070fc493b9ee4..1fb99de25ff34 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svaddhnb_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svaddhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svaddhnb_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svaddhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svaddhnb_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svaddhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svaddhnb_n_s64(svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svaddhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svaddhnb_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svaddhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svaddhnb_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svaddhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnt.c
index 8bc757ec9356a..5e75799db4b8c 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svaddhnt_u64(svuint32_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svaddhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svaddhnt_n_s16(svint8_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svaddhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svaddhnt_n_s32(svint16_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svaddhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svaddhnt_n_s64(svint32_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svaddhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svaddhnt_n_u16(svuint8_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svaddhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svaddhnt_n_u32(svuint16_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svaddhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlb.c
index 8eed03d66bee2..8efc990b57920 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svaddlb_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svaddlb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svaddlb_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svaddlb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svaddlb_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svaddlb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svaddlb_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svaddlb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svaddlb_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svaddlb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svaddlb_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svaddlb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlbt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlbt.c
index adbc91e186691..b494bd77387b2 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlbt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlbt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svaddlbt_s64(svint32_t op1, svint32_t op2)
// CHECK-LABEL: @test_svaddlbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svaddlbt_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svaddlbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svaddlbt_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svaddlbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlt.c
index 80bbcec99e8cb..57313698caef8 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svaddlt_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svaddlt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svaddlt_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svaddlt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svaddlt_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svaddlt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svaddlt_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svaddlt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svaddlt_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svaddlt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svaddlt_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svaddlt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwb.c
index dd0fecaf0fcd0..8b6efd906b1db 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svaddwb_u64(svuint64_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svaddwb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svaddwb_n_s16(svint16_t op1, int8_t op2)
// CHECK-LABEL: @test_svaddwb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svaddwb_n_s32(svint32_t op1, int16_t op2)
// CHECK-LABEL: @test_svaddwb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svaddwb_n_s64(svint64_t op1, int32_t op2)
// CHECK-LABEL: @test_svaddwb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svaddwb_n_u16(svuint16_t op1, uint8_t op2)
// CHECK-LABEL: @test_svaddwb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svaddwb_n_u32(svuint32_t op1, uint16_t op2)
// CHECK-LABEL: @test_svaddwb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwt.c
index b059f5c698ab4..aa9e7e5848204 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svaddwt_u64(svuint64_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svaddwt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svaddwt_n_s16(svint16_t op1, int8_t op2)
// CHECK-LABEL: @test_svaddwt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svaddwt_n_s32(svint32_t op1, int16_t op2)
// CHECK-LABEL: @test_svaddwt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svaddwt_n_s64(svint64_t op1, int32_t op2)
// CHECK-LABEL: @test_svaddwt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svaddwt_n_u16(svuint16_t op1, uint8_t op2)
// CHECK-LABEL: @test_svaddwt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svaddwt_n_u32(svuint32_t op1, uint16_t op2)
// CHECK-LABEL: @test_svaddwt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bcax.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bcax.c
index 7d68aef34a446..9fc6391ccb8b7 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bcax.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bcax.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_svbcax_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svbcax_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbcax_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_svbcax_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svbcax_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_svbcax_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svbcax_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_svbcax_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svbcax_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_svbcax_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svbcax_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbcax_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_svbcax_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svbcax_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_svbcax_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svbcax_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_svbcax_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svbcax_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbcax_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bdep.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bdep.c
index feb02c035e30a..619f1e77c5a2b 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bdep.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bdep.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -85,13 +87,15 @@ svuint64_t test_svbdep_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svbdep_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bdep.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbdep_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bdep.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -104,13 +108,15 @@ svuint8_t test_svbdep_n_u8(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svbdep_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bdep.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbdep_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bdep.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -123,13 +129,15 @@ svuint16_t test_svbdep_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svbdep_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bdep.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbdep_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bdep.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -142,13 +150,15 @@ svuint32_t test_svbdep_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svbdep_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bdep.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbdep_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bdep.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bext.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bext.c
index 2e351c65b83a0..67c601ce8922d 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bext.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bext.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -85,13 +87,15 @@ svuint64_t test_svbext_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svbext_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bext.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbext_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bext.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -104,13 +108,15 @@ svuint8_t test_svbext_n_u8(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svbext_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bext.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbext_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bext.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -123,13 +129,15 @@ svuint16_t test_svbext_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svbext_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bext.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbext_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bext.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -142,13 +150,15 @@ svuint32_t test_svbext_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svbext_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bext.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbext_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bext.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bgrp.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bgrp.c
index 88d2abb048081..1408011c1c2e4 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bgrp.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bgrp.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-bitperm -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -85,13 +87,15 @@ svuint64_t test_svbgrp_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svbgrp_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bgrp.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbgrp_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bgrp.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -104,13 +108,15 @@ svuint8_t test_svbgrp_n_u8(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svbgrp_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bgrp.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bgrp.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -123,13 +129,15 @@ svuint16_t test_svbgrp_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svbgrp_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bgrp.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bgrp.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -142,13 +150,15 @@ svuint32_t test_svbgrp_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svbgrp_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bgrp.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bgrp.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl.c
index dc75f24ecc99a..8ef7bfc4e4a16 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_svbsl_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svbsl_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svbsl_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_svbsl_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svbsl_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_svbsl_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svbsl_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_svbsl_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svbsl_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_svbsl_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svbsl_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z15test_svbsl_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_svbsl_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svbsl_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_svbsl_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svbsl_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_svbsl_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svbsl_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svbsl_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl1n.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl1n.c
index b3661c2c06900..3c41f009f5b9d 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl1n.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl1n.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_svbsl1n_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svbsl1n_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl1n_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_svbsl1n_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svbsl1n_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_svbsl1n_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svbsl1n_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_svbsl1n_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svbsl1n_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_svbsl1n_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svbsl1n_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl1n_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_svbsl1n_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svbsl1n_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_svbsl1n_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svbsl1n_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_svbsl1n_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svbsl1n_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl2n.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl2n.c
index 1fc3dce061c7c..e228720cc5784 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl2n.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl2n.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_svbsl2n_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svbsl2n_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl2n_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_svbsl2n_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svbsl2n_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_svbsl2n_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svbsl2n_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_svbsl2n_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svbsl2n_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_svbsl2n_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svbsl2n_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svbsl2n_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_svbsl2n_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svbsl2n_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_svbsl2n_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svbsl2n_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_svbsl2n_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svbsl2n_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eor3.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eor3.c
index bcb020225ad2e..fcdc1a44092fa 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eor3.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eor3.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_sveor3_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_sveor3_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_sveor3_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_sveor3_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_sveor3_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_sveor3_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_sveor3_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_sveor3_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_sveor3_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_sveor3_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_sveor3_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_sveor3_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_sveor3_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_sveor3_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_sveor3_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_sveor3_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_sveor3_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_sveor3_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveor3_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eorbt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eorbt.c
index c813ebc049ea1..50d20d68b34c9 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eorbt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eorbt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_sveorbt_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_sveorbt_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveorbt_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_sveorbt_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_sveorbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_sveorbt_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_sveorbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_sveorbt_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_sveorbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_sveorbt_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_sveorbt_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveorbt_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_sveorbt_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_sveorbt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_sveorbt_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_sveorbt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_sveorbt_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_sveorbt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eortb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eortb.c
index fdcac8f2edd25..7aa8397dd8c97 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eortb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eortb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_sveortb_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_sveortb_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveortb_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_sveortb_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_sveortb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_sveortb_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_sveortb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_sveortb_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_sveortb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_sveortb_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_sveortb_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_sveortb_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_sveortb_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_sveortb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_sveortb_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_sveortb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_sveortb_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_sveortb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_sveortb_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hadd.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hadd.c
index bdcc297bbe840..ed630275fffa4 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hadd.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hadd.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -166,13 +168,15 @@ svuint64_t test_svhadd_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svhadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -186,14 +190,16 @@ svint8_t test_svhadd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -207,14 +213,16 @@ svint16_t test_svhadd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -228,14 +236,16 @@ svint32_t test_svhadd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -248,13 +258,15 @@ svint64_t test_svhadd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -268,14 +280,16 @@ svuint8_t test_svhadd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -289,14 +303,16 @@ svuint16_t test_svhadd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -310,14 +326,16 @@ svuint32_t test_svhadd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -495,14 +513,16 @@ svuint64_t test_svhadd_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svhadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -517,7 +537,8 @@ svint8_t test_svhadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -525,7 +546,8 @@ svint8_t test_svhadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -540,7 +562,8 @@ svint16_t test_svhadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -548,7 +571,8 @@ svint16_t test_svhadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -563,7 +587,8 @@ svint32_t test_svhadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -571,7 +596,8 @@ svint32_t test_svhadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -585,14 +611,16 @@ svint64_t test_svhadd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -607,7 +635,8 @@ svuint8_t test_svhadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -615,7 +644,8 @@ svuint8_t test_svhadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -630,7 +660,8 @@ svuint16_t test_svhadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -638,7 +669,8 @@ svuint16_t test_svhadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -653,7 +685,8 @@ svuint32_t test_svhadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -661,7 +694,8 @@ svuint32_t test_svhadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -824,13 +858,15 @@ svuint64_t test_svhadd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svhadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -844,14 +880,16 @@ svint8_t test_svhadd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -865,14 +903,16 @@ svint16_t test_svhadd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -886,14 +926,16 @@ svint32_t test_svhadd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -906,13 +948,15 @@ svint64_t test_svhadd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -926,14 +970,16 @@ svuint8_t test_svhadd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -947,14 +993,16 @@ svuint16_t test_svhadd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -968,14 +1016,16 @@ svuint32_t test_svhadd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsub.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsub.c
index 47bdeea578370..c19a0247d6e6a 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsub.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsub.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -477,14 +479,16 @@ svuint64_t test_svhsub_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svhsub_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -499,7 +503,8 @@ svint8_t test_svhsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhsub_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -507,7 +512,8 @@ svint8_t test_svhsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -522,7 +528,8 @@ svint16_t test_svhsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhsub_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -530,7 +537,8 @@ svint16_t test_svhsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -545,7 +553,8 @@ svint32_t test_svhsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhsub_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -553,7 +562,8 @@ svint32_t test_svhsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -567,14 +577,16 @@ svint64_t test_svhsub_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhsub_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -589,7 +601,8 @@ svuint8_t test_svhsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhsub_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -597,7 +610,8 @@ svuint8_t test_svhsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -612,7 +626,8 @@ svuint16_t test_svhsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhsub_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -620,7 +635,8 @@ svuint16_t test_svhsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -635,7 +651,8 @@ svuint32_t test_svhsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhsub_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -643,7 +660,8 @@ svuint32_t test_svhsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -657,13 +675,15 @@ svuint64_t test_svhsub_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svhsub_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -677,14 +697,16 @@ svint8_t test_svhsub_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhsub_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -698,14 +720,16 @@ svint16_t test_svhsub_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhsub_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -719,14 +743,16 @@ svint32_t test_svhsub_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhsub_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -739,13 +765,15 @@ svint64_t test_svhsub_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhsub_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -759,14 +787,16 @@ svuint8_t test_svhsub_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhsub_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -780,14 +810,16 @@ svuint16_t test_svhsub_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhsub_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -801,14 +833,16 @@ svuint32_t test_svhsub_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhsub_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -821,13 +855,15 @@ svuint64_t test_svhsub_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svhsub_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -841,14 +877,16 @@ svint8_t test_svhsub_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhsub_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -862,14 +900,16 @@ svint16_t test_svhsub_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhsub_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -883,14 +923,16 @@ svint32_t test_svhsub_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhsub_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -903,13 +945,15 @@ svint64_t test_svhsub_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhsub_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -923,14 +967,16 @@ svuint8_t test_svhsub_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhsub_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -944,14 +990,16 @@ svuint16_t test_svhsub_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhsub_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -965,14 +1013,16 @@ svuint32_t test_svhsub_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhsub_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsubr.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsubr.c
index 823cfd8681757..76726523202d6 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsubr.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsubr.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -477,14 +479,16 @@ svuint64_t test_svhsubr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svhsubr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -499,7 +503,8 @@ svint8_t test_svhsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhsubr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -507,7 +512,8 @@ svint8_t test_svhsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -522,7 +528,8 @@ svint16_t test_svhsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhsubr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -530,7 +537,8 @@ svint16_t test_svhsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -545,7 +553,8 @@ svint32_t test_svhsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhsubr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -553,7 +562,8 @@ svint32_t test_svhsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -567,14 +577,16 @@ svint64_t test_svhsubr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhsubr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -589,7 +601,8 @@ svuint8_t test_svhsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhsubr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -597,7 +610,8 @@ svuint8_t test_svhsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -612,7 +626,8 @@ svuint16_t test_svhsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhsubr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -620,7 +635,8 @@ svuint16_t test_svhsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -635,7 +651,8 @@ svuint32_t test_svhsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhsubr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -643,7 +660,8 @@ svuint32_t test_svhsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -657,13 +675,15 @@ svuint64_t test_svhsubr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svhsubr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -677,14 +697,16 @@ svint8_t test_svhsubr_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhsubr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -698,14 +720,16 @@ svint16_t test_svhsubr_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhsubr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -719,14 +743,16 @@ svint32_t test_svhsubr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhsubr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -739,13 +765,15 @@ svint64_t test_svhsubr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhsubr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -759,14 +787,16 @@ svuint8_t test_svhsubr_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhsubr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -780,14 +810,16 @@ svuint16_t test_svhsubr_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhsubr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -801,14 +833,16 @@ svuint32_t test_svhsubr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhsubr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -821,13 +855,15 @@ svuint64_t test_svhsubr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svhsubr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -841,14 +877,16 @@ svint8_t test_svhsubr_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svhsubr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -862,14 +900,16 @@ svint16_t test_svhsubr_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svhsubr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -883,14 +923,16 @@ svint32_t test_svhsubr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svhsubr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -903,13 +945,15 @@ svint64_t test_svhsubr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svhsubr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -923,14 +967,16 @@ svuint8_t test_svhsubr_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svhsubr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -944,14 +990,16 @@ svuint16_t test_svhsubr_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svhsubr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -965,14 +1013,16 @@ svuint32_t test_svhsubr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svhsubr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalb.c
index f3385a5408167..24db332c84846 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svmlalb_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
// CHECK-LABEL: @test_svmlalb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svmlalb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmlalb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svmlalb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svmlalb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svmlalb_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svmlalb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svmlalb_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svmlalb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svmlalb_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svmlalb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -386,13 +400,15 @@ svfloat32_t test_svmlalb_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
// CHECK-LABEL: @test_svmlalb_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalb_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalt.c
index a5c75f811a7d5..0d7bb6b10aaae 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svmlalt_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
// CHECK-LABEL: @test_svmlalt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svmlalt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmlalt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svmlalt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svmlalt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svmlalt_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svmlalt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svmlalt_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svmlalt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svmlalt_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svmlalt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -386,13 +400,15 @@ svfloat32_t test_svmlalt_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
// CHECK-LABEL: @test_svmlalt_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlalt_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslb.c
index 457abe252c3f5..622677b702cfa 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svmlslb_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
// CHECK-LABEL: @test_svmlslb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svmlslb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmlslb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svmlslb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svmlslb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svmlslb_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svmlslb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svmlslb_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svmlslb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svmlslb_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svmlslb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -386,13 +400,15 @@ svfloat32_t test_svmlslb_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
// CHECK-LABEL: @test_svmlslb_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslb_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslt.c
index 66ee032b91b13..3be97c6db902f 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svmlslt_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
// CHECK-LABEL: @test_svmlslt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svmlslt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svmlslt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svmlslt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svmlslt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svmlslt_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svmlslt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u16u12__SVUint16_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svmlslt_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svmlslt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u32u12__SVUint32_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svmlslt_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svmlslt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u64u12__SVUint64_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -386,13 +400,15 @@ svfloat32_t test_svmlslt_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
// CHECK-LABEL: @test_svmlslt_n_f32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmlslt_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullb.c
index 70723deb57bf0..10abdafd1bb2d 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svmullb_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svmullb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svmullb_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmullb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svmullb_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmullb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svmullb_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmullb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svmullb_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmullb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svmullb_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmullb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullt.c
index b02989e681e5b..a08a30dbd443b 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svmullt_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svmullt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svmullt_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svmullt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svmullt_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svmullt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svmullt_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svmullt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svmullt_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svmullt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svmullt_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svmullt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svmullt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_nbsl.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_nbsl.c
index 54e5a9314a2a6..03c0500033c85 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_nbsl.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_nbsl.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -153,13 +155,15 @@ svuint64_t test_svnbsl_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svnbsl_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svnbsl_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -172,13 +176,15 @@ svint8_t test_svnbsl_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svnbsl_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -191,13 +197,15 @@ svint16_t test_svnbsl_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svnbsl_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -210,13 +218,15 @@ svint32_t test_svnbsl_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svnbsl_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -229,13 +239,15 @@ svint64_t test_svnbsl_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svnbsl_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svnbsl_n_u8u11__SVUint8_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -248,13 +260,15 @@ svuint8_t test_svnbsl_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
// CHECK-LABEL: @test_svnbsl_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u16u12__SVUint16_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -267,13 +281,15 @@ svuint16_t test_svnbsl_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svnbsl_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -286,13 +302,15 @@ svuint32_t test_svnbsl_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svnbsl_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmul.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmul.c
index 9e85b316b1208..c0e68957780ab 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmul.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmul.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -34,13 +36,15 @@ svuint8_t test_svpmul_u8(svuint8_t op1, svuint8_t op2)
// CHECK-LABEL: @test_svpmul_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmul.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z16test_svpmul_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmul.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb.c
index 767f4ff1fdb63..811f3de46b6c8 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -51,13 +53,15 @@ svuint32_t test_svpmullb_pair_u32(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svpmullb_pair_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z23test_svpmullb_pair_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -70,13 +74,15 @@ svuint8_t test_svpmullb_pair_n_u8(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svpmullb_pair_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullb_pair_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -127,14 +133,16 @@ svuint64_t test_svpmullb_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svpmullb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
@@ -148,14 +156,16 @@ svuint16_t test_svpmullb_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svpmullb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb_128.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb_128.c
index 43b3f0d271669..eb701007d7a72 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb_128.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb_128.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -34,13 +36,15 @@ svuint64_t test_svpmullb_pair_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svpmullb_pair_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullb.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullb_pair_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullb.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt.c
index 355a6ddafe7ac..e55bedfac4a6b 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -51,13 +53,15 @@ svuint32_t test_svpmullt_pair_u32(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svpmullt_pair_n_u8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z23test_svpmullt_pair_n_u8u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -70,13 +74,15 @@ svuint8_t test_svpmullt_pair_n_u8(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svpmullt_pair_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullt_pair_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -127,14 +133,16 @@ svuint64_t test_svpmullt_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svpmullt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
@@ -148,14 +156,16 @@ svuint16_t test_svpmullt_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svpmullt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt_128.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt_128.c
index 298073adb3217..8c33762ef56c6 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt_128.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt_128.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2-aes -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -34,13 +36,15 @@ svuint64_t test_svpmullt_pair_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svpmullt_pair_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullt.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z24test_svpmullt_pair_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullt.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qadd.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qadd.c
index b6f35b2e7ff14..1e289fa6376ab 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qadd.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qadd.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -166,13 +168,15 @@ svuint64_t test_svqadd_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svqadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -186,14 +190,16 @@ svint8_t test_svqadd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -207,14 +213,16 @@ svint16_t test_svqadd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -228,14 +236,16 @@ svint32_t test_svqadd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -248,13 +258,15 @@ svint64_t test_svqadd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -268,14 +280,16 @@ svuint8_t test_svqadd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -289,14 +303,16 @@ svuint16_t test_svqadd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -310,14 +326,16 @@ svuint32_t test_svqadd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -495,14 +513,16 @@ svuint64_t test_svqadd_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svqadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -517,7 +537,8 @@ svint8_t test_svqadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -525,7 +546,8 @@ svint8_t test_svqadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -540,7 +562,8 @@ svint16_t test_svqadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -548,7 +571,8 @@ svint16_t test_svqadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -563,7 +587,8 @@ svint32_t test_svqadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -571,7 +596,8 @@ svint32_t test_svqadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -585,14 +611,16 @@ svint64_t test_svqadd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -607,7 +635,8 @@ svuint8_t test_svqadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -615,7 +644,8 @@ svuint8_t test_svqadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -630,7 +660,8 @@ svuint16_t test_svqadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -638,7 +669,8 @@ svuint16_t test_svqadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -653,7 +685,8 @@ svuint32_t test_svqadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -661,7 +694,8 @@ svuint32_t test_svqadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -824,13 +858,15 @@ svuint64_t test_svqadd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svqadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -844,14 +880,16 @@ svint8_t test_svqadd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -865,14 +903,16 @@ svint16_t test_svqadd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -886,14 +926,16 @@ svint32_t test_svqadd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -906,13 +948,15 @@ svint64_t test_svqadd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -926,14 +970,16 @@ svuint8_t test_svqadd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -947,14 +993,16 @@ svuint16_t test_svqadd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -968,14 +1016,16 @@ svuint32_t test_svqadd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalb.c
index b9b5e3cc23d14..6d8b495424b1b 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmlalb_s64(svint64_t op1, svint32_t op2, svint32_t op3)
// CHECK-LABEL: @test_svqdmlalb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmlalb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqdmlalb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmlalb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqdmlalb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalbt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalbt.c
index 8d323d689706a..fe12a33d95c6a 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalbt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalbt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmlalbt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
// CHECK-LABEL: @test_svqdmlalbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmlalbt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqdmlalbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmlalbt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqdmlalbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalt.c
index 7a1691303e830..cdd2b2c791b2c 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmlalt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
// CHECK-LABEL: @test_svqdmlalt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmlalt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqdmlalt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmlalt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqdmlalt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslb.c
index e4fd60422eba2..bbdd2047340e3 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmlslb_s64(svint64_t op1, svint32_t op2, svint32_t op3)
// CHECK-LABEL: @test_svqdmlslb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmlslb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqdmlslb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmlslb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqdmlslb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslbt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslbt.c
index 56a5fe1a1788e..a463f3610bebe 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslbt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslbt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmlslbt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
// CHECK-LABEL: @test_svqdmlslbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmlslbt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqdmlslbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmlslbt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqdmlslbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslt.c
index 30444905ac21c..a1b5706657730 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmlslt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
// CHECK-LABEL: @test_svqdmlslt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s16u11__SVInt16_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmlslt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqdmlslt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s32u11__SVInt32_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmlslt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqdmlslt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s64u11__SVInt64_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmulh.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmulh.c
index ad3279d67ecc0..bc588bc0a71ca 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmulh.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmulh.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -85,13 +87,15 @@ svint64_t test_svqdmulh_s64(svint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svqdmulh_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqdmulh_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -104,13 +108,15 @@ svint8_t test_svqdmulh_n_s8(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqdmulh_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -123,13 +129,15 @@ svint16_t test_svqdmulh_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqdmulh_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -142,13 +150,15 @@ svint32_t test_svqdmulh_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqdmulh_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullb.c
index 61fedbfc0bb35..2f1c33f615db0 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmullb_s64(svint32_t op1, svint32_t op2)
// CHECK-LABEL: @test_svqdmullb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmullb_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqdmullb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmullb_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqdmullb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullt.c
index d14c758274cd9..5e1622f4a38cb 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svqdmullt_s64(svint32_t op1, svint32_t op2)
// CHECK-LABEL: @test_svqdmullt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svqdmullt_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqdmullt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svqdmullt_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqdmullt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlah.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlah.c
index af000b6284024..eb0ca21daef6f 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlah.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlah.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -85,13 +87,15 @@ svint64_t test_svqrdmlah_s64(svint64_t op1, svint64_t op2, svint64_t op3)
// CHECK-LABEL: @test_svqrdmlah_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlah.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrdmlah_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlah.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -104,13 +108,15 @@ svint8_t test_svqrdmlah_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqrdmlah_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlah.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlah_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlah.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -123,13 +129,15 @@ svint16_t test_svqrdmlah_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqrdmlah_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlah.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlah_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlah.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -142,13 +150,15 @@ svint32_t test_svqrdmlah_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svqrdmlah_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlah.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlah_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlah.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlsh.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlsh.c
index 07b36721bf3b4..f6c42805ffc62 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlsh.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmlsh.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -85,13 +87,15 @@ svint64_t test_svqrdmlsh_s64(svint64_t op1, svint64_t op2, svint64_t op3)
// CHECK-LABEL: @test_svqrdmlsh_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrdmlsh_n_s8u10__SVInt8_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmlsh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -104,13 +108,15 @@ svint8_t test_svqrdmlsh_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
// CHECK-LABEL: @test_svqrdmlsh_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s16u11__SVInt16_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmlsh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -123,13 +129,15 @@ svint16_t test_svqrdmlsh_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svqrdmlsh_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s32u11__SVInt32_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmlsh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -142,13 +150,15 @@ svint32_t test_svqrdmlsh_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svqrdmlsh_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmlsh_n_s64u11__SVInt64_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmlsh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmulh.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmulh.c
index 0a5727d8599f2..7cc7df2d90692 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmulh.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrdmulh.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -85,13 +87,15 @@ svint64_t test_svqrdmulh_s64(svint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svqrdmulh_n_s8(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrdmulh_n_s8u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -104,13 +108,15 @@ svint8_t test_svqrdmulh_n_s8(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqrdmulh_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmulh_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -123,13 +129,15 @@ svint16_t test_svqrdmulh_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqrdmulh_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmulh_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -142,13 +150,15 @@ svint32_t test_svqrdmulh_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqrdmulh_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrdmulh_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrshl.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrshl.c
index 38c6dbf5816eb..4783e0b00a42b 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrshl.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qrshl.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -477,14 +479,16 @@ svuint64_t test_svqrshl_u64_x(svbool_t pg, svuint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svqrshl_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -499,7 +503,8 @@ svint8_t test_svqrshl_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqrshl_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -507,7 +512,8 @@ svint8_t test_svqrshl_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -522,7 +528,8 @@ svint16_t test_svqrshl_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqrshl_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -530,7 +537,8 @@ svint16_t test_svqrshl_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -545,7 +553,8 @@ svint32_t test_svqrshl_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqrshl_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -553,7 +562,8 @@ svint32_t test_svqrshl_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -567,14 +577,16 @@ svint64_t test_svqrshl_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqrshl_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -589,7 +601,8 @@ svuint8_t test_svqrshl_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqrshl_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -597,7 +610,8 @@ svuint8_t test_svqrshl_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -612,7 +626,8 @@ svuint16_t test_svqrshl_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqrshl_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -620,7 +635,8 @@ svuint16_t test_svqrshl_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -635,7 +651,8 @@ svuint32_t test_svqrshl_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqrshl_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -643,7 +660,8 @@ svuint32_t test_svqrshl_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -657,13 +675,15 @@ svuint64_t test_svqrshl_n_u64_z(svbool_t pg, svuint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqrshl_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -677,14 +697,16 @@ svint8_t test_svqrshl_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqrshl_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -698,14 +720,16 @@ svint16_t test_svqrshl_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqrshl_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -719,14 +743,16 @@ svint32_t test_svqrshl_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqrshl_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -739,13 +765,15 @@ svint64_t test_svqrshl_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqrshl_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -759,14 +787,16 @@ svuint8_t test_svqrshl_n_u8_m(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqrshl_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -780,14 +810,16 @@ svuint16_t test_svqrshl_n_u16_m(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqrshl_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -801,14 +833,16 @@ svuint32_t test_svqrshl_n_u32_m(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqrshl_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -821,13 +855,15 @@ svuint64_t test_svqrshl_n_u64_m(svbool_t pg, svuint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqrshl_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -841,14 +877,16 @@ svint8_t test_svqrshl_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqrshl_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -862,14 +900,16 @@ svint16_t test_svqrshl_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqrshl_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -883,14 +923,16 @@ svint32_t test_svqrshl_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqrshl_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -903,13 +945,15 @@ svint64_t test_svqrshl_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqrshl_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqrshl_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqrshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -923,14 +967,16 @@ svuint8_t test_svqrshl_n_u8_x(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqrshl_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqrshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -944,14 +990,16 @@ svuint16_t test_svqrshl_n_u16_x(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqrshl_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqrshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -965,14 +1013,16 @@ svuint32_t test_svqrshl_n_u32_x(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqrshl_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqrshl_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqrshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qshl.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qshl.c
index 3fde6854a1b4d..a5c396b5a74b3 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qshl.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qshl.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -477,14 +479,16 @@ svuint64_t test_svqshl_u64_x(svbool_t pg, svuint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svqshl_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -499,7 +503,8 @@ svint8_t test_svqshl_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqshl_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -507,7 +512,8 @@ svint8_t test_svqshl_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -522,7 +528,8 @@ svint16_t test_svqshl_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqshl_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -530,7 +537,8 @@ svint16_t test_svqshl_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -545,7 +553,8 @@ svint32_t test_svqshl_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqshl_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -553,7 +562,8 @@ svint32_t test_svqshl_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -567,14 +577,16 @@ svint64_t test_svqshl_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqshl_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -589,7 +601,8 @@ svuint8_t test_svqshl_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqshl_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -597,7 +610,8 @@ svuint8_t test_svqshl_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -612,7 +626,8 @@ svuint16_t test_svqshl_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqshl_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -620,7 +635,8 @@ svuint16_t test_svqshl_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -635,7 +651,8 @@ svuint32_t test_svqshl_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqshl_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -643,7 +660,8 @@ svuint32_t test_svqshl_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -657,13 +675,15 @@ svuint64_t test_svqshl_n_u64_z(svbool_t pg, svuint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqshl_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -677,14 +697,16 @@ svint8_t test_svqshl_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqshl_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -698,14 +720,16 @@ svint16_t test_svqshl_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqshl_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -719,14 +743,16 @@ svint32_t test_svqshl_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqshl_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -739,13 +765,15 @@ svint64_t test_svqshl_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqshl_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -759,14 +787,16 @@ svuint8_t test_svqshl_n_u8_m(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqshl_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -780,14 +810,16 @@ svuint16_t test_svqshl_n_u16_m(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqshl_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -801,14 +833,16 @@ svuint32_t test_svqshl_n_u32_m(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqshl_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -821,13 +855,15 @@ svuint64_t test_svqshl_n_u64_m(svbool_t pg, svuint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqshl_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -841,14 +877,16 @@ svint8_t test_svqshl_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqshl_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -862,14 +900,16 @@ svint16_t test_svqshl_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqshl_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -883,14 +923,16 @@ svint32_t test_svqshl_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqshl_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -903,13 +945,15 @@ svint64_t test_svqshl_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqshl_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqshl_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -923,14 +967,16 @@ svuint8_t test_svqshl_n_u8_x(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqshl_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -944,14 +990,16 @@ svuint16_t test_svqshl_n_u16_x(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqshl_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -965,14 +1013,16 @@ svuint32_t test_svqshl_n_u32_x(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqshl_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqshl_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsub.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsub.c
index f669fd9d2cbaa..72451d33f28af 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsub.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsub.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -477,14 +479,16 @@ svuint64_t test_svqsub_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svqsub_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -499,7 +503,8 @@ svint8_t test_svqsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqsub_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -507,7 +512,8 @@ svint8_t test_svqsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -522,7 +528,8 @@ svint16_t test_svqsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqsub_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -530,7 +537,8 @@ svint16_t test_svqsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -545,7 +553,8 @@ svint32_t test_svqsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqsub_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -553,7 +562,8 @@ svint32_t test_svqsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -567,14 +577,16 @@ svint64_t test_svqsub_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqsub_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -589,7 +601,8 @@ svuint8_t test_svqsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqsub_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -597,7 +610,8 @@ svuint8_t test_svqsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -612,7 +626,8 @@ svuint16_t test_svqsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqsub_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -620,7 +635,8 @@ svuint16_t test_svqsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -635,7 +651,8 @@ svuint32_t test_svqsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqsub_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -643,7 +660,8 @@ svuint32_t test_svqsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -657,13 +675,15 @@ svuint64_t test_svqsub_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svqsub_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -677,14 +697,16 @@ svint8_t test_svqsub_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqsub_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -698,14 +720,16 @@ svint16_t test_svqsub_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqsub_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -719,14 +743,16 @@ svint32_t test_svqsub_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqsub_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -739,13 +765,15 @@ svint64_t test_svqsub_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqsub_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -759,14 +787,16 @@ svuint8_t test_svqsub_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqsub_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -780,14 +810,16 @@ svuint16_t test_svqsub_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqsub_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -801,14 +833,16 @@ svuint32_t test_svqsub_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqsub_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -821,13 +855,15 @@ svuint64_t test_svqsub_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svqsub_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -841,14 +877,16 @@ svint8_t test_svqsub_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqsub_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -862,14 +900,16 @@ svint16_t test_svqsub_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqsub_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -883,14 +923,16 @@ svint32_t test_svqsub_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqsub_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -903,13 +945,15 @@ svint64_t test_svqsub_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqsub_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svqsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -923,14 +967,16 @@ svuint8_t test_svqsub_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqsub_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -944,14 +990,16 @@ svuint16_t test_svqsub_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqsub_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -965,14 +1013,16 @@ svuint32_t test_svqsub_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqsub_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsubr.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsubr.c
index 69784968550f6..d59d16490241d 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsubr.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qsubr.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -477,14 +479,16 @@ svuint64_t test_svqsubr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svqsubr_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -499,7 +503,8 @@ svint8_t test_svqsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqsubr_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -507,7 +512,8 @@ svint8_t test_svqsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -522,7 +528,8 @@ svint16_t test_svqsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqsubr_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -530,7 +537,8 @@ svint16_t test_svqsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -545,7 +553,8 @@ svint32_t test_svqsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqsubr_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -553,7 +562,8 @@ svint32_t test_svqsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -567,14 +577,16 @@ svint64_t test_svqsubr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqsubr_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -589,7 +601,8 @@ svuint8_t test_svqsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqsubr_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -597,7 +610,8 @@ svuint8_t test_svqsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -612,7 +626,8 @@ svuint16_t test_svqsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqsubr_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -620,7 +635,8 @@ svuint16_t test_svqsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -635,7 +651,8 @@ svuint32_t test_svqsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqsubr_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -643,7 +660,8 @@ svuint32_t test_svqsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -657,13 +675,15 @@ svuint64_t test_svqsubr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svqsubr_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -677,14 +697,16 @@ svint8_t test_svqsubr_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqsubr_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -698,14 +720,16 @@ svint16_t test_svqsubr_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqsubr_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -719,14 +743,16 @@ svint32_t test_svqsubr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqsubr_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -739,13 +765,15 @@ svint64_t test_svqsubr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqsubr_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -759,14 +787,16 @@ svuint8_t test_svqsubr_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqsubr_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -780,14 +810,16 @@ svuint16_t test_svqsubr_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqsubr_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -801,14 +833,16 @@ svuint32_t test_svqsubr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqsubr_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -821,13 +855,15 @@ svuint64_t test_svqsubr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
// CHECK-LABEL: @test_svqsubr_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -841,14 +877,16 @@ svint8_t test_svqsubr_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svqsubr_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -862,14 +900,16 @@ svint16_t test_svqsubr_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svqsubr_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -883,14 +923,16 @@ svint32_t test_svqsubr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svqsubr_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -903,13 +945,15 @@ svint64_t test_svqsubr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svqsubr_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svqsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -923,14 +967,16 @@ svuint8_t test_svqsubr_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svqsubr_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -944,14 +990,16 @@ svuint16_t test_svqsubr_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svqsubr_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -965,14 +1013,16 @@ svuint32_t test_svqsubr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svqsubr_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svqsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnb.c
index af12457ee4799..95ee9b8162839 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svraddhnb_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svraddhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svraddhnb_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svraddhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svraddhnb_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svraddhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svraddhnb_n_s64(svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svraddhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svraddhnb_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svraddhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svraddhnb_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svraddhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnt.c
index 8d42071762085..6c91d3e15e901 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_raddhnt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svraddhnt_u64(svuint32_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svraddhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svraddhnt_n_s16(svint8_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svraddhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svraddhnt_n_s32(svint16_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svraddhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svraddhnt_n_s64(svint32_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svraddhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.raddhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svraddhnt_n_u16(svuint8_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svraddhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.raddhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svraddhnt_n_u32(svuint16_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svraddhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svraddhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.raddhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rhadd.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rhadd.c
index d37487073a7e8..bcf2a13890968 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rhadd.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rhadd.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -165,13 +167,15 @@ svuint64_t test_svrhadd_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svrhadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -185,14 +189,16 @@ svint8_t test_svrhadd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrhadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -206,14 +212,16 @@ svint16_t test_svrhadd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrhadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -227,14 +235,16 @@ svint32_t test_svrhadd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrhadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -247,13 +257,15 @@ svint64_t test_svrhadd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrhadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -267,14 +279,16 @@ svuint8_t test_svrhadd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svrhadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -288,14 +302,16 @@ svuint16_t test_svrhadd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svrhadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -309,14 +325,16 @@ svuint32_t test_svrhadd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svrhadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -494,14 +512,16 @@ svuint64_t test_svrhadd_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svrhadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -516,7 +536,8 @@ svint8_t test_svrhadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrhadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -524,7 +545,8 @@ svint8_t test_svrhadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -539,7 +561,8 @@ svint16_t test_svrhadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrhadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -547,7 +570,8 @@ svint16_t test_svrhadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -562,7 +586,8 @@ svint32_t test_svrhadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrhadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -570,7 +595,8 @@ svint32_t test_svrhadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -584,14 +610,16 @@ svint64_t test_svrhadd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrhadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -606,7 +634,8 @@ svuint8_t test_svrhadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svrhadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -614,7 +643,8 @@ svuint8_t test_svrhadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -629,7 +659,8 @@ svuint16_t test_svrhadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svrhadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -637,7 +668,8 @@ svuint16_t test_svrhadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -652,7 +684,8 @@ svuint32_t test_svrhadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svrhadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -660,7 +693,8 @@ svuint32_t test_svrhadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -823,13 +857,15 @@ svuint64_t test_svrhadd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svrhadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -843,14 +879,16 @@ svint8_t test_svrhadd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrhadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -864,14 +902,16 @@ svint16_t test_svrhadd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrhadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -885,14 +925,16 @@ svint32_t test_svrhadd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrhadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -905,13 +947,15 @@ svint64_t test_svrhadd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrhadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svrhadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -925,14 +969,16 @@ svuint8_t test_svrhadd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svrhadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -946,14 +992,16 @@ svuint16_t test_svrhadd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svrhadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -967,14 +1015,16 @@ svuint32_t test_svrhadd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svrhadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svrhadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rshl.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rshl.c
index 6df89c4aace80..a12fd5ccf47d7 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rshl.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rshl.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -477,14 +479,16 @@ svuint64_t test_svrshl_u64_x(svbool_t pg, svuint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svrshl_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_s8_zu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -499,7 +503,8 @@ svint8_t test_svrshl_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrshl_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -507,7 +512,8 @@ svint8_t test_svrshl_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s16_zu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -522,7 +528,8 @@ svint16_t test_svrshl_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrshl_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -530,7 +537,8 @@ svint16_t test_svrshl_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s32_zu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -545,7 +553,8 @@ svint32_t test_svrshl_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrshl_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -553,7 +562,8 @@ svint32_t test_svrshl_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s64_zu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -567,14 +577,16 @@ svint64_t test_svrshl_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrshl_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -589,7 +601,8 @@ svuint8_t test_svrshl_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrshl_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -597,7 +610,8 @@ svuint8_t test_svrshl_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -612,7 +626,8 @@ svuint16_t test_svrshl_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrshl_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -620,7 +635,8 @@ svuint16_t test_svrshl_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -635,7 +651,8 @@ svuint32_t test_svrshl_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrshl_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -643,7 +660,8 @@ svuint32_t test_svrshl_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -657,13 +675,15 @@ svuint64_t test_svrshl_n_u64_z(svbool_t pg, svuint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrshl_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_s8_mu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -677,14 +697,16 @@ svint8_t test_svrshl_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrshl_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s16_mu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -698,14 +720,16 @@ svint16_t test_svrshl_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrshl_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s32_mu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -719,14 +743,16 @@ svint32_t test_svrshl_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrshl_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s64_mu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -739,13 +765,15 @@ svint64_t test_svrshl_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrshl_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -759,14 +787,16 @@ svuint8_t test_svrshl_n_u8_m(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrshl_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -780,14 +810,16 @@ svuint16_t test_svrshl_n_u16_m(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrshl_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -801,14 +833,16 @@ svuint32_t test_svrshl_n_u32_m(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrshl_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -821,13 +855,15 @@ svuint64_t test_svrshl_n_u64_m(svbool_t pg, svuint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrshl_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_s8_xu10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.srshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -841,14 +877,16 @@ svint8_t test_svrshl_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrshl_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s16_xu10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.srshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -862,14 +900,16 @@ svint16_t test_svrshl_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrshl_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s32_xu10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.srshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -883,14 +923,16 @@ svint32_t test_svrshl_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrshl_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_s64_xu10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.srshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -903,13 +945,15 @@ svint64_t test_svrshl_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrshl_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svrshl_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.urshl.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -923,14 +967,16 @@ svuint8_t test_svrshl_n_u8_x(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svrshl_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.urshl.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -944,14 +990,16 @@ svuint16_t test_svrshl_n_u16_x(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrshl_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.urshl.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -965,14 +1013,16 @@ svuint32_t test_svrshl_n_u32_x(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrshl_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svrshl_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.urshl.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnb.c
index 10f81f0e65538..52eb55dafcded 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svrsubhnb_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svrsubhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svrsubhnb_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svrsubhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svrsubhnb_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svrsubhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svrsubhnb_n_s64(svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svrsubhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svrsubhnb_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svrsubhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svrsubhnb_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svrsubhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnt.c
index 27084aaba0d80..97ccb87ef8e32 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_rsubhnt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svrsubhnt_u64(svuint32_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svrsubhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svrsubhnt_n_s16(svint8_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svrsubhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svrsubhnt_n_s32(svint16_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svrsubhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svrsubhnt_n_s64(svint32_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svrsubhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.rsubhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svrsubhnt_n_u16(svuint8_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svrsubhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.rsubhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svrsubhnt_n_u32(svuint16_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svrsubhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z20test_svrsubhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.rsubhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclb.c
index 821bb94873fe2..7eb5427990c63 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -51,13 +53,15 @@ svuint64_t test_svsbclb_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svsbclb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclb_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -70,13 +74,15 @@ svuint32_t test_svsbclb_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svsbclb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclb_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclt.c
index 85d62531288c6..f950f70e8ae02 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sbclt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -51,13 +53,15 @@ svuint64_t test_svsbclt_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svsbclt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclt_n_u32u12__SVUint32_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sbclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -70,13 +74,15 @@ svuint32_t test_svsbclt_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svsbclt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsbclt_n_u64u12__SVUint64_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sbclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sqadd.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sqadd.c
index bf162093c7ded..985e0a64e6498 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sqadd.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sqadd.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -91,13 +93,15 @@ svuint64_t test_svsqadd_u64_m(svbool_t pg, svuint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svsqadd_n_u8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsqadd_n_u8_mu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -111,14 +115,16 @@ svuint8_t test_svsqadd_n_u8_m(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsqadd_n_u16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u16_mu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -132,14 +138,16 @@ svuint16_t test_svsqadd_n_u16_m(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsqadd_n_u32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u32_mu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -153,14 +161,16 @@ svuint32_t test_svsqadd_n_u32_m(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsqadd_n_u64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u64_mu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -255,14 +265,16 @@ svuint64_t test_svsqadd_u64_z(svbool_t pg, svuint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svsqadd_n_u8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svsqadd_n_u8_zu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -277,7 +289,8 @@ svuint8_t test_svsqadd_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsqadd_n_u16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -285,7 +298,8 @@ svuint8_t test_svsqadd_n_u8_z(svbool_t pg, svuint8_t op1, int8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u16_zu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -300,7 +314,8 @@ svuint16_t test_svsqadd_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsqadd_n_u32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -308,7 +323,8 @@ svuint16_t test_svsqadd_n_u16_z(svbool_t pg, svuint16_t op1, int16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u32_zu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -323,7 +339,8 @@ svuint32_t test_svsqadd_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsqadd_n_u64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -331,7 +348,8 @@ svuint32_t test_svsqadd_n_u32_z(svbool_t pg, svuint32_t op1, int32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u64_zu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -420,13 +438,15 @@ svuint64_t test_svsqadd_u64_x(svbool_t pg, svuint64_t op1, svint64_t op2)
// CHECK-LABEL: @test_svsqadd_n_u8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsqadd_n_u8_xu10__SVBool_tu11__SVUint8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.usqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -440,14 +460,16 @@ svuint8_t test_svsqadd_n_u8_x(svbool_t pg, svuint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsqadd_n_u16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u16_xu10__SVBool_tu12__SVUint16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -461,14 +483,16 @@ svuint16_t test_svsqadd_n_u16_x(svbool_t pg, svuint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsqadd_n_u32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u32_xu10__SVBool_tu12__SVUint32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -482,14 +506,16 @@ svuint32_t test_svsqadd_n_u32_x(svbool_t pg, svuint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsqadd_n_u64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svsqadd_n_u64_xu10__SVBool_tu12__SVUint64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnb.c
index 6aac7783901ee..58ddaee18409d 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svsubhnb_u64(svuint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svsubhnb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_s16u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svsubhnb_n_s16(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsubhnb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_s32u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svsubhnb_n_s32(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsubhnb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_s64u11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svsubhnb_n_s64(svint64_t op1, int64_t op2)
// CHECK-LABEL: @test_svsubhnb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_u16u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svsubhnb_n_u16(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsubhnb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_u32u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svsubhnb_n_u32(svuint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svsubhnb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnb_n_u64u12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnt.c
index 8a00938b6aab5..661a65b767a50 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subhnt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint32_t test_svsubhnt_u64(svuint32_t op1, svuint64_t op2, svuint64_t op3)
// CHECK-LABEL: @test_svsubhnt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint8_t test_svsubhnt_n_s16(svint8_t op1, svint16_t op2, int16_t op3)
// CHECK-LABEL: @test_svsubhnt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint16_t test_svsubhnt_n_s32(svint16_t op1, svint32_t op2, int32_t op3)
// CHECK-LABEL: @test_svsubhnt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint32_t test_svsubhnt_n_s64(svint32_t op1, svint64_t op2, int64_t op3)
// CHECK-LABEL: @test_svsubhnt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint8_t test_svsubhnt_n_u16(svuint8_t op1, svuint16_t op2, uint16_t op3)
// CHECK-LABEL: @test_svsubhnt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint16_t test_svsubhnt_n_u32(svuint16_t op1, svuint32_t op2, uint32_t op3)
// CHECK-LABEL: @test_svsubhnt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublb.c
index 93845fcbf7604..673768ebb6fea 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svsublb_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svsublb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svsublb_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsublb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svsublb_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsublb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svsublb_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsublb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svsublb_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsublb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svsublb_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsublb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublb_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublbt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublbt.c
index 674b2f7c64b5a..345ea12c3c558 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublbt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublbt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svsublbt_s64(svint32_t op1, svint32_t op2)
// CHECK-LABEL: @test_svsublbt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsublbt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svsublbt_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsublbt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsublbt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svsublbt_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsublbt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsublbt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublt.c
index 25742e28ff143..73d62f41795dd 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_sublt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svsublt_u64(svuint32_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svsublt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svsublt_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsublt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svsublt_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsublt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svsublt_n_s64(svint32_t op1, int32_t op2)
// CHECK-LABEL: @test_svsublt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_u16u11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usublt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svsublt_n_u16(svuint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsublt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_u32u12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usublt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svsublt_n_u32(svuint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsublt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsublt_n_u64u12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usublt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subltb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subltb.c
index 89cb511ca6684..f350e11031636 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subltb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subltb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -68,13 +70,15 @@ svint64_t test_svsubltb_s64(svint32_t op1, svint32_t op2)
// CHECK-LABEL: @test_svsubltb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubltb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubltb_n_s16u10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubltb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -87,13 +91,15 @@ svint16_t test_svsubltb_n_s16(svint8_t op1, int8_t op2)
// CHECK-LABEL: @test_svsubltb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubltb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubltb_n_s32u11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubltb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -106,13 +112,15 @@ svint32_t test_svsubltb_n_s32(svint16_t op1, int16_t op2)
// CHECK-LABEL: @test_svsubltb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubltb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svsubltb_n_s64u11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubltb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwb.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwb.c
index b3519e4110b2b..268f4dbda91e3 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwb.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwb.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svsubwb_u64(svuint64_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svsubwb_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svsubwb_n_s16(svint16_t op1, int8_t op2)
// CHECK-LABEL: @test_svsubwb_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svsubwb_n_s32(svint32_t op1, int16_t op2)
// CHECK-LABEL: @test_svsubwb_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svsubwb_n_s64(svint64_t op1, int32_t op2)
// CHECK-LABEL: @test_svsubwb_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svsubwb_n_u16(svuint16_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsubwb_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svsubwb_n_u32(svuint32_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsubwb_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwb_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwt.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwt.c
index 3b2a2da7710ce..037176b4bbb85 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwt.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_subwt.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -119,13 +121,15 @@ svuint64_t test_svsubwt_u64(svuint64_t op1, svuint32_t op2)
// CHECK-LABEL: @test_svsubwt_n_s16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_s16u11__SVInt16_ta(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.ssubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -138,13 +142,15 @@ svint16_t test_svsubwt_n_s16(svint16_t op1, int8_t op2)
// CHECK-LABEL: @test_svsubwt_n_s32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_s32u11__SVInt32_ts(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.ssubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -157,13 +163,15 @@ svint32_t test_svsubwt_n_s32(svint32_t op1, int16_t op2)
// CHECK-LABEL: @test_svsubwt_n_s64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_s64u11__SVInt64_ti(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.ssubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
@@ -176,13 +184,15 @@ svint64_t test_svsubwt_n_s64(svint64_t op1, int32_t op2)
// CHECK-LABEL: @test_svsubwt_n_u16(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_u16u12__SVUint16_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.usubwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
//
@@ -195,13 +205,15 @@ svuint16_t test_svsubwt_n_u16(svuint16_t op1, uint8_t op2)
// CHECK-LABEL: @test_svsubwt_n_u32(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_u32u12__SVUint32_tt(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usubwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
//
@@ -214,13 +226,15 @@ svuint32_t test_svsubwt_n_u32(svuint32_t op1, uint16_t op2)
// CHECK-LABEL: @test_svsubwt_n_u64(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z18test_svsubwt_n_u64u12__SVUint64_tj(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.usubwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
//
diff --git a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_uqadd.c b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_uqadd.c
index 35fc04b811914..a61ab70e3cee5 100644
--- a/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_uqadd.c
+++ b/clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_uqadd.c
@@ -1,4 +1,6 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2 -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
@@ -91,13 +93,15 @@ svint64_t test_svuqadd_u64_m(svbool_t pg, svint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svuqadd_n_s8_m(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svuqadd_n_s8_mu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -111,14 +115,16 @@ svint8_t test_svuqadd_n_s8_m(svbool_t pg, svint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svuqadd_n_s16_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s16_mu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -132,14 +138,16 @@ svint16_t test_svuqadd_n_s16_m(svbool_t pg, svint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svuqadd_n_s32_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s32_mu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -153,14 +161,16 @@ svint32_t test_svuqadd_n_s32_m(svbool_t pg, svint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svuqadd_n_s64_m(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s64_mu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
@@ -255,14 +265,16 @@ svint64_t test_svuqadd_u64_z(svbool_t pg, svint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svuqadd_n_s8_z(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z19test_svuqadd_n_s8_zu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP2]]
@@ -277,7 +289,8 @@ svint8_t test_svuqadd_n_s8_z(svbool_t pg, svint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svuqadd_n_s16_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -285,7 +298,8 @@ svint8_t test_svuqadd_n_s8_z(svbool_t pg, svint8_t op1, uint8_t op2)
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s16_zu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP3]]
@@ -300,7 +314,8 @@ svint16_t test_svuqadd_n_s16_z(svbool_t pg, svint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svuqadd_n_s32_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -308,7 +323,8 @@ svint16_t test_svuqadd_n_s16_z(svbool_t pg, svint16_t op1, uint16_t op2)
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s32_zu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP3]]
@@ -323,7 +339,8 @@ svint32_t test_svuqadd_n_s32_z(svbool_t pg, svint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svuqadd_n_s64_z(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -331,7 +348,8 @@ svint32_t test_svuqadd_n_s32_z(svbool_t pg, svint32_t op1, uint32_t op2)
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s64_zu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP3]]
@@ -420,13 +438,15 @@ svint64_t test_svuqadd_u64_x(svbool_t pg, svint64_t op1, svuint64_t op2)
// CHECK-LABEL: @test_svuqadd_n_s8_x(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z19test_svuqadd_n_s8_xu10__SVBool_tu10__SVInt8_th(
// CPP-CHECK-NEXT: entry:
-// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.suqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i8> [[TMP1]]
//
@@ -440,14 +460,16 @@ svint8_t test_svuqadd_n_s8_x(svbool_t pg, svint8_t op1, uint8_t op2)
// CHECK-LABEL: @test_svuqadd_n_s16_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s16_xu10__SVBool_tu11__SVInt16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.suqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
//
@@ -461,14 +483,16 @@ svint16_t test_svuqadd_n_s16_x(svbool_t pg, svint16_t op1, uint16_t op2)
// CHECK-LABEL: @test_svuqadd_n_s32_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s32_xu10__SVBool_tu11__SVInt32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.suqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
//
@@ -482,14 +506,16 @@ svint32_t test_svuqadd_n_s32_x(svbool_t pg, svint32_t op1, uint32_t op2)
// CHECK-LABEL: @test_svuqadd_n_s64_x(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z20test_svuqadd_n_s64_xu10__SVBool_tu11__SVInt64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
+// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.suqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
//
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
index 6abc8caaab811..f12854af9ab23 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -438,6 +438,18 @@ static Optional<Instruction *> instCombineSVEDup(InstCombiner &IC,
return IC.replaceInstUsesWith(II, Insert);
}
+static Optional<Instruction *> instCombineSVEDupX(InstCombiner &IC,
+ IntrinsicInst &II) {
+ // Replace DupX with a regular IR splat.
+ IRBuilder<> Builder(II.getContext());
+ Builder.SetInsertPoint(&II);
+ auto *RetTy = cast<ScalableVectorType>(II.getType());
+ Value *Splat =
+ Builder.CreateVectorSplat(RetTy->getElementCount(), II.getArgOperand(0));
+ Splat->takeName(&II);
+ return IC.replaceInstUsesWith(II, Splat);
+}
+
static Optional<Instruction *> instCombineSVECmpNE(InstCombiner &IC,
IntrinsicInst &II) {
LLVMContext &Ctx = II.getContext();
@@ -455,12 +467,9 @@ static Optional<Instruction *> instCombineSVECmpNE(InstCombiner &IC,
return None;
// Check that we have a compare of zero..
- auto *DupX = dyn_cast<IntrinsicInst>(II.getArgOperand(2));
- if (!DupX || DupX->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
- return None;
-
- auto *DupXArg = dyn_cast<ConstantInt>(DupX->getArgOperand(0));
- if (!DupXArg || !DupXArg->isZero())
+ auto *SplatValue =
+ dyn_cast_or_null<ConstantInt>(getSplatValue(II.getArgOperand(2)));
+ if (!SplatValue || !SplatValue->isZero())
return None;
// ..against a dupq
@@ -693,14 +702,11 @@ static Optional<Instruction *> instCombineSVEVectorMul(InstCombiner &IC,
IRBuilder<> Builder(II.getContext());
Builder.SetInsertPoint(&II);
- // Return true if a given instruction is an aarch64_sve_dup_x intrinsic call
- // with a unit splat value, false otherwise.
- auto IsUnitDupX = [](auto *I) {
- auto *IntrI = dyn_cast<IntrinsicInst>(I);
- if (!IntrI || IntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
+ // Return true if a given instruction is a unit splat value, false otherwise.
+ auto IsUnitSplat = [](auto *I) {
+ auto *SplatValue = getSplatValue(I);
+ if (!SplatValue)
return false;
-
- auto *SplatValue = IntrI->getOperand(0);
return match(SplatValue, m_FPOne()) || match(SplatValue, m_One());
};
@@ -717,10 +723,10 @@ static Optional<Instruction *> instCombineSVEVectorMul(InstCombiner &IC,
// The OpMultiplier variable should always point to the dup (if any), so
// swap if necessary.
- if (IsUnitDup(OpMultiplicand) || IsUnitDupX(OpMultiplicand))
+ if (IsUnitDup(OpMultiplicand) || IsUnitSplat(OpMultiplicand))
std::swap(OpMultiplier, OpMultiplicand);
- if (IsUnitDupX(OpMultiplier)) {
+ if (IsUnitSplat(OpMultiplier)) {
// [f]mul pg (dupx 1) %n => %n
OpMultiplicand->takeName(&II);
return IC.replaceInstUsesWith(II, OpMultiplicand);
@@ -767,13 +773,9 @@ static Optional<Instruction *> instCombineSVETBL(InstCombiner &IC,
auto *OpIndices = II.getOperand(1);
VectorType *VTy = cast<VectorType>(II.getType());
- // Check whether OpIndices is an aarch64_sve_dup_x intrinsic call with
- // constant splat value < minimal element count of result.
- auto *DupXIntrI = dyn_cast<IntrinsicInst>(OpIndices);
- if (!DupXIntrI || DupXIntrI->getIntrinsicID() != Intrinsic::aarch64_sve_dup_x)
- return None;
-
- auto *SplatValue = dyn_cast<ConstantInt>(DupXIntrI->getOperand(0));
+ // Check whether OpIndices is a constant splat value < minimal element count
+ // of result.
+ auto *SplatValue = dyn_cast_or_null<ConstantInt>(getSplatValue(OpIndices));
if (!SplatValue ||
SplatValue->getValue().uge(VTy->getElementCount().getKnownMinValue()))
return None;
@@ -801,6 +803,8 @@ AArch64TTIImpl::instCombineIntrinsic(InstCombiner &IC,
return instCombineConvertFromSVBool(IC, II);
case Intrinsic::aarch64_sve_dup:
return instCombineSVEDup(IC, II);
+ case Intrinsic::aarch64_sve_dup_x:
+ return instCombineSVEDupX(IC, II);
case Intrinsic::aarch64_sve_cmpne:
case Intrinsic::aarch64_sve_cmpne_wide:
return instCombineSVECmpNE(IC, II);
diff --git a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
index d13f76c2be301..9bc48d8777954 100644
--- a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
+++ b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
@@ -53,8 +53,7 @@ define <vscale x 8 x half> @idempotent_fmul_two_dups(<vscale x 8 x i1> %pg, <vsc
; Edge case -- make sure that the case where we're fmultiplying two dups
; together is sane.
; CHECK-LABEL: @idempotent_fmul_two_dups(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 0xH3C00)
-; CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
+; CHECK-NEXT: ret <vscale x 8 x half> shufflevector (<vscale x 8 x half> insertelement (<vscale x 8 x half> poison, half 0xH3C00, i32 0), <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer)
;
%1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 1.0)
%2 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 1.0)
@@ -65,9 +64,8 @@ define <vscale x 8 x half> @idempotent_fmul_two_dups(<vscale x 8 x i1> %pg, <vsc
; Non-idempotent fmuls -- we don't expect these to be optimised out.
define <vscale x 8 x half> @non_idempotent_fmul_f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a) #0 {
; CHECK-LABEL: @non_idempotent_fmul_f16(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 0xH4000)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 8 x half> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x half> [[A:%.*]], <vscale x 8 x half> shufflevector (<vscale x 8 x half> insertelement (<vscale x 8 x half> poison, half 0xH4000, i32 0), <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 8 x half> [[TMP1]]
;
%1 = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half 2.0)
%2 = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> %pg, <vscale x 8 x half> %a, <vscale x 8 x half> %1)
@@ -76,9 +74,8 @@ define <vscale x 8 x half> @non_idempotent_fmul_f16(<vscale x 8 x i1> %pg, <vsca
define <vscale x 4 x float> @non_idempotent_fmul_f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a) #0 {
; CHECK-LABEL: @non_idempotent_fmul_f32(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 2.000000e+00)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x float> [[A:%.*]], <vscale x 4 x float> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 4 x float> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x float> [[A:%.*]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 2.000000e+00, i32 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 4 x float> [[TMP1]]
;
%1 = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float 2.0)
%2 = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> %pg, <vscale x 4 x float> %a, <vscale x 4 x float> %1)
@@ -87,9 +84,8 @@ define <vscale x 4 x float> @non_idempotent_fmul_f32(<vscale x 4 x i1> %pg, <vsc
define <vscale x 2 x double> @non_idempotent_fmul_f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) #0 {
; CHECK-LABEL: @non_idempotent_fmul_f64(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 2.000000e+00)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x double> [[A:%.*]], <vscale x 2 x double> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 2 x double> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x double> [[A:%.*]], <vscale x 2 x double> shufflevector (<vscale x 2 x double> insertelement (<vscale x 2 x double> poison, double 2.000000e+00, i32 0), <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 2 x double> [[TMP1]]
;
%1 = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 2.0)
%2 = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a, <vscale x 2 x double> %1)
diff --git a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll
index 8e10050714a6f..e07fa9263a355 100644
--- a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll
+++ b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll
@@ -53,8 +53,7 @@ define <vscale x 8 x i16> @idempotent_mul_two_dups(<vscale x 8 x i1> %pg, <vscal
; Edge case -- make sure that the case where we're multiplying two dups
; together is sane.
; CHECK-LABEL: @idempotent_mul_two_dups(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
-; CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
+; CHECK-NEXT: ret <vscale x 8 x i16> shufflevector (<vscale x 8 x i16> insertelement (<vscale x 8 x i16> poison, i16 1, i32 0), <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer)
;
%1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
%2 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 1)
@@ -65,9 +64,8 @@ define <vscale x 8 x i16> @idempotent_mul_two_dups(<vscale x 8 x i1> %pg, <vscal
; Non-idempotent muls -- we don't expect these to be optimised out.
define <vscale x 8 x i16> @non_idempotent_mul_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) #0 {
; CHECK-LABEL: @non_idempotent_mul_i16(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 2)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x i16> [[A:%.*]], <vscale x 8 x i16> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 8 x i16> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[PG:%.*]], <vscale x 8 x i16> [[A:%.*]], <vscale x 8 x i16> shufflevector (<vscale x 8 x i16> insertelement (<vscale x 8 x i16> poison, i16 2, i32 0), <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 8 x i16> [[TMP1]]
;
%1 = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 2)
%2 = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a, <vscale x 8 x i16> %1)
@@ -76,9 +74,8 @@ define <vscale x 8 x i16> @non_idempotent_mul_i16(<vscale x 8 x i1> %pg, <vscale
define <vscale x 4 x i32> @non_idempotent_mul_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: @non_idempotent_mul_i32(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 2)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x i32> [[A:%.*]], <vscale x 4 x i32> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 4 x i32> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x i32> [[A:%.*]], <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 2, i32 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 4 x i32> [[TMP1]]
;
%1 = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 2)
%2 = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a, <vscale x 4 x i32> %1)
@@ -87,9 +84,8 @@ define <vscale x 4 x i32> @non_idempotent_mul_i32(<vscale x 4 x i1> %pg, <vscale
define <vscale x 2 x i64> @non_idempotent_mul_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) #0 {
; CHECK-LABEL: @non_idempotent_mul_i64(
-; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 2)
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x i64> [[A:%.*]], <vscale x 2 x i64> [[TMP1]])
-; CHECK-NEXT: ret <vscale x 2 x i64> [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x i64> [[A:%.*]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 2, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+; CHECK-NEXT: ret <vscale x 2 x i64> [[TMP1]]
;
%1 = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 2)
%2 = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a, <vscale x 2 x i64> %1)
diff --git a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-dup.ll b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-dup.ll
index add9cfb56f1e4..668946f9eda44 100644
--- a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-dup.ll
+++ b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-dup.ll
@@ -40,6 +40,17 @@ define <vscale x 8 x i16> @dup_insertelement_0_convert(<vscale x 8 x i16> %v, i1
ret <vscale x 8 x i16> %insert
}
+define <vscale x 8 x i16> @dupx_splat_convert(i16 %s) #0 {
+; CHECK-LABEL: @dupx_splat_convert(
+; CHECK-NEXT: %.splatinsert = insertelement <vscale x 8 x i16> poison, i16 %s, i32 0
+; CHECK-NEXT: %splat = shufflevector <vscale x 8 x i16> %.splatinsert, <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
+; CHECK-NEXT: ret <vscale x 8 x i16> %splat
+ %splat = tail call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 %s)
+ ret <vscale x 8 x i16> %splat
+}
+
+declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16)
+
declare <vscale x 16 x i8> @llvm.aarch64.sve.dup.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i1>, i8)
declare <vscale x 8 x i16> @llvm.aarch64.sve.dup.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i1>, i16)
More information about the llvm-commits
mailing list