[clang] [llvm] [CLANG][LLVM][AArch64]SME2.1 intrinsics for MOVAZ tile to 2/4 vectors (PR #88710)

Mon Jun 3 08:57:30 PDT 2024

https://github.com/CarolineConcatto updated https://github.com/llvm/llvm-project/pull/88710

>From a4ba2e63ca1dc40422cdf398925cdb4a25f7ca83 Mon Sep 17 00:00:00 2001
From: Caroline Concatto <caroline.concatto at arm.com>
Date: Mon, 15 Apr 2024 10:35:03 +0000
Subject: [PATCH 1/5] [CLANG][LLVM][AArch64]SME2.1 intrinsics for MOVAZ tile to
 2/4 vectors

According to the specification in
ARM-software/acle#309 this adds the intrinsics

// Variants are also available for _za8_u8, _za16_s16, _za16_u16,
// _za16_f16, _za16_bf16, _za32_s32, _za32_u32, _za32_f32,
// _za64_s64, _za64_u64 and _za64_f64
svint8x2_t svreadz_hor_za8_s8_vg2(uint64_t tile, uint32_t slice)
__arm_streaming __arm_inout("za");

// Variants are also available for _za8_u8, _za16_s16, _za16_u16,
// _za16_f16, _za16_bf16, _za32_s32, _za32_u32, _za32_f32,
// _za64_s64, _za64_u64 and _za64_f64
svint8x4_t svreadz_hor_za8_s8_vg4(uint64_t tile, uint32_t slice)
__arm_streaming __arm_inout("za");

// Variants are also available for _za8_u8, _za16_s16, _za16_u16,
// _za16_f16, _za16_bf16, _za32_s32, _za32_u32, _za32_f32,
// _za64_s64, _za64_u64 and _za64_f64
svint8x2_t svreadz_ver_za8_s8_vg2(uint64_t tile, uint32_t slice)
__arm_streaming __arm_inout("za");

// Variants are also available for _za8_u8, _za16_s16, _za16_u16,
// _za16_f16, _za16_bf16, _za32_s32, _za32_u32, _za32_f32,
// _za64_s64, _za64_u64 and _za64_f64
svint8x4_t svreadz_ver_za8_s8_vg4(uint64_t tile, uint32_t slice)
__arm_streaming __arm_inout("za");
---
 clang/include/clang/Basic/arm_sme.td          |   23 +
 .../acle_sme2p1_movaz.c                       | 1414 +++++++++++++++++
 llvm/include/llvm/IR/IntrinsicsAArch64.td     |   20 +-
 .../Target/AArch64/AArch64ISelDAGToDAG.cpp    |   99 +-
 .../Target/AArch64/AArch64ISelLowering.cpp    |   49 +
 llvm/lib/Target/AArch64/AArch64ISelLowering.h |    3 +
 llvm/lib/Target/AArch64/SMEInstrFormats.td    |   27 +
 .../AArch64/sme2p1-intrinsics-movaz.ll        |  457 ++++++
 .../CodeGen/AArch64/sme2p1-intrinsics-movaz.s |    0
 9 files changed, 2090 insertions(+), 2 deletions(-)
 create mode 100644 clang/test/CodeGen/aarch64-sme2p1-intrinsics/acle_sme2p1_movaz.c
 create mode 100644 llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.ll
 create mode 100644 llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.s

diff --git a/clang/include/clang/Basic/arm_sme.td b/clang/include/clang/Basic/arm_sme.td
index 1ac6d5170ea28..44bb32916d86b 100644
--- a/clang/include/clang/Basic/arm_sme.td
+++ b/clang/include/clang/Basic/arm_sme.td
@@ -674,3 +674,26 @@ let TargetGuard = "sme2" in {
   def SVLUTI2_LANE_ZT_X2 : Inst<"svluti2_lane_zt_{d}_x2", "2.di[i", "cUcsUsiUibhf", MergeNone, "aarch64_sme_luti2_lane_zt_x2", [IsStreaming, IsInZT0], [ImmCheck<0, ImmCheck0_0>, ImmCheck<2, ImmCheck0_7>]>;
   def SVLUTI4_LANE_ZT_X2 : Inst<"svluti4_lane_zt_{d}_x2", "2.di[i", "cUcsUsiUibhf", MergeNone, "aarch64_sme_luti4_lane_zt_x2", [IsStreaming, IsInZT0], [ImmCheck<0, ImmCheck0_0>, ImmCheck<2, ImmCheck0_3>]>;
 }
+
+multiclass ZAReadz<string n_suffix, string vg_num, string t, string i_prefix, list<ImmCheck> ch> {
+  let TargetGuard = "sme2p1" in {
+    def NAME # _H : SInst<"svreadz_hor_" # n_suffix # "_{d}_vg" # vg_num, vg_num # "im", t,
+                          MergeNone, i_prefix # "_horiz_x" # vg_num,
+                          [IsStreaming, IsInOutZA], ch>;
+
+    def NAME # _V : SInst<"svreadz_ver_" # n_suffix # "_{d}_vg" # vg_num, vg_num # "im", t,
+                          MergeNone, i_prefix # "_vert_x" #vg_num,
+                          [IsStreaming, IsInOutZA], ch>;
+  }
+}
+
+defm SVREADZ_ZA8_X2  : ZAReadz<"za8",  "2", "cUc",   "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_0>]>;
+defm SVREADZ_ZA16_X2 : ZAReadz<"za16", "2", "sUshb", "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_1>]>;
+defm SVREADZ_ZA32_X2 : ZAReadz<"za32", "2", "iUif",  "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_3>]>;
+defm SVREADZ_ZA64_X2 : ZAReadz<"za64", "2", "lUld",  "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_7>]>;
+
+defm SVREADZ_ZA8_X4  : ZAReadz<"za8",  "4", "cUc",   "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_0>]>;
+defm SVREADZ_ZA16_X4 : ZAReadz<"za16", "4", "sUshb", "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_1>]>;
+defm SVREADZ_ZA32_X4 : ZAReadz<"za32", "4", "iUif",  "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_3>]>;
+defm SVREADZ_ZA64_X4 : ZAReadz<"za64", "4", "lUld",  "aarch64_sme_readz", [ImmCheck<0, ImmCheck0_7>]>;
+
diff --git a/clang/test/CodeGen/aarch64-sme2p1-intrinsics/acle_sme2p1_movaz.c b/clang/test/CodeGen/aarch64-sme2p1-intrinsics/acle_sme2p1_movaz.c
new file mode 100644
index 0000000000000..2b80c76e3d999
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2p1-intrinsics/acle_sme2p1_movaz.c
@@ -0,0 +1,1414 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 4
+ //RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2p1 -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2p1 -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2p1 -target-feature +bf16 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+
+#include <arm_sme.h>
+
+//
+// X2- hor
+// CHECK-LABEL: define dso_local <vscale x 32 x i8> @test_svreadz_hor_za8_s8_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i8> @_Z26test_svreadz_hor_za8_s8_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0:[0-9]+]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svint8x2_t test_svreadz_hor_za8_s8_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za8_s8_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x i8> @test_svreadz_hor_za8_u8_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i8> @_Z26test_svreadz_hor_za8_u8_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svuint8x2_t test_svreadz_hor_za8_u8_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za8_u8_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i16> @test_svreadz_hor_za16_s16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x2.nxv8i16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i16> @_Z28test_svreadz_hor_za16_s16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x2.nxv8i16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svint16x2_t test_svreadz_hor_za16_s16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_s16_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i16> @test_svreadz_hor_za16_u16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x2.nxv8i16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i16> @_Z28test_svreadz_hor_za16_u16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x2.nxv8i16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svuint16x2_t test_svreadz_hor_za16_u16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_u16_vg2(1, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x half> @test_svreadz_hor_za16_f16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.horiz.x2.nxv8f16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x half> @_Z28test_svreadz_hor_za16_f16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.horiz.x2.nxv8f16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+svfloat16x2_t test_svreadz_hor_za16_f16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_f16_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x bfloat> @test_svreadz_hor_za16_bf16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.horiz.x2.nxv8bf16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x bfloat> @_Z29test_svreadz_hor_za16_bf16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.horiz.x2.nxv8bf16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+svbfloat16x2_t test_svreadz_hor_za16_bf16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_bf16_vg2(1, slice);
+}
+
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i32> @test_svreadz_hor_za32_s32_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x2.nxv4i32(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i32> @_Z28test_svreadz_hor_za32_s32_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x2.nxv4i32(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svint32x2_t test_svreadz_hor_za32_s32_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za32_s32_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i32> @test_svreadz_hor_za32_u32_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x2.nxv4i32(i32 2, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i32> @_Z28test_svreadz_hor_za32_u32_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x2.nxv4i32(i32 2, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svuint32x2_t test_svreadz_hor_za32_u32_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za32_u32_vg2(2, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x float> @test_svreadz_hor_za32_f32_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.horiz.x2.nxv4f32(i32 3, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x float> @_Z28test_svreadz_hor_za32_f32_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.horiz.x2.nxv4f32(i32 3, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+svfloat32x2_t test_svreadz_hor_za32_f32_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za32_f32_vg2(3, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 4 x i64> @test_svreadz_hor_za64_s64_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x2.nxv2i64(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 4 x i64> @_Z28test_svreadz_hor_za64_s64_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x2.nxv2i64(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svint64x2_t test_svreadz_hor_za64_s64_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za64_s64_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 4 x i64> @test_svreadz_hor_za64_u64_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x2.nxv2i64(i32 4, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 4 x i64> @_Z28test_svreadz_hor_za64_u64_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x2.nxv2i64(i32 4, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svuint64x2_t test_svreadz_hor_za64_u64_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za64_u64_vg2(4, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 4 x double> @test_svreadz_hor_za64_f64_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.horiz.x2.nxv2f64(i32 7, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 4 x double> @_Z28test_svreadz_hor_za64_f64_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.horiz.x2.nxv2f64(i32 7, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+svfloat64x2_t test_svreadz_hor_za64_f64_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za64_f64_vg2(7, slice);
+}
+
+
+//
+// X2- ver
+//
+
+// CHECK-LABEL: define dso_local <vscale x 32 x i8> @test_svreadz_ver_za8_s8_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i8> @_Z26test_svreadz_ver_za8_s8_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svint8x2_t test_svreadz_ver_za8_s8_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za8_s8_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x i8> @test_svreadz_ver_za8_u8_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i8> @_Z26test_svreadz_ver_za8_u8_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x2.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svuint8x2_t test_svreadz_ver_za8_u8_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za8_u8_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i16> @test_svreadz_ver_za16_s16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x2.nxv8i16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i16> @_Z28test_svreadz_ver_za16_s16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x2.nxv8i16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svint16x2_t test_svreadz_ver_za16_s16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_s16_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i16> @test_svreadz_ver_za16_u16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x2.nxv8i16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i16> @_Z28test_svreadz_ver_za16_u16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x2.nxv8i16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svuint16x2_t test_svreadz_ver_za16_u16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_u16_vg2(1, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x half> @test_svreadz_ver_za16_f16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.vert.x2.nxv8f16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x half> @_Z28test_svreadz_ver_za16_f16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.vert.x2.nxv8f16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+svfloat16x2_t test_svreadz_ver_za16_f16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_f16_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x bfloat> @test_svreadz_ver_za16_bf16_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.vert.x2.nxv8bf16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x bfloat> @_Z29test_svreadz_ver_za16_bf16_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.vert.x2.nxv8bf16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+svbfloat16x2_t test_svreadz_ver_za16_bf16_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_bf16_vg2(1, slice);
+}
+
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i32> @test_svreadz_ver_za32_s32_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x2.nxv4i32(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i32> @_Z28test_svreadz_ver_za32_s32_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x2.nxv4i32(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svint32x2_t test_svreadz_ver_za32_s32_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za32_s32_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i32> @test_svreadz_ver_za32_u32_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x2.nxv4i32(i32 2, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i32> @_Z28test_svreadz_ver_za32_u32_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x2.nxv4i32(i32 2, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svuint32x2_t test_svreadz_ver_za32_u32_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za32_u32_vg2(2, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x float> @test_svreadz_ver_za32_f32_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.vert.x2.nxv4f32(i32 3, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x float> @_Z28test_svreadz_ver_za32_f32_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.vert.x2.nxv4f32(i32 3, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+svfloat32x2_t test_svreadz_ver_za32_f32_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za32_f32_vg2(3, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 4 x i64> @test_svreadz_ver_za64_s64_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x2.nxv2i64(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 4 x i64> @_Z28test_svreadz_ver_za64_s64_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x2.nxv2i64(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svint64x2_t test_svreadz_ver_za64_s64_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za64_s64_vg2(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 4 x i64> @test_svreadz_ver_za64_u64_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x2.nxv2i64(i32 4, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 4 x i64> @_Z28test_svreadz_ver_za64_u64_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x2.nxv2i64(i32 4, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svuint64x2_t test_svreadz_ver_za64_u64_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za64_u64_vg2(4, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 4 x double> @test_svreadz_ver_za64_f64_x2(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.vert.x2.nxv2f64(i32 7, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 4 x double> @_Z28test_svreadz_ver_za64_f64_x2j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.vert.x2.nxv2f64(i32 7, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+svfloat64x2_t test_svreadz_ver_za64_f64_x2(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za64_f64_vg2(7, slice);
+}
+
+
+//
+// X4 - hor
+// CHECK-LABEL: define dso_local <vscale x 64 x i8> @test_svreadz_hor_za8_s8_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 64 x i8> @_Z26test_svreadz_hor_za8_s8_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svint8x4_t test_svreadz_hor_za8_s8_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za8_s8_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 64 x i8> @test_svreadz_hor_za8_u8_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 64 x i8> @_Z26test_svreadz_hor_za8_u8_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.horiz.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svuint8x4_t test_svreadz_hor_za8_u8_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za8_u8_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x i16> @test_svreadz_hor_za16_s16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x4.nxv8i16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i16> @_Z28test_svreadz_hor_za16_s16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x4.nxv8i16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svint16x4_t test_svreadz_hor_za16_s16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_s16_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x i16> @test_svreadz_hor_za16_u16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x4.nxv8i16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i16> @_Z28test_svreadz_hor_za16_u16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.horiz.x4.nxv8i16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svuint16x4_t test_svreadz_hor_za16_u16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_u16_vg4(1, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x half> @test_svreadz_hor_za16_f16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.horiz.x4.nxv8f16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x half> @_Z28test_svreadz_hor_za16_f16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.horiz.x4.nxv8f16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+svfloat16x4_t test_svreadz_hor_za16_f16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_f16_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x bfloat> @test_svreadz_hor_za16_bf16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.horiz.x4.nxv8bf16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x bfloat> @_Z29test_svreadz_hor_za16_bf16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.horiz.x4.nxv8bf16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+svbfloat16x4_t test_svreadz_hor_za16_bf16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za16_bf16_vg4(1, slice);
+}
+
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i32> @test_svreadz_hor_za32_s32_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x4.nxv4i32(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i32> @_Z28test_svreadz_hor_za32_s32_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x4.nxv4i32(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svint32x4_t test_svreadz_hor_za32_s32_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za32_s32_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i32> @test_svreadz_hor_za32_u32_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x4.nxv4i32(i32 2, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i32> @_Z28test_svreadz_hor_za32_u32_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.horiz.x4.nxv4i32(i32 2, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svuint32x4_t test_svreadz_hor_za32_u32_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za32_u32_vg4(2, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x float> @test_svreadz_hor_za32_f32_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.horiz.x4.nxv4f32(i32 3, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x float> @_Z28test_svreadz_hor_za32_f32_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.horiz.x4.nxv4f32(i32 3, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+svfloat32x4_t test_svreadz_hor_za32_f32_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za32_f32_vg4(3, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i64> @test_svreadz_hor_za64_s64_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x4.nxv2i64(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i64> @_Z28test_svreadz_hor_za64_s64_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x4.nxv2i64(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svint64x4_t test_svreadz_hor_za64_s64_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za64_s64_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i64> @test_svreadz_hor_za64_u64_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x4.nxv2i64(i32 4, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i64> @_Z28test_svreadz_hor_za64_u64_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.horiz.x4.nxv2i64(i32 4, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svuint64x4_t test_svreadz_hor_za64_u64_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za64_u64_vg4(4, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x double> @test_svreadz_hor_za64_f64_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.horiz.x4.nxv2f64(i32 7, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x double> @_Z28test_svreadz_hor_za64_f64_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.horiz.x4.nxv2f64(i32 7, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+svfloat64x4_t test_svreadz_hor_za64_f64_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_hor_za64_f64_vg4(7, slice);
+}
+
+//
+// X4 - ver
+// CHECK-LABEL: define dso_local <vscale x 64 x i8> @test_svreadz_ver_za8_s8_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 64 x i8> @_Z26test_svreadz_ver_za8_s8_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svint8x4_t test_svreadz_ver_za8_s8_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za8_s8_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 64 x i8> @test_svreadz_ver_za8_u8_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 64 x i8> @_Z26test_svreadz_ver_za8_u8_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.readz.vert.x4.nxv16i8(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svuint8x4_t test_svreadz_ver_za8_u8_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za8_u8_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x i16> @test_svreadz_ver_za16_s16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x4.nxv8i16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i16> @_Z28test_svreadz_ver_za16_s16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x4.nxv8i16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svint16x4_t test_svreadz_ver_za16_s16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_s16_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x i16> @test_svreadz_ver_za16_u16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x4.nxv8i16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x i16> @_Z28test_svreadz_ver_za16_u16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.readz.vert.x4.nxv8i16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svuint16x4_t test_svreadz_ver_za16_u16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_u16_vg4(1, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x half> @test_svreadz_ver_za16_f16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.vert.x4.nxv8f16(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x half> @_Z28test_svreadz_ver_za16_f16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.readz.vert.x4.nxv8f16(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+svfloat16x4_t test_svreadz_ver_za16_f16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_f16_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 32 x bfloat> @test_svreadz_ver_za16_bf16_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.vert.x4.nxv8bf16(i32 1, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 32 x bfloat> @_Z29test_svreadz_ver_za16_bf16_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.readz.vert.x4.nxv8bf16(i32 1, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+svbfloat16x4_t test_svreadz_ver_za16_bf16_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za16_bf16_vg4(1, slice);
+}
+
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i32> @test_svreadz_ver_za32_s32_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x4.nxv4i32(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i32> @_Z28test_svreadz_ver_za32_s32_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x4.nxv4i32(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svint32x4_t test_svreadz_ver_za32_s32_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za32_s32_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x i32> @test_svreadz_ver_za32_u32_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x4.nxv4i32(i32 2, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x i32> @_Z28test_svreadz_ver_za32_u32_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.readz.vert.x4.nxv4i32(i32 2, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svuint32x4_t test_svreadz_ver_za32_u32_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za32_u32_vg4(2, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 16 x float> @test_svreadz_ver_za32_f32_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.vert.x4.nxv4f32(i32 3, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 16 x float> @_Z28test_svreadz_ver_za32_f32_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.readz.vert.x4.nxv4f32(i32 3, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+svfloat32x4_t test_svreadz_ver_za32_f32_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za32_f32_vg4(3, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i64> @test_svreadz_ver_za64_s64_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x4.nxv2i64(i32 0, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i64> @_Z28test_svreadz_ver_za64_s64_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x4.nxv2i64(i32 0, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svint64x4_t test_svreadz_ver_za64_s64_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za64_s64_vg4(0, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x i64> @test_svreadz_ver_za64_u64_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x4.nxv2i64(i32 4, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x i64> @_Z28test_svreadz_ver_za64_u64_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.readz.vert.x4.nxv2i64(i32 4, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svuint64x4_t test_svreadz_ver_za64_u64_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za64_u64_vg4(4, slice);
+}
+
+// CHECK-LABEL: define dso_local <vscale x 8 x double> @test_svreadz_ver_za64_f64_x4(
+// CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.vert.x4.nxv2f64(i32 7, i32 [[SLICE]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+// CPP-CHECK-LABEL: define dso_local <vscale x 8 x double> @_Z28test_svreadz_ver_za64_f64_x4j(
+// CPP-CHECK-SAME: i32 noundef [[SLICE:%.*]]) #[[ATTR0]] {
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.readz.vert.x4.nxv2f64(i32 7, i32 [[SLICE]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+svfloat64x4_t test_svreadz_ver_za64_f64_x4(uint32_t slice) __arm_streaming __arm_inout("za")
+{
+   return svreadz_ver_za64_f64_vg4(7, slice);
+}
diff --git a/llvm/include/llvm/IR/IntrinsicsAArch64.td b/llvm/include/llvm/IR/IntrinsicsAArch64.td
index bcaa37de74b63..572fd5ecb8dae 100644
--- a/llvm/include/llvm/IR/IntrinsicsAArch64.td
+++ b/llvm/include/llvm/IR/IntrinsicsAArch64.td
@@ -2839,6 +2839,24 @@ let TargetPrefix = "aarch64" in {
   def int_aarch64_sme_writeq_horiz : SME_VectorToTile_Intrinsic;
   def int_aarch64_sme_writeq_vert  : SME_VectorToTile_Intrinsic;
 
+  class SME_MOVAZ_TileToVector_X2_Intrinsic
+      : DefaultAttrsIntrinsic<[llvm_anyvector_ty, LLVMMatchType<0>],
+          [llvm_i32_ty, llvm_i32_ty],
+          [IntrNoMem, IntrHasSideEffects, ImmArg<ArgIndex<0>>]>;
+
+  class SME_MOVAZ_TileToVector_X4_Intrinsic
+      : DefaultAttrsIntrinsic<[llvm_anyvector_ty, LLVMMatchType<0>,
+           LLVMMatchType<0>,LLVMMatchType<0>],
+          [llvm_i32_ty, llvm_i32_ty],
+          [IntrNoMem, IntrHasSideEffects, ImmArg<ArgIndex<0>>]>;
+
+  def int_aarch64_sme_readz_horiz_x2 : SME_MOVAZ_TileToVector_X2_Intrinsic;
+  def int_aarch64_sme_readz_vert_x2  : SME_MOVAZ_TileToVector_X2_Intrinsic;
+
+  def int_aarch64_sme_readz_horiz_x4 : SME_MOVAZ_TileToVector_X4_Intrinsic;
+  def int_aarch64_sme_readz_vert_x4  : SME_MOVAZ_TileToVector_X4_Intrinsic;
+
+
   def int_aarch64_sme_zero : DefaultAttrsIntrinsic<[], [llvm_i32_ty], [ImmArg<ArgIndex<0>>]>;
 
   class SME_OuterProduct_Intrinsic
@@ -3646,4 +3664,4 @@ def int_aarch64_sve_pmov_to_pred_lane_zero : SVE2_1VectorArg_Pred_Intrinsic;
 
 def int_aarch64_sve_pmov_to_vector_lane_merging : SVE2_Pred_1VectorArgIndexed_Intrinsic;
    
-def int_aarch64_sve_pmov_to_vector_lane_zeroing : SVE2_Pred_1VectorArg_Intrinsic;
\ No newline at end of file
+def int_aarch64_sve_pmov_to_vector_lane_zeroing : SVE2_Pred_1VectorArg_Intrinsic;
diff --git a/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp b/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
index 51bec3604026b..07e6ac4e1b828 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
@@ -392,7 +392,8 @@ class AArch64DAGToDAGISel : public SelectionDAGISel {
   template <unsigned MaxIdx, unsigned Scale>
   void SelectMultiVectorMove(SDNode *N, unsigned NumVecs, unsigned BaseReg,
                              unsigned Op);
-
+  template <unsigned MaxIdx, unsigned Scale>
+  void SelectMultiVectorMoveZ(SDNode *N, unsigned NumVecs, unsigned Op);
   bool SelectAddrModeFrameIndexSVE(SDValue N, SDValue &Base, SDValue &OffImm);
   /// SVE Reg+Imm addressing mode.
   template <int64_t Min, int64_t Max>
@@ -1985,6 +1986,34 @@ void AArch64DAGToDAGISel::SelectMultiVectorMove(SDNode *N, unsigned NumVecs,
   CurDAG->RemoveDeadNode(N);
 }
 
+template <unsigned MaxIdx, unsigned Scale>
+void AArch64DAGToDAGISel::SelectMultiVectorMoveZ(SDNode *N, unsigned NumVecs,
+                                                 unsigned Op) {
+
+  SDValue SliceBase = N->getOperand(3);
+  SDValue Base, Offset;
+  if (!SelectSMETileSlice(SliceBase, MaxIdx, Base, Offset, Scale))
+    return;
+  // The correct Za tile number is computed in Machine Instruction
+  // See EmitTileMovaz
+  // DAG cannot select Za tile as an output register with ZReg
+  SDLoc DL(N);
+  SDValue Ops[] = {/*TileNum*/ N->getOperand(2), Base, Offset,
+                   /*Chain*/ N->getOperand(0)};
+  SDNode *Mov = CurDAG->getMachineNode(Op, DL, {MVT::Untyped, MVT::Other}, Ops);
+
+  EVT VT = N->getValueType(0);
+  for (unsigned I = 0; I < NumVecs; ++I)
+    ReplaceUses(SDValue(N, I),
+                CurDAG->getTargetExtractSubreg(AArch64::zsub0 + I, DL, VT,
+                                               SDValue(Mov, 0)));
+
+  // Copy chain
+  unsigned ChainIdx = NumVecs;
+  ReplaceUses(SDValue(N, ChainIdx), SDValue(Mov, 1));
+  CurDAG->RemoveDeadNode(N);
+}
+
 void AArch64DAGToDAGISel::SelectUnaryMultiIntrinsic(SDNode *N,
                                                     unsigned NumOutVecs,
                                                     bool IsTupleInput,
@@ -5175,6 +5204,74 @@ void AArch64DAGToDAGISel::Select(SDNode *Node) {
                                   AArch64::MOVA_VG4_4ZMXI);
       return;
     }
+    case Intrinsic::aarch64_sme_readz_horiz_x2: {
+      if (VT == MVT::nxv16i8) {
+        SelectMultiVectorMoveZ<14, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_B_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
+                 VT == MVT::nxv8bf16) {
+        SelectMultiVectorMoveZ<6, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_H_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
+        SelectMultiVectorMoveZ<2, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_S_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
+        SelectMultiVectorMoveZ<0, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_D_PSEUDO);
+        return;
+      }
+      break;
+    }
+    case Intrinsic::aarch64_sme_readz_vert_x2: {
+      if (VT == MVT::nxv16i8) {
+        SelectMultiVectorMoveZ<14, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_B_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
+                 VT == MVT::nxv8bf16) {
+        SelectMultiVectorMoveZ<6, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_H_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
+        SelectMultiVectorMoveZ<2, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_S_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
+        SelectMultiVectorMoveZ<0, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_D_PSEUDO);
+        return;
+      }
+      break;
+    }
+    case Intrinsic::aarch64_sme_readz_horiz_x4: {
+      if (VT == MVT::nxv16i8) {
+        SelectMultiVectorMoveZ<12, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_B_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
+                 VT == MVT::nxv8bf16) {
+        SelectMultiVectorMoveZ<4, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_H_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
+        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_S_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
+        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_D_PSEUDO);
+        return;
+      }
+      break;
+    }
+    case Intrinsic::aarch64_sme_readz_vert_x4: {
+      if (VT == MVT::nxv16i8) {
+        SelectMultiVectorMoveZ<12, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_B_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
+                 VT == MVT::nxv8bf16) {
+        SelectMultiVectorMoveZ<4, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_H_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
+        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_S_PSEUDO);
+        return;
+      } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
+        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_D_PSEUDO);
+        return;
+      }
+      break;
+    }
     case Intrinsic::swift_async_context_addr: {
       SDLoc DL(Node);
       SDValue Chain = Node->getOperand(0);
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index 80181a77c9d23..1b7a5f180c484 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -2832,6 +2832,23 @@ AArch64TargetLowering::EmitTileLoad(unsigned Opc, unsigned BaseReg,
   return BB;
 }
 
+MachineBasicBlock *
+AArch64TargetLowering::EmitTileMovaz(unsigned Opc, unsigned BaseReg,
+                                     MachineInstr &MI,
+                                     MachineBasicBlock *BB) const {
+  const TargetInstrInfo *TII = Subtarget->getInstrInfo();
+  MachineInstrBuilder MIB = BuildMI(*BB, MI, MI.getDebugLoc(), TII->get(Opc));
+
+  MIB.add(MI.getOperand(0)); // ZReg
+  MIB.addReg(BaseReg + MI.getOperand(1).getImm(),
+             RegState::Define);                    // add as output
+  MIB.addReg(BaseReg + MI.getOperand(1).getImm()); // add as input
+  MIB.add(MI.getOperand(2));                       // slice index register
+  MIB.add(MI.getOperand(3));                       // slice index offset
+  MI.eraseFromParent();                            // The pseudo is gone now.
+  return BB;
+}
+
 MachineBasicBlock *
 AArch64TargetLowering::EmitFill(MachineInstr &MI, MachineBasicBlock *BB) const {
   const TargetInstrInfo *TII = Subtarget->getInstrInfo();
@@ -2992,6 +3009,38 @@ MachineBasicBlock *AArch64TargetLowering::EmitInstrWithCustomInserter(
     return EmitZero(MI, BB);
   case AArch64::ZERO_T_PSEUDO:
     return EmitZTInstr(MI, BB, AArch64::ZERO_T, /*Op0IsDef=*/true);
+  case AArch64::MOVAZ_2ZMI_H_B_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_B, AArch64::ZAB0, MI, BB);
+  case AArch64::MOVAZ_2ZMI_H_H_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_H, AArch64::ZAH0, MI, BB);
+  case AArch64::MOVAZ_2ZMI_H_S_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_S, AArch64::ZAS0, MI, BB);
+  case AArch64::MOVAZ_2ZMI_H_D_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_D, AArch64::ZAD0, MI, BB);
+  case AArch64::MOVAZ_2ZMI_V_B_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_B, AArch64::ZAB0, MI, BB);
+  case AArch64::MOVAZ_2ZMI_V_H_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_H, AArch64::ZAH0, MI, BB);
+  case AArch64::MOVAZ_2ZMI_V_S_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_S, AArch64::ZAS0, MI, BB);
+  case AArch64::MOVAZ_2ZMI_V_D_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_D, AArch64::ZAD0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_H_B_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_B, AArch64::ZAB0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_H_H_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_H, AArch64::ZAH0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_H_S_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_S, AArch64::ZAS0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_H_D_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_D, AArch64::ZAD0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_V_B_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_B, AArch64::ZAB0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_V_H_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_H, AArch64::ZAH0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_V_S_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_S, AArch64::ZAS0, MI, BB);
+  case AArch64::MOVAZ_4ZMI_V_D_PSEUDO:
+    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_D, AArch64::ZAD0, MI, BB);
   }
 }
 
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.h b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
index 18439dc7f0102..8c86b47a701cd 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.h
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
@@ -635,6 +635,9 @@ class AArch64TargetLowering : public TargetLowering {
   MachineBasicBlock *EmitTileLoad(unsigned Opc, unsigned BaseReg,
                                   MachineInstr &MI,
                                   MachineBasicBlock *BB) const;
+  MachineBasicBlock *EmitTileMovaz(unsigned Opc, unsigned BaseReg,
+                                   MachineInstr &MI,
+                                   MachineBasicBlock *BB) const;
   MachineBasicBlock *EmitFill(MachineInstr &MI, MachineBasicBlock *BB) const;
   MachineBasicBlock *EmitZAInstr(unsigned Opc, unsigned BaseReg,
                                  MachineInstr &MI, MachineBasicBlock *BB,
diff --git a/llvm/lib/Target/AArch64/SMEInstrFormats.td b/llvm/lib/Target/AArch64/SMEInstrFormats.td
index 3363aab4b093c..d59df7e3f23bc 100644
--- a/llvm/lib/Target/AArch64/SMEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SMEInstrFormats.td
@@ -104,6 +104,12 @@ class sme2_move_to_tile_pseudo<string name, Operand tile_imm, Operand imm_ty, Re
   let usesCustomInserter = 1;
 }
 
+class sme2_movez_to_tile_multi_pseudo<string name, Operand tile_imm, Operand imm_ty, RegisterOperand vector_ty, SMEMatrixTypeEnum za_flag>
+    : SMEPseudo2Instr<name, 0>,
+      Pseudo<(outs vector_ty:$Zn), (ins tile_imm:$tile, MatrixIndexGPR32Op12_15:$Rs, imm_ty:$imm), []> {
+  let SMEMatrixType = za_flag;
+  let usesCustomInserter = 1;
+}
 //===----------------------------------------------------------------------===//
 // SME pattern match helpers.
 //===----------------------------------------------------------------------===//
@@ -4033,6 +4039,17 @@ multiclass sme2_mova_tile_to_vec_vg2_multi<string mnemonic>{
 multiclass sme2p1_movaz_tile_to_vec_vg2<string mnemonic>{
  defm _H : sme2_mova_tile_to_vec_vg2_multi_inst<0b0, 0b010, mnemonic>;
  defm _V : sme2_mova_tile_to_vec_vg2_multi_inst<0b1, 0b010, mnemonic>;
+
+
+ def NAME # _H_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm3s2range, ZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _H_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm2s2range, ZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _H_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm1s2range, ZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _H_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s2range, ZZ_d_mul_r, SMEMatrixTileD>;
+
+ def NAME # _V_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm3s2range, ZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _V_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm2s2range, ZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _V_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm1s2range, ZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _V_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s2range, ZZ_d_mul_r, SMEMatrixTileD>;
 }
 
 class sme2_mova_tile_to_vec_vg4_multi_base<bits<2> sz, bit v, bits<6> op,
@@ -4164,6 +4181,16 @@ multiclass sme2_mova_tile_to_vec_vg4_multi<string mnemonic>{
 multiclass sme2p1_movaz_tile_to_vec_vg4<string mnemonic>{
  defm _H : sme2_mova_tile_to_vec_vg4_multi_base<0b0, 0b110, mnemonic>;
  defm _V : sme2_mova_tile_to_vec_vg4_multi_base<0b1, 0b110, mnemonic>;
+
+ def NAME # _H_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm2s4range, ZZZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _H_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm1s4range, ZZZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _H_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm0s4range, ZZZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _H_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s4range, ZZZZ_d_mul_r, SMEMatrixTileD>;
+
+ def NAME # _V_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm2s4range, ZZZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _V_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm1s4range, ZZZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _V_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm0s4range, ZZZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _V_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s4range, ZZZZ_d_mul_r, SMEMatrixTileD>;
 }
 
 
diff --git a/llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.ll b/llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.ll
new file mode 100644
index 0000000000000..53207e543dd39
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.ll
@@ -0,0 +1,457 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
+; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sme2p1 -verify-machineinstrs < %s | FileCheck %s
+
+;MOVAZ (tile to vector, Multi)
+
+
+;;
+; X2 - Horiz
+;;
+
+define {<vscale x 16 x i8>, <vscale x 16 x i8>} @test_readz_hor_z8_i8_x2(i32 %tile, i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z8_i8_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w1
+; CHECK-NEXT:    movaz { z0.b, z1.b }, za0h.b[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.b, z1.b }, za0h.b[w12, 14:15]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.horiz.x2.nxv16i8(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 14
+  %res2 = call {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.horiz.x2.nxv16i8(i32 0, i32 %slice.max)
+  ret {<vscale x 16 x i8>, <vscale x 16 x i8>} %res2
+}
+define {<vscale x 8 x i16>, <vscale x 8 x i16>} @test_readz_hor_z16_i16_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z16_i16_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za0h.h[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za1h.h[w12, 6:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.horiz.x2.nxv8i16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 6
+  %res2 = call {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.horiz.x2.nxv8i16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x i16>, <vscale x 8 x i16>} %res2
+}
+
+define {<vscale x 4 x i32>, <vscale x 4 x i32>} @test_readz_hor_z32_i32_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z32_i32_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za0h.s[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za3h.s[w12, 2:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.horiz.x2.nxv4i32(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 2
+  %res2 = call {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.horiz.x2.nxv4i32(i32 3, i32 %slice.max)
+  ret {<vscale x 4 x i32>, <vscale x 4 x i32>} %res2
+}
+
+define {<vscale x 2 x i64>, <vscale x 2 x i64>} @test_readz_hor_z64_i64_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z64_i64_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d, z1.d }, za0h.d[w12, 0:1]
+; CHECK-NEXT:    movaz { z2.d, z3.d }, za7h.d[w12, 0:1]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.horiz.x2.nxv2i64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.horiz.x2.nxv2i64(i32 7, i32 %slice)
+  ret {<vscale x 2 x i64>, <vscale x 2 x i64>} %res
+}
+
+define {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @test_readz_hor_z16_bf16_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z16_bf16_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za0h.h[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za1h.h[w12, 6:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.horiz.x2.nxv8bf16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 6
+  %res2 = call {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.horiz.x2.nxv8bf16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} %res2
+}
+
+define {<vscale x 8 x half>, <vscale x 8 x half>} @test_readz_hor_z16_f16_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z16_f16_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za0h.h[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za1h.h[w12, 6:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.horiz.x2.nxv8f16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 6
+  %res2 = call {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.horiz.x2.nxv8f16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x half>, <vscale x 8 x half>} %res2
+}
+
+define {<vscale x 4 x float>, <vscale x 4 x float>} @test_readz_hor_z32_f32_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z32_f32_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za0h.s[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za3h.s[w12, 2:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.horiz.x2.nxv4f32(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 2
+  %res2 = call {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.horiz.x2.nxv4f32(i32 3, i32 %slice.max)
+  ret {<vscale x 4 x float>, <vscale x 4 x float>} %res2
+}
+
+define {<vscale x 2 x double>, <vscale x 2 x double>} @test_readz_hor_z64_f64_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z64_f64_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d, z1.d }, za0h.d[w12, 0:1]
+; CHECK-NEXT:    movaz { z2.d, z3.d }, za7h.d[w12, 0:1]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.horiz.x2.nxv2f64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.horiz.x2.nxv2f64(i32 7, i32 %slice)
+  ret {<vscale x 2 x double>, <vscale x 2 x double>} %res
+}
+
+;;
+; X2- Vert
+;;
+
+define {<vscale x 16 x i8>, <vscale x 16 x i8>} @test_readz_ver_z8_i8_x2(i32 %tile, i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z8_i8_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w1
+; CHECK-NEXT:    movaz { z0.b, z1.b }, za0v.b[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.b, z1.b }, za0v.b[w12, 14:15]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.vert.x2.nxv16i8(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 14
+  %res2 = call {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.vert.x2.nxv16i8(i32 0, i32 %slice.max)
+  ret {<vscale x 16 x i8>, <vscale x 16 x i8>} %res2
+}
+define {<vscale x 8 x i16>, <vscale x 8 x i16>} @test_readz_ver_z16_i16_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z16_i16_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za0v.h[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za1v.h[w12, 6:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.vert.x2.nxv8i16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 6
+  %res2 = call {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.vert.x2.nxv8i16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x i16>, <vscale x 8 x i16>} %res2
+}
+
+define {<vscale x 4 x i32>, <vscale x 4 x i32>} @test_readz_ver_z32_i32_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z32_i32_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za0v.s[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za3v.s[w12, 2:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.vert.x2.nxv4i32(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 2
+  %res2 = call {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.vert.x2.nxv4i32(i32 3, i32 %slice.max)
+  ret {<vscale x 4 x i32>, <vscale x 4 x i32>} %res2
+}
+
+define {<vscale x 2 x i64>, <vscale x 2 x i64>} @test_readz_ver_z64_i64_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z64_i64_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d, z1.d }, za0v.d[w12, 0:1]
+; CHECK-NEXT:    movaz { z2.d, z3.d }, za7v.d[w12, 0:1]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.vert.x2.nxv2i64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.vert.x2.nxv2i64(i32 7, i32 %slice)
+  ret {<vscale x 2 x i64>, <vscale x 2 x i64>} %res
+}
+
+define {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @test_readz_ver_z16_bf16_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z16_bf16_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za0v.h[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za1v.h[w12, 6:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.vert.x2.nxv8bf16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 6
+  %res2 = call {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.vert.x2.nxv8bf16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} %res2
+}
+
+define {<vscale x 8 x half>, <vscale x 8 x half>} @test_readz_ver_z16_f16_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z16_f16_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za0v.h[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.h, z1.h }, za1v.h[w12, 6:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.vert.x2.nxv8f16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 6
+  %res2 = call {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.vert.x2.nxv8f16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x half>, <vscale x 8 x half>} %res2
+}
+
+define {<vscale x 4 x float>, <vscale x 4 x float>} @test_readz_ver_z32_f32_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z32_f32_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za0v.s[w12, 0:1]
+; CHECK-NEXT:    movaz { z0.s, z1.s }, za3v.s[w12, 2:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.vert.x2.nxv4f32(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 2
+  %res2 = call {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.vert.x2.nxv4f32(i32 3, i32 %slice.max)
+  ret {<vscale x 4 x float>, <vscale x 4 x float>} %res2
+}
+
+define {<vscale x 2 x double>, <vscale x 2 x double>} @test_readz_ver_z64_f64_x2(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z64_f64_x2:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d, z1.d }, za0v.d[w12, 0:1]
+; CHECK-NEXT:    movaz { z2.d, z3.d }, za7v.d[w12, 0:1]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.vert.x2.nxv2f64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.vert.x2.nxv2f64(i32 7, i32 %slice)
+  ret {<vscale x 2 x double>, <vscale x 2 x double>} %res
+}
+
+;;
+; X4 - Horiz
+;;
+
+define {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @test_readz_hor_z8_i8_x4(i32 %tile, i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z8_i8_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w1
+; CHECK-NEXT:    movaz { z0.b - z3.b }, za0h.b[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.b - z3.b }, za0h.b[w12, 12:15]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.horiz.x4.nxv16i8(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 12
+  %res2 = call {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.horiz.x4.nxv16i8(i32 0, i32 %slice.max)
+  ret {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} %res2
+}
+define {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @test_readz_hor_z16_i16_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z16_i16_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za0h.h[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za1h.h[w12, 4:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.horiz.x4.nxv8i16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 4
+  %res2 = call {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.horiz.x4.nxv8i16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} %res2
+}
+
+define {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @test_readz_hor_z32_i32_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z32_i32_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za0h.s[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za3h.s[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.horiz.x4.nxv4i32(i32 0, i32 %slice)
+  %res2 = call {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.horiz.x4.nxv4i32(i32 3, i32 %slice)
+  ret {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} %res2
+}
+
+define {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @test_readz_hor_z64_i64_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z64_i64_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d - z3.d }, za0h.d[w12, 0:3]
+; CHECK-NEXT:    movaz { z4.d - z7.d }, za7h.d[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.horiz.x4.nxv2i64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.horiz.x4.nxv2i64(i32 7, i32 %slice)
+  ret {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} %res
+}
+
+define {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @test_readz_hor_z16_bf16_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z16_bf16_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za0h.h[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za1h.h[w12, 4:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.horiz.x4.nxv8bf16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 4
+  %res2 = call {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.horiz.x4.nxv8bf16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} %res2
+}
+
+define {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @test_readz_hor_z16_f16_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z16_f16_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za0h.h[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za1h.h[w12, 4:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.horiz.x4.nxv8f16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 4
+  %res2 = call {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.horiz.x4.nxv8f16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} %res2
+}
+
+define {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @test_readz_hor_z32_f32_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z32_f32_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za0h.s[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za3h.s[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.horiz.x4.nxv4f32(i32 0, i32 %slice)
+  %res2 = call {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.horiz.x4.nxv4f32(i32 3, i32 %slice)
+  ret {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} %res2
+}
+
+define {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @test_readz_hor_z64_f64_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_hor_z64_f64_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d - z3.d }, za0h.d[w12, 0:3]
+; CHECK-NEXT:    movaz { z4.d - z7.d }, za7h.d[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.horiz.x4.nxv2f64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.horiz.x4.nxv2f64(i32 7, i32 %slice)
+  ret {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} %res
+}
+
+;;
+; X4 - Vert
+;;
+
+define {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @test_readz_ver_z8_i8_x4(i32 %tile, i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z8_i8_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w1
+; CHECK-NEXT:    movaz { z0.b - z3.b }, za0v.b[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.b - z3.b }, za0v.b[w12, 12:15]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.vert.x4.nxv16i8(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 12
+  %res2 = call {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.vert.x4.nxv16i8(i32 0, i32 %slice.max)
+  ret {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} %res2
+}
+define {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @test_readz_ver_z16_i16_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z16_i16_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za0v.h[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za1v.h[w12, 4:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.vert.x4.nxv8i16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 4
+  %res2 = call {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.vert.x4.nxv8i16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} %res2
+}
+
+define {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @test_readz_ver_z32_i32_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z32_i32_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za0v.s[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za3v.s[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.vert.x4.nxv4i32(i32 0, i32 %slice)
+  %res2 = call {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.vert.x4.nxv4i32(i32 3, i32 %slice)
+  ret {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} %res2
+}
+
+define {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @test_readz_ver_z64_i64_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z64_i64_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d - z3.d }, za0v.d[w12, 0:3]
+; CHECK-NEXT:    movaz { z4.d - z7.d }, za7v.d[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.vert.x4.nxv2i64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.vert.x4.nxv2i64(i32 7, i32 %slice)
+  ret {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} %res
+}
+
+define {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @test_readz_ver_z16_bf16_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z16_bf16_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za0v.h[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za1v.h[w12, 4:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.vert.x4.nxv8bf16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 4
+  %res2 = call {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.vert.x4.nxv8bf16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} %res2
+}
+
+define {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @test_readz_ver_z16_f16_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z16_f16_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za0v.h[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.h - z3.h }, za1v.h[w12, 4:7]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.vert.x4.nxv8f16(i32 0, i32 %slice)
+  %slice.max = add i32 %slice, 4
+  %res2 = call {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.vert.x4.nxv8f16(i32 1, i32 %slice.max)
+  ret {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} %res2
+}
+
+define {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @test_readz_ver_z32_f32_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z32_f32_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za0v.s[w12, 0:3]
+; CHECK-NEXT:    movaz { z0.s - z3.s }, za3v.s[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.vert.x4.nxv4f32(i32 0, i32 %slice)
+  %res2 = call {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.vert.x4.nxv4f32(i32 3, i32 %slice)
+  ret {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} %res2
+}
+
+define {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @test_readz_ver_z64_f64_x4(i32 %slice) {
+; CHECK-LABEL: test_readz_ver_z64_f64_x4:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w12, w0
+; CHECK-NEXT:    movaz { z0.d - z3.d }, za0v.d[w12, 0:3]
+; CHECK-NEXT:    movaz { z4.d - z7.d }, za7v.d[w12, 0:3]
+; CHECK-NEXT:    ret
+  %res = call  {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.vert.x4.nxv2f64(i32 0, i32 %slice)
+  %res2 = call {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.vert.x4.nxv2f64(i32 7, i32 %slice)
+  ret {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} %res
+}
+
+declare {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.horiz.za8.x2.nxv16i8(i32, i32)
+declare {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.horiz.x2.nxv8i16(i32, i32)
+declare {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.horiz.x2.nxv4i32(i32, i32)
+declare {<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.horiz.x2.nxv2i64(i32, i32)
+declare {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.horiz.x2.nxv8bf16(i32, i32)
+declare {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.horiz.x2.nxv8f16(i32, i32)
+declare {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.horiz.x2.nxv4f32(i32, i32)
+declare {<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.horiz.x2.nxv2f64(i32, i32)
+
+declare {<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.vert.za8.x2.nxv16i8(i32, i32)
+declare {<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.vert.x2.nxv8i16(i32, i32)
+declare {<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.vert.x2.nxv4i32(i32, i32)
+declare {<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.vert.x2.nxv2i64(i32, i32)
+declare {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.vert.x2.nxv8bf16(i32, i32)
+declare {<vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.vert.x2.nxv8f16(i32, i32)
+declare {<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.vert.x2.nxv4f32(i32, i32)
+declare {<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.vert.x2.nxv2f64(i32, i32)
+
+declare {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.horiz.za8.x4.nxv16i8(i32, i32)
+declare {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.horiz.x4.nxv8i16(i32, i32)
+declare {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.horiz.x4.nxv4i32(i32, i32)
+declare {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.horiz.x4.nxv2i64(i32, i32)
+declare {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.horiz.x4.nxv8bf16(i32, i32)
+declare {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.horiz.x4.nxv8f16(i32, i32)
+declare {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.horiz.x4.nxv4f32(i32, i32)
+declare {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.horiz.x4.nxv2f64(i32, i32)
+
+declare {<vscale x 16 x i8>, <vscale x 16 x i8>,<vscale x 16 x i8>, <vscale x 16 x i8>} @llvm.aarch64.sme.readz.vert.za8.x4.nxv16i8(i32, i32)
+declare {<vscale x 8 x i16>, <vscale x 8 x i16>,<vscale x 8 x i16>, <vscale x 8 x i16>} @llvm.aarch64.sme.readz.vert.x4.nxv8i16(i32, i32)
+declare {<vscale x 4 x i32>, <vscale x 4 x i32>,<vscale x 4 x i32>, <vscale x 4 x i32>} @llvm.aarch64.sme.readz.vert.x4.nxv4i32(i32, i32)
+declare {<vscale x 2 x i64>, <vscale x 2 x i64>,<vscale x 2 x i64>, <vscale x 2 x i64>} @llvm.aarch64.sme.readz.vert.x4.nxv2i64(i32, i32)
+declare {<vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>} @llvm.aarch64.sme.readz.vert.x4.nxv8bf16(i32, i32)
+declare {<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>} @llvm.aarch64.sme.readz.vert.x4.nxv8f16(i32, i32)
+declare {<vscale x 4 x float>, <vscale x 4 x float>,<vscale x 4 x float>, <vscale x 4 x float>} @llvm.aarch64.sme.readz.vert.x4.nxv4f32(i32, i32)
+declare {<vscale x 2 x double>, <vscale x 2 x double>,<vscale x 2 x double>, <vscale x 2 x double>} @llvm.aarch64.sme.readz.vert.x4.nxv2f64(i32, i32)
diff --git a/llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.s b/llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.s
new file mode 100644
index 0000000000000..e69de29bb2d1d

>From bf03a05f3c1ba402d3991f68307106eb3494cf60 Mon Sep 17 00:00:00 2001
From: Caroline Concatto <caroline.concatto at arm.com>
Date: Mon, 15 Apr 2024 10:42:36 +0000
Subject: [PATCH 2/5] Remove unwanted test file

---
 llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.s | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 delete mode 100644 llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.s

diff --git a/llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.s b/llvm/test/CodeGen/AArch64/sme2p1-intrinsics-movaz.s
deleted file mode 100644
index e69de29bb2d1d..0000000000000

>From b74fd7a18c808bc98f31710aea17b8b4d8cae6aa Mon Sep 17 00:00:00 2001
From: Caroline Concatto <caroline.concatto at arm.com>
Date: Mon, 29 Apr 2024 14:37:35 +0000
Subject: [PATCH 3/5] Reuse-EmitZAInstr-to-add-Za-Matrix

---
 .../Target/AArch64/AArch64ISelDAGToDAG.cpp    |  40 +++----
 .../Target/AArch64/AArch64ISelLowering.cpp    | 113 ++++++++++--------
 llvm/lib/Target/AArch64/AArch64ISelLowering.h |   2 +-
 llvm/lib/Target/AArch64/SMEInstrFormats.td    |  48 ++++----
 4 files changed, 111 insertions(+), 92 deletions(-)

diff --git a/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp b/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
index 07e6ac4e1b828..8eca0f704dbcd 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelDAGToDAG.cpp
@@ -392,8 +392,8 @@ class AArch64DAGToDAGISel : public SelectionDAGISel {
   template <unsigned MaxIdx, unsigned Scale>
   void SelectMultiVectorMove(SDNode *N, unsigned NumVecs, unsigned BaseReg,
                              unsigned Op);
-  template <unsigned MaxIdx, unsigned Scale>
-  void SelectMultiVectorMoveZ(SDNode *N, unsigned NumVecs, unsigned Op);
+  void SelectMultiVectorMoveZ(SDNode *N, unsigned NumVecs, unsigned Op,
+                              unsigned MaxIdx, unsigned Scale);
   bool SelectAddrModeFrameIndexSVE(SDValue N, SDValue &Base, SDValue &OffImm);
   /// SVE Reg+Imm addressing mode.
   template <int64_t Min, int64_t Max>
@@ -1986,9 +1986,9 @@ void AArch64DAGToDAGISel::SelectMultiVectorMove(SDNode *N, unsigned NumVecs,
   CurDAG->RemoveDeadNode(N);
 }
 
-template <unsigned MaxIdx, unsigned Scale>
 void AArch64DAGToDAGISel::SelectMultiVectorMoveZ(SDNode *N, unsigned NumVecs,
-                                                 unsigned Op) {
+                                                 unsigned Op, unsigned MaxIdx,
+                                                 unsigned Scale) {
 
   SDValue SliceBase = N->getOperand(3);
   SDValue Base, Offset;
@@ -5206,68 +5206,68 @@ void AArch64DAGToDAGISel::Select(SDNode *Node) {
     }
     case Intrinsic::aarch64_sme_readz_horiz_x2: {
       if (VT == MVT::nxv16i8) {
-        SelectMultiVectorMoveZ<14, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_B_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_H_B_PSEUDO, 14, 2);
         return;
       } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
                  VT == MVT::nxv8bf16) {
-        SelectMultiVectorMoveZ<6, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_H_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_H_H_PSEUDO, 6, 2);
         return;
       } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
-        SelectMultiVectorMoveZ<2, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_S_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_H_S_PSEUDO, 2, 2);
         return;
       } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
-        SelectMultiVectorMoveZ<0, 2>(Node, 2, AArch64::MOVAZ_2ZMI_H_D_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_H_D_PSEUDO, 0, 2);
         return;
       }
       break;
     }
     case Intrinsic::aarch64_sme_readz_vert_x2: {
       if (VT == MVT::nxv16i8) {
-        SelectMultiVectorMoveZ<14, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_B_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_V_B_PSEUDO, 14, 2);
         return;
       } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
                  VT == MVT::nxv8bf16) {
-        SelectMultiVectorMoveZ<6, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_H_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_V_H_PSEUDO, 6, 2);
         return;
       } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
-        SelectMultiVectorMoveZ<2, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_S_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_V_S_PSEUDO, 2, 2);
         return;
       } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
-        SelectMultiVectorMoveZ<0, 2>(Node, 2, AArch64::MOVAZ_2ZMI_V_D_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 2, AArch64::MOVAZ_2ZMI_V_D_PSEUDO, 0, 2);
         return;
       }
       break;
     }
     case Intrinsic::aarch64_sme_readz_horiz_x4: {
       if (VT == MVT::nxv16i8) {
-        SelectMultiVectorMoveZ<12, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_B_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_H_B_PSEUDO, 12, 4);
         return;
       } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
                  VT == MVT::nxv8bf16) {
-        SelectMultiVectorMoveZ<4, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_H_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_H_H_PSEUDO, 4, 4);
         return;
       } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
-        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_S_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_H_S_PSEUDO, 0, 4);
         return;
       } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
-        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_H_D_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_H_D_PSEUDO, 0, 4);
         return;
       }
       break;
     }
     case Intrinsic::aarch64_sme_readz_vert_x4: {
       if (VT == MVT::nxv16i8) {
-        SelectMultiVectorMoveZ<12, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_B_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_V_B_PSEUDO, 12, 4);
         return;
       } else if (VT == MVT::nxv8i16 || VT == MVT::nxv8f16 ||
                  VT == MVT::nxv8bf16) {
-        SelectMultiVectorMoveZ<4, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_H_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_V_H_PSEUDO, 4, 4);
         return;
       } else if (VT == MVT::nxv4i32 || VT == MVT::nxv4f32) {
-        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_S_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_V_S_PSEUDO, 0, 4);
         return;
       } else if (VT == MVT::nxv2i64 || VT == MVT::nxv2f64) {
-        SelectMultiVectorMoveZ<0, 4>(Node, 4, AArch64::MOVAZ_4ZMI_V_D_PSEUDO);
+        SelectMultiVectorMoveZ(Node, 4, AArch64::MOVAZ_4ZMI_V_D_PSEUDO, 0, 4);
         return;
       }
       break;
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index 1b7a5f180c484..c0d22b7d06966 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -2883,19 +2883,28 @@ MachineBasicBlock *AArch64TargetLowering::EmitZTInstr(MachineInstr &MI,
 
 MachineBasicBlock *
 AArch64TargetLowering::EmitZAInstr(unsigned Opc, unsigned BaseReg,
-                                   MachineInstr &MI,
-                                   MachineBasicBlock *BB, bool HasTile) const {
+                                   MachineInstr &MI, MachineBasicBlock *BB,
+                                   bool HasTile, bool HasZPROut) const {
   const TargetInstrInfo *TII = Subtarget->getInstrInfo();
   MachineInstrBuilder MIB = BuildMI(*BB, MI, MI.getDebugLoc(), TII->get(Opc));
   unsigned StartIdx = 0;
 
-  if (HasTile) {
-    MIB.addReg(BaseReg + MI.getOperand(0).getImm(), RegState::Define);
-    MIB.addReg(BaseReg + MI.getOperand(0).getImm());
-    StartIdx = 1;
-  } else
-    MIB.addReg(BaseReg, RegState::Define).addReg(BaseReg);
-
+  if (HasZPROut) {
+    if (HasTile) {
+      MIB.add(MI.getOperand(0)); // Output ZPR
+      MIB.addReg(BaseReg + MI.getOperand(1).getImm(),
+                 RegState::Define);                    // Output ZA Tile
+      MIB.addReg(BaseReg + MI.getOperand(1).getImm()); // Input Za Tile
+      StartIdx = 2;
+    }
+  } else {
+    if (HasTile) {
+      MIB.addReg(BaseReg + MI.getOperand(0).getImm(), RegState::Define);
+      MIB.addReg(BaseReg + MI.getOperand(0).getImm());
+      StartIdx = 1;
+    } else
+      MIB.addReg(BaseReg, RegState::Define).addReg(BaseReg);
+  }
   for (unsigned I = StartIdx; I < MI.getNumOperands(); ++I)
     MIB.add(MI.getOperand(I));
 
@@ -2930,17 +2939,59 @@ MachineBasicBlock *AArch64TargetLowering::EmitInstrWithCustomInserter(
         TII->get(MI.getOpcode()).TSFlags & AArch64::SMEMatrixTypeMask;
     switch (SMEMatrixType) {
     case (AArch64::SMEMatrixArray):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZA, MI, BB, /*HasTile*/ false);
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZA, MI, BB, /*HasTile*/ false,
+                         /*HasZPROut*/ false);
     case (AArch64::SMEMatrixTileB):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZAB0, MI, BB, /*HasTile*/ true);
+      switch (MI.getOpcode()) {
+      case AArch64::MOVAZ_2ZMI_H_B_PSEUDO:
+      case AArch64::MOVAZ_2ZMI_V_B_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_H_B_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_V_B_PSEUDO:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAB0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ true);
+      default:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAB0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ false);
+      }
     case (AArch64::SMEMatrixTileH):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZAH0, MI, BB, /*HasTile*/ true);
+      switch (MI.getOpcode()) {
+      case AArch64::MOVAZ_2ZMI_H_H_PSEUDO:
+      case AArch64::MOVAZ_2ZMI_V_H_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_H_H_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_V_H_PSEUDO:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAH0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ true);
+      default:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAH0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ false);
+      }
     case (AArch64::SMEMatrixTileS):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZAS0, MI, BB, /*HasTile*/ true);
+      switch (MI.getOpcode()) {
+      case AArch64::MOVAZ_2ZMI_H_S_PSEUDO:
+      case AArch64::MOVAZ_2ZMI_V_S_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_H_S_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_V_S_PSEUDO:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAS0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ true);
+      default:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAS0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ false);
+      }
     case (AArch64::SMEMatrixTileD):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZAD0, MI, BB, /*HasTile*/ true);
+      switch (MI.getOpcode()) {
+      case AArch64::MOVAZ_2ZMI_H_D_PSEUDO:
+      case AArch64::MOVAZ_2ZMI_V_D_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_H_D_PSEUDO:
+      case AArch64::MOVAZ_4ZMI_V_D_PSEUDO:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAD0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ true);
+      default:
+        return EmitZAInstr(SMEOrigInstr, AArch64::ZAD0, MI, BB,
+                           /*HasTile*/ true, /*HasZPROut*/ false);
+      }
     case (AArch64::SMEMatrixTileQ):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZAQ0, MI, BB, /*HasTile*/ true);
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZAQ0, MI, BB, /*HasTile*/ true,
+                         /*HasZPROut*/ false);
     }
   }
 
@@ -3009,38 +3060,6 @@ MachineBasicBlock *AArch64TargetLowering::EmitInstrWithCustomInserter(
     return EmitZero(MI, BB);
   case AArch64::ZERO_T_PSEUDO:
     return EmitZTInstr(MI, BB, AArch64::ZERO_T, /*Op0IsDef=*/true);
-  case AArch64::MOVAZ_2ZMI_H_B_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_B, AArch64::ZAB0, MI, BB);
-  case AArch64::MOVAZ_2ZMI_H_H_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_H, AArch64::ZAH0, MI, BB);
-  case AArch64::MOVAZ_2ZMI_H_S_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_S, AArch64::ZAS0, MI, BB);
-  case AArch64::MOVAZ_2ZMI_H_D_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_H_D, AArch64::ZAD0, MI, BB);
-  case AArch64::MOVAZ_2ZMI_V_B_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_B, AArch64::ZAB0, MI, BB);
-  case AArch64::MOVAZ_2ZMI_V_H_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_H, AArch64::ZAH0, MI, BB);
-  case AArch64::MOVAZ_2ZMI_V_S_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_S, AArch64::ZAS0, MI, BB);
-  case AArch64::MOVAZ_2ZMI_V_D_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_2ZMI_V_D, AArch64::ZAD0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_H_B_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_B, AArch64::ZAB0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_H_H_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_H, AArch64::ZAH0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_H_S_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_S, AArch64::ZAS0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_H_D_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_H_D, AArch64::ZAD0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_V_B_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_B, AArch64::ZAB0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_V_H_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_H, AArch64::ZAH0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_V_S_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_S, AArch64::ZAS0, MI, BB);
-  case AArch64::MOVAZ_4ZMI_V_D_PSEUDO:
-    return EmitTileMovaz(AArch64::MOVAZ_4ZMI_V_D, AArch64::ZAD0, MI, BB);
   }
 }
 
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.h b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
index 8c86b47a701cd..da1ab2a10add4 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.h
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
@@ -641,7 +641,7 @@ class AArch64TargetLowering : public TargetLowering {
   MachineBasicBlock *EmitFill(MachineInstr &MI, MachineBasicBlock *BB) const;
   MachineBasicBlock *EmitZAInstr(unsigned Opc, unsigned BaseReg,
                                  MachineInstr &MI, MachineBasicBlock *BB,
-                                 bool HasTile) const;
+                                 bool HasTile, bool HasZPROut) const;
   MachineBasicBlock *EmitZTInstr(MachineInstr &MI, MachineBasicBlock *BB,
                                  unsigned Opcode, bool Op0IsDef) const;
   MachineBasicBlock *EmitZero(MachineInstr &MI, MachineBasicBlock *BB) const;
diff --git a/llvm/lib/Target/AArch64/SMEInstrFormats.td b/llvm/lib/Target/AArch64/SMEInstrFormats.td
index d59df7e3f23bc..9ece7e36a0521 100644
--- a/llvm/lib/Target/AArch64/SMEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SMEInstrFormats.td
@@ -3942,7 +3942,7 @@ multiclass sme2_mova_tile_to_vec_vg2_multi_inst<bit v, bits<3> opc, string mnemo
   def _B : sme2_mova_tile_to_vec_vg2_multi_base<0b00, v, opc, ZZ_b_mul_r,
                                                 !if(v, TileVectorOpV8,
                                                        TileVectorOpH8),
-                                                 uimm3s2range, mnemonic> {
+                                                 uimm3s2range, mnemonic>, SMEPseudo2Instr<NAME # _B, 1> {
     bits<3> imm;
     let Inst{7-5} = imm;
   }
@@ -3950,7 +3950,7 @@ multiclass sme2_mova_tile_to_vec_vg2_multi_inst<bit v, bits<3> opc, string mnemo
   def _H : sme2_mova_tile_to_vec_vg2_multi_base<0b01, v, opc, ZZ_h_mul_r,
                                                 !if(v, TileVectorOpV16,
                                                        TileVectorOpH16),
-                                                 uimm2s2range, mnemonic> {
+                                                 uimm2s2range, mnemonic>, SMEPseudo2Instr<NAME # _H, 1> {
     bits<1> ZAn;
     bits<2> imm;
     let Inst{7}   = ZAn;
@@ -3960,7 +3960,7 @@ multiclass sme2_mova_tile_to_vec_vg2_multi_inst<bit v, bits<3> opc, string mnemo
   def _S : sme2_mova_tile_to_vec_vg2_multi_base<0b10, v, opc, ZZ_s_mul_r,
                                                 !if(v, TileVectorOpV32,
                                                        TileVectorOpH32),
-                                                 uimm1s2range, mnemonic> {
+                                                 uimm1s2range, mnemonic>, SMEPseudo2Instr<NAME # _S, 1> {
     bits<2> ZAn;
     bits<1> imm;
     let Inst{7-6} = ZAn;
@@ -3970,7 +3970,7 @@ multiclass sme2_mova_tile_to_vec_vg2_multi_inst<bit v, bits<3> opc, string mnemo
   def _D : sme2_mova_tile_to_vec_vg2_multi_base<0b11, v, opc, ZZ_d_mul_r,
                                                 !if(v, TileVectorOpV64,
                                                        TileVectorOpH64),
-                                                uimm0s2range, mnemonic> {
+                                                uimm0s2range, mnemonic>, SMEPseudo2Instr<NAME # _D, 1> {
     bits<3> ZAn;
     let Inst{7-5} = ZAn;
   }
@@ -4041,15 +4041,15 @@ multiclass sme2p1_movaz_tile_to_vec_vg2<string mnemonic>{
  defm _V : sme2_mova_tile_to_vec_vg2_multi_inst<0b1, 0b010, mnemonic>;
 
 
- def NAME # _H_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm3s2range, ZZ_b_mul_r, SMEMatrixTileB>;
- def NAME # _H_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm2s2range, ZZ_h_mul_r, SMEMatrixTileH>;
- def NAME # _H_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm1s2range, ZZ_s_mul_r, SMEMatrixTileS>;
- def NAME # _H_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s2range, ZZ_d_mul_r, SMEMatrixTileD>;
+ def NAME # _H_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_B, sme_elm_idx0_0, uimm3s2range, ZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _H_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_H, sme_elm_idx0_1, uimm2s2range, ZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _H_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_S, sme_elm_idx0_3, uimm1s2range, ZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _H_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_D, sme_elm_idx0_7, uimm0s2range, ZZ_d_mul_r, SMEMatrixTileD>;
 
- def NAME # _V_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm3s2range, ZZ_b_mul_r, SMEMatrixTileB>;
- def NAME # _V_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm2s2range, ZZ_h_mul_r, SMEMatrixTileH>;
- def NAME # _V_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm1s2range, ZZ_s_mul_r, SMEMatrixTileS>;
- def NAME # _V_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s2range, ZZ_d_mul_r, SMEMatrixTileD>;
+ def NAME # _V_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_B, sme_elm_idx0_0, uimm3s2range, ZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _V_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_H, sme_elm_idx0_1, uimm2s2range, ZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _V_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_S, sme_elm_idx0_3, uimm1s2range, ZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _V_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_D, sme_elm_idx0_7, uimm0s2range, ZZ_d_mul_r, SMEMatrixTileD>;
 }
 
 class sme2_mova_tile_to_vec_vg4_multi_base<bits<2> sz, bit v, bits<6> op,
@@ -4083,7 +4083,7 @@ multiclass sme2_mova_tile_to_vec_vg4_multi_base<bit v, bits<3> opc, string mnemo
                                                 ZZZZ_b_mul_r,
                                                 !if(v, TileVectorOpV8,
                                                        TileVectorOpH8),
-                                                uimm2s4range, mnemonic> {
+                                                uimm2s4range, mnemonic>, SMEPseudo2Instr<NAME # _B, 1> {
     bits<2> imm;
     let Inst{6-5} = imm;
   }
@@ -4092,7 +4092,7 @@ multiclass sme2_mova_tile_to_vec_vg4_multi_base<bit v, bits<3> opc, string mnemo
                                                 ZZZZ_h_mul_r,
                                                 !if(v, TileVectorOpV16,
                                                        TileVectorOpH16),
-                                                uimm1s4range, mnemonic> {
+                                                uimm1s4range, mnemonic>, SMEPseudo2Instr<NAME # _H, 1> {
     bits<1> ZAn;
     bits<1> imm;
     let Inst{6}   = ZAn;
@@ -4103,7 +4103,7 @@ multiclass sme2_mova_tile_to_vec_vg4_multi_base<bit v, bits<3> opc, string mnemo
                                                 ZZZZ_s_mul_r,
                                                 !if(v, TileVectorOpV32,
                                                        TileVectorOpH32),
-                                                 uimm0s4range, mnemonic> {
+                                                 uimm0s4range, mnemonic>, SMEPseudo2Instr<NAME # _S, 1> {
     bits<2> ZAn;
     let Inst{6-5} = ZAn;
   }
@@ -4112,7 +4112,7 @@ multiclass sme2_mova_tile_to_vec_vg4_multi_base<bit v, bits<3> opc, string mnemo
                                                 ZZZZ_d_mul_r,
                                                 !if(v, TileVectorOpV64,
                                                        TileVectorOpH64),
-                                                uimm0s4range, mnemonic> {
+                                                uimm0s4range, mnemonic>, SMEPseudo2Instr<NAME # _D, 1> {
     bits<3> ZAn;
     let Inst{7-5} = ZAn;
   }
@@ -4182,15 +4182,15 @@ multiclass sme2p1_movaz_tile_to_vec_vg4<string mnemonic>{
  defm _H : sme2_mova_tile_to_vec_vg4_multi_base<0b0, 0b110, mnemonic>;
  defm _V : sme2_mova_tile_to_vec_vg4_multi_base<0b1, 0b110, mnemonic>;
 
- def NAME # _H_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm2s4range, ZZZZ_b_mul_r, SMEMatrixTileB>;
- def NAME # _H_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm1s4range, ZZZZ_h_mul_r, SMEMatrixTileH>;
- def NAME # _H_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm0s4range, ZZZZ_s_mul_r, SMEMatrixTileS>;
- def NAME # _H_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s4range, ZZZZ_d_mul_r, SMEMatrixTileD>;
+ def NAME # _H_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_B, sme_elm_idx0_0, uimm2s4range, ZZZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _H_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_H, sme_elm_idx0_1, uimm1s4range, ZZZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _H_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_S, sme_elm_idx0_3, uimm0s4range, ZZZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _H_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _H_D, sme_elm_idx0_7, uimm0s4range, ZZZZ_d_mul_r, SMEMatrixTileD>;
 
- def NAME # _V_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_0, uimm2s4range, ZZZZ_b_mul_r, SMEMatrixTileB>;
- def NAME # _V_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_1, uimm1s4range, ZZZZ_h_mul_r, SMEMatrixTileH>;
- def NAME # _V_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_3, uimm0s4range, ZZZZ_s_mul_r, SMEMatrixTileS>;
- def NAME # _V_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME, sme_elm_idx0_7, uimm0s4range, ZZZZ_d_mul_r, SMEMatrixTileD>;
+ def NAME # _V_B_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_B, sme_elm_idx0_0, uimm2s4range, ZZZZ_b_mul_r, SMEMatrixTileB>;
+ def NAME # _V_H_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_H, sme_elm_idx0_1, uimm1s4range, ZZZZ_h_mul_r, SMEMatrixTileH>;
+ def NAME # _V_S_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_S, sme_elm_idx0_3, uimm0s4range, ZZZZ_s_mul_r, SMEMatrixTileS>;
+ def NAME # _V_D_PSEUDO : sme2_movez_to_tile_multi_pseudo<NAME # _V_D, sme_elm_idx0_7, uimm0s4range, ZZZZ_d_mul_r, SMEMatrixTileD>;
 }
 
 

>From 6331343214b99a2c2642c5f65c3f7adb920981c8 Mon Sep 17 00:00:00 2001
From: Caroline Concatto <caroline.concatto at arm.com>
Date: Fri, 17 May 2024 10:24:29 +0000
Subject: [PATCH 4/5] Simplify EmitZaInstr

---
 .../Target/AArch64/AArch64ISelLowering.cpp    | 90 ++++---------------
 llvm/lib/Target/AArch64/AArch64ISelLowering.h |  6 +-
 2 files changed, 17 insertions(+), 79 deletions(-)

diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index c0d22b7d06966..86b280e67bc7a 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -2832,23 +2832,6 @@ AArch64TargetLowering::EmitTileLoad(unsigned Opc, unsigned BaseReg,
   return BB;
 }
 
-MachineBasicBlock *
-AArch64TargetLowering::EmitTileMovaz(unsigned Opc, unsigned BaseReg,
-                                     MachineInstr &MI,
-                                     MachineBasicBlock *BB) const {
-  const TargetInstrInfo *TII = Subtarget->getInstrInfo();
-  MachineInstrBuilder MIB = BuildMI(*BB, MI, MI.getDebugLoc(), TII->get(Opc));
-
-  MIB.add(MI.getOperand(0)); // ZReg
-  MIB.addReg(BaseReg + MI.getOperand(1).getImm(),
-             RegState::Define);                    // add as output
-  MIB.addReg(BaseReg + MI.getOperand(1).getImm()); // add as input
-  MIB.add(MI.getOperand(2));                       // slice index register
-  MIB.add(MI.getOperand(3));                       // slice index offset
-  MI.eraseFromParent();                            // The pseudo is gone now.
-  return BB;
-}
-
 MachineBasicBlock *
 AArch64TargetLowering::EmitFill(MachineInstr &MI, MachineBasicBlock *BB) const {
   const TargetInstrInfo *TII = Subtarget->getInstrInfo();
@@ -2883,20 +2866,20 @@ MachineBasicBlock *AArch64TargetLowering::EmitZTInstr(MachineInstr &MI,
 
 MachineBasicBlock *
 AArch64TargetLowering::EmitZAInstr(unsigned Opc, unsigned BaseReg,
-                                   MachineInstr &MI, MachineBasicBlock *BB,
-                                   bool HasTile, bool HasZPROut) const {
+                                   MachineInstr &MI,
+                                   MachineBasicBlock *BB) const {
   const TargetInstrInfo *TII = Subtarget->getInstrInfo();
   MachineInstrBuilder MIB = BuildMI(*BB, MI, MI.getDebugLoc(), TII->get(Opc));
   unsigned StartIdx = 0;
 
+  bool HasTile = BaseReg != AArch64::ZA;
+  bool HasZPROut = HasTile && MI.getOperand(0).isReg();
   if (HasZPROut) {
-    if (HasTile) {
-      MIB.add(MI.getOperand(0)); // Output ZPR
-      MIB.addReg(BaseReg + MI.getOperand(1).getImm(),
-                 RegState::Define);                    // Output ZA Tile
-      MIB.addReg(BaseReg + MI.getOperand(1).getImm()); // Input Za Tile
-      StartIdx = 2;
-    }
+    MIB.add(MI.getOperand(0)); // Output ZPR
+    MIB.addReg(BaseReg + MI.getOperand(1).getImm(),
+               RegState::Define);                    // Output ZA Tile
+    MIB.addReg(BaseReg + MI.getOperand(1).getImm()); // Input Za Tile
+    StartIdx = 2;
   } else {
     if (HasTile) {
       MIB.addReg(BaseReg + MI.getOperand(0).getImm(), RegState::Define);
@@ -2939,59 +2922,18 @@ MachineBasicBlock *AArch64TargetLowering::EmitInstrWithCustomInserter(
         TII->get(MI.getOpcode()).TSFlags & AArch64::SMEMatrixTypeMask;
     switch (SMEMatrixType) {
     case (AArch64::SMEMatrixArray):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZA, MI, BB, /*HasTile*/ false,
-                         /*HasZPROut*/ false);
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZA, MI, BB);
     case (AArch64::SMEMatrixTileB):
-      switch (MI.getOpcode()) {
-      case AArch64::MOVAZ_2ZMI_H_B_PSEUDO:
-      case AArch64::MOVAZ_2ZMI_V_B_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_H_B_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_V_B_PSEUDO:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAB0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ true);
-      default:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAB0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ false);
-      }
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZAB0, MI, BB);
     case (AArch64::SMEMatrixTileH):
-      switch (MI.getOpcode()) {
-      case AArch64::MOVAZ_2ZMI_H_H_PSEUDO:
-      case AArch64::MOVAZ_2ZMI_V_H_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_H_H_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_V_H_PSEUDO:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAH0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ true);
-      default:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAH0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ false);
-      }
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZAH0, MI, BB);
+      ///*HasTile*/ true, /*HasZPROut*/ false);
     case (AArch64::SMEMatrixTileS):
-      switch (MI.getOpcode()) {
-      case AArch64::MOVAZ_2ZMI_H_S_PSEUDO:
-      case AArch64::MOVAZ_2ZMI_V_S_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_H_S_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_V_S_PSEUDO:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAS0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ true);
-      default:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAS0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ false);
-      }
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZAS0, MI, BB);
     case (AArch64::SMEMatrixTileD):
-      switch (MI.getOpcode()) {
-      case AArch64::MOVAZ_2ZMI_H_D_PSEUDO:
-      case AArch64::MOVAZ_2ZMI_V_D_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_H_D_PSEUDO:
-      case AArch64::MOVAZ_4ZMI_V_D_PSEUDO:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAD0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ true);
-      default:
-        return EmitZAInstr(SMEOrigInstr, AArch64::ZAD0, MI, BB,
-                           /*HasTile*/ true, /*HasZPROut*/ false);
-      }
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZAD0, MI, BB);
     case (AArch64::SMEMatrixTileQ):
-      return EmitZAInstr(SMEOrigInstr, AArch64::ZAQ0, MI, BB, /*HasTile*/ true,
-                         /*HasZPROut*/ false);
+      return EmitZAInstr(SMEOrigInstr, AArch64::ZAQ0, MI, BB);
     }
   }
 
diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.h b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
index da1ab2a10add4..8aa95d4bfe6bd 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.h
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.h
@@ -635,13 +635,9 @@ class AArch64TargetLowering : public TargetLowering {
   MachineBasicBlock *EmitTileLoad(unsigned Opc, unsigned BaseReg,
                                   MachineInstr &MI,
                                   MachineBasicBlock *BB) const;
-  MachineBasicBlock *EmitTileMovaz(unsigned Opc, unsigned BaseReg,
-                                   MachineInstr &MI,
-                                   MachineBasicBlock *BB) const;
   MachineBasicBlock *EmitFill(MachineInstr &MI, MachineBasicBlock *BB) const;
   MachineBasicBlock *EmitZAInstr(unsigned Opc, unsigned BaseReg,
-                                 MachineInstr &MI, MachineBasicBlock *BB,
-                                 bool HasTile, bool HasZPROut) const;
+                                 MachineInstr &MI, MachineBasicBlock *BB) const;
   MachineBasicBlock *EmitZTInstr(MachineInstr &MI, MachineBasicBlock *BB,
                                  unsigned Opcode, bool Op0IsDef) const;
   MachineBasicBlock *EmitZero(MachineInstr &MI, MachineBasicBlock *BB) const;

>From c6fdfea8e3cc3789512a9b019b699d4461e05791 Mon Sep 17 00:00:00 2001
From: Caroline Concatto <caroline.concatto at arm.com>
Date: Mon, 3 Jun 2024 15:27:38 +0000
Subject: [PATCH 5/5] Simplify EmitZaInstr with StartIdx

---
 .../Target/AArch64/AArch64ISelLowering.cpp    | 19 ++++++++-----------
 1 file changed, 8 insertions(+), 11 deletions(-)

diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
index 86b280e67bc7a..e0594a14f7ef7 100644
--- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -2876,17 +2876,15 @@ AArch64TargetLowering::EmitZAInstr(unsigned Opc, unsigned BaseReg,
   bool HasZPROut = HasTile && MI.getOperand(0).isReg();
   if (HasZPROut) {
     MIB.add(MI.getOperand(0)); // Output ZPR
-    MIB.addReg(BaseReg + MI.getOperand(1).getImm(),
-               RegState::Define);                    // Output ZA Tile
-    MIB.addReg(BaseReg + MI.getOperand(1).getImm()); // Input Za Tile
-    StartIdx = 2;
+    ++StartIdx;
+  }
+  if (HasTile) {
+    MIB.addReg(BaseReg + MI.getOperand(StartIdx).getImm(),
+               RegState::Define);                           // Output ZA Tile
+    MIB.addReg(BaseReg + MI.getOperand(StartIdx).getImm()); // Input Za Tile
+    StartIdx++;
   } else {
-    if (HasTile) {
-      MIB.addReg(BaseReg + MI.getOperand(0).getImm(), RegState::Define);
-      MIB.addReg(BaseReg + MI.getOperand(0).getImm());
-      StartIdx = 1;
-    } else
-      MIB.addReg(BaseReg, RegState::Define).addReg(BaseReg);
+    MIB.addReg(BaseReg, RegState::Define).addReg(BaseReg);
   }
   for (unsigned I = StartIdx; I < MI.getNumOperands(); ++I)
     MIB.add(MI.getOperand(I));
@@ -2927,7 +2925,6 @@ MachineBasicBlock *AArch64TargetLowering::EmitInstrWithCustomInserter(
       return EmitZAInstr(SMEOrigInstr, AArch64::ZAB0, MI, BB);
     case (AArch64::SMEMatrixTileH):
       return EmitZAInstr(SMEOrigInstr, AArch64::ZAH0, MI, BB);
-      ///*HasTile*/ true, /*HasZPROut*/ false);
     case (AArch64::SMEMatrixTileS):
       return EmitZAInstr(SMEOrigInstr, AArch64::ZAS0, MI, BB);
     case (AArch64::SMEMatrixTileD):

