[flang-commits] [flang] [clang] [llvm] [InstCombine] Canonicalize constant GEPs to i8 source element type (PR #68882)
Nikita Popov via flang-commits
flang-commits at lists.llvm.org
Tue Jan 2 02:42:56 PST 2024
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/68882
>From 0250cce4519a58389c9e94cc27059fa447d9d89e Mon Sep 17 00:00:00 2001
From: Nikita Popov <npopov at redhat.com>
Date: Thu, 12 Oct 2023 14:41:21 +0200
Subject: [PATCH] [InstCombine] Canonicalize constant GEPs to i8 source element
type
---
.../CodeGen/PowerPC/builtins-ppc-pair-mma.c | 10 +-
clang/test/CodeGen/aarch64-ls64-inline-asm.c | 18 +-
.../attr-arm-sve-vector-bits-bitcast.c | 48 +-
.../attr-riscv-rvv-vector-bits-bitcast.c | 24 +-
clang/test/CodeGen/cleanup-destslot-simple.c | 4 +-
.../test/CodeGen/hexagon-brev-ld-ptr-incdec.c | 6 +-
clang/test/CodeGen/ms-intrinsics.c | 12 +-
clang/test/CodeGen/nofpclass.c | 8 +-
clang/test/CodeGen/union-tbaa1.c | 4 +-
.../RelativeVTablesABI/dynamic-cast.cpp | 2 +-
.../RelativeVTablesABI/type-info.cpp | 4 +-
.../CodeGenCXX/microsoft-abi-dynamic-cast.cpp | 12 +-
.../test/CodeGenCXX/microsoft-abi-typeid.cpp | 2 +-
clang/test/CodeGenObjC/arc-foreach.m | 4 +-
.../test/CodeGenObjCXX/arc-cxx11-init-list.mm | 2 +-
clang/test/Headers/__clang_hip_math.hip | 24 +-
clang/test/OpenMP/bug57757.cpp | 12 +-
flang/test/HLFIR/no-block-merging.fir | 2 +-
.../InstCombine/InstructionCombining.cpp | 9 +
llvm/test/Analysis/BasicAA/featuretest.ll | 6 +-
.../CodeGen/AMDGPU/vector-alloca-bitcast.ll | 12 +-
.../BPF/preserve-static-offset/load-inline.ll | 4 +-
.../load-unroll-inline.ll | 4 +-
.../BPF/preserve-static-offset/load-unroll.ll | 8 +-
.../store-unroll-inline.ll | 4 +-
.../Hexagon/autohvx/vector-align-tbaa.ll | 54 +-
llvm/test/Transforms/Coroutines/coro-async.ll | 2 +-
.../coro-retcon-alloca-opaque-ptr.ll | 2 +-
.../Coroutines/coro-retcon-alloca.ll | 2 +-
.../Coroutines/coro-retcon-once-value.ll | 6 +-
.../Coroutines/coro-retcon-resume-values.ll | 8 +-
.../Transforms/Coroutines/coro-swifterror.ll | 2 +-
.../InstCombine/2007-03-25-BadShiftMask.ll | 2 +-
.../InstCombine/2009-01-08-AlignAlloca.ll | 2 +-
.../2009-02-20-InstCombine-SROA.ll | 32 +-
.../X86/x86-addsub-inseltpoison.ll | 2 +-
.../Transforms/InstCombine/X86/x86-addsub.ll | 2 +-
llvm/test/Transforms/InstCombine/add3.ll | 2 +-
llvm/test/Transforms/InstCombine/array.ll | 4 +-
llvm/test/Transforms/InstCombine/assume.ll | 2 +-
llvm/test/Transforms/InstCombine/cast_phi.ll | 4 +-
.../Transforms/InstCombine/catchswitch-phi.ll | 4 +-
.../Transforms/InstCombine/compare-alloca.ll | 2 +-
.../Transforms/InstCombine/extractvalue.ll | 4 +-
.../Transforms/InstCombine/gep-addrspace.ll | 2 +-
.../gep-canonicalize-constant-indices.ll | 18 +-
.../InstCombine/gep-combine-loop-invariant.ll | 6 +-
.../Transforms/InstCombine/gep-custom-dl.ll | 4 +-
.../InstCombine/gep-merge-constant-indices.ll | 14 +-
.../InstCombine/gep-vector-indices.ll | 8 +-
.../test/Transforms/InstCombine/gep-vector.ll | 2 +-
llvm/test/Transforms/InstCombine/gepphigep.ll | 2 +-
.../Transforms/InstCombine/getelementptr.ll | 49 +-
.../Transforms/InstCombine/icmp-custom-dl.ll | 4 +-
llvm/test/Transforms/InstCombine/icmp-gep.ll | 8 +-
.../InstCombine/indexed-gep-compares.ll | 16 +-
llvm/test/Transforms/InstCombine/intptr1.ll | 20 +-
llvm/test/Transforms/InstCombine/intptr2.ll | 4 +-
llvm/test/Transforms/InstCombine/intptr3.ll | 4 +-
llvm/test/Transforms/InstCombine/intptr4.ll | 4 +-
llvm/test/Transforms/InstCombine/intptr5.ll | 4 +-
llvm/test/Transforms/InstCombine/intptr7.ll | 4 +-
.../InstCombine/load-store-forward.ll | 2 +-
llvm/test/Transforms/InstCombine/load.ll | 2 +-
.../InstCombine/loadstore-metadata.ll | 2 +-
llvm/test/Transforms/InstCombine/memchr-5.ll | 48 +-
llvm/test/Transforms/InstCombine/memchr-9.ll | 46 +-
llvm/test/Transforms/InstCombine/memcmp-3.ll | 56 +-
llvm/test/Transforms/InstCombine/memcmp-4.ll | 8 +-
llvm/test/Transforms/InstCombine/memcmp-5.ll | 26 +-
llvm/test/Transforms/InstCombine/memcmp-6.ll | 12 +-
llvm/test/Transforms/InstCombine/memcmp-7.ll | 22 +-
.../Transforms/InstCombine/memcpy_alloca.ll | 2 +-
llvm/test/Transforms/InstCombine/memrchr-5.ll | 64 +-
llvm/test/Transforms/InstCombine/memset2.ll | 2 +-
.../multi-size-address-space-pointer.ll | 14 +-
.../InstCombine/non-integral-pointers.ll | 4 +-
.../test/Transforms/InstCombine/opaque-ptr.ll | 42 +-
.../phi-equal-incoming-pointers.ll | 2 +-
.../Transforms/InstCombine/phi-timeout.ll | 2 +-
llvm/test/Transforms/InstCombine/phi.ll | 4 +-
llvm/test/Transforms/InstCombine/pr39908.ll | 6 +-
llvm/test/Transforms/InstCombine/pr44242.ll | 2 +-
llvm/test/Transforms/InstCombine/pr58901.ll | 7 +-
.../InstCombine/ptr-replace-alloca.ll | 10 +-
.../Transforms/InstCombine/select-cmp-br.ll | 16 +-
.../test/Transforms/InstCombine/select-gep.ll | 16 +-
llvm/test/Transforms/InstCombine/shift.ll | 8 +-
.../sink_sideeffecting_instruction.ll | 2 +-
llvm/test/Transforms/InstCombine/sprintf-2.ll | 16 +-
.../InstCombine/statepoint-cleanup.ll | 8 +-
llvm/test/Transforms/InstCombine/str-int-3.ll | 46 +-
llvm/test/Transforms/InstCombine/str-int-4.ll | 68 +-
llvm/test/Transforms/InstCombine/str-int-5.ll | 54 +-
llvm/test/Transforms/InstCombine/str-int.ll | 2 +-
.../Transforms/InstCombine/strcall-bad-sig.ll | 8 +-
.../Transforms/InstCombine/strcall-no-nul.ll | 26 +-
llvm/test/Transforms/InstCombine/strlen-7.ll | 40 +-
llvm/test/Transforms/InstCombine/strlen-9.ll | 12 +-
llvm/test/Transforms/InstCombine/strncmp-4.ll | 28 +-
llvm/test/Transforms/InstCombine/strncmp-5.ll | 42 +-
llvm/test/Transforms/InstCombine/strncmp-6.ll | 12 +-
llvm/test/Transforms/InstCombine/sub.ll | 6 +-
.../test/Transforms/InstCombine/unpack-fca.ll | 54 +-
.../vec_demanded_elts-inseltpoison.ll | 4 +-
.../InstCombine/vec_demanded_elts.ll | 4 +-
.../vec_gep_scalar_arg-inseltpoison.ll | 2 +-
.../InstCombine/vec_gep_scalar_arg.ll | 2 +-
.../test/Transforms/InstCombine/vscale_gep.ll | 2 +-
llvm/test/Transforms/InstCombine/wcslen-5.ll | 2 +-
.../Transforms/LoopUnroll/ARM/upperbound.ll | 4 +-
llvm/test/Transforms/LoopUnroll/peel-loop.ll | 16 +-
.../AArch64/deterministic-type-shrinkage.ll | 2 +-
.../LoopVectorize/AArch64/intrinsiccost.ll | 8 +-
.../AArch64/sve-cond-inv-loads.ll | 4 +-
.../AArch64/sve-interleaved-accesses.ll | 622 ++++++-------
.../LoopVectorize/AArch64/sve-widen-phi.ll | 166 ++--
.../AArch64/vector-reverse-mask4.ll | 8 +-
.../LoopVectorize/AMDGPU/packed-math.ll | 44 +-
.../Transforms/LoopVectorize/ARM/mve-qabs.ll | 18 +-
.../LoopVectorize/ARM/mve-reductions.ll | 2 +-
.../LoopVectorize/ARM/mve-selectandorcost.ll | 6 +-
.../LoopVectorize/ARM/pointer_iv.ll | 60 +-
.../LoopVectorize/X86/float-induction-x86.ll | 18 +-
.../LoopVectorize/X86/interleaving.ll | 60 +-
.../LoopVectorize/X86/intrinsiccost.ll | 27 +-
.../X86/invariant-store-vectorization.ll | 48 +-
.../LoopVectorize/X86/metadata-enable.ll | 824 +++++++++---------
.../Transforms/LoopVectorize/X86/pr23997.ll | 28 +-
.../LoopVectorize/X86/small-size.ll | 8 +-
...86-interleaved-store-accesses-with-gaps.ll | 4 +-
.../LoopVectorize/consecutive-ptr-uniforms.ll | 4 +-
.../LoopVectorize/extract-last-veclane.ll | 2 +-
.../LoopVectorize/float-induction.ll | 16 +-
.../Transforms/LoopVectorize/induction.ll | 52 +-
.../LoopVectorize/interleaved-accesses.ll | 20 +-
.../LoopVectorize/reduction-inloop-uf4.ll | 6 +-
.../Transforms/LoopVectorize/runtime-check.ll | 2 +-
.../scalar_after_vectorization.ll | 2 +-
.../Transforms/LoopVectorize/vector-geps.ll | 14 +-
.../multiply-fused-dominance.ll | 112 +--
.../multiply-fused-loops.ll | 24 +-
.../multiply-fused-multiple-blocks.ll | 72 +-
.../LowerMatrixIntrinsics/multiply-fused.ll | 134 +--
.../LowerMatrixIntrinsics/multiply-minimal.ll | 10 +-
...ting-sinking-required-for-vectorization.ll | 26 +-
...ple-unreachable-exits-for-vectorization.ll | 22 +-
.../PhaseOrdering/AArch64/quant_4x4.ll | 96 +-
.../AArch64/sinking-vs-if-conversion.ll | 8 +-
.../PhaseOrdering/ARM/arm_mult_q15.ll | 12 +-
.../PhaseOrdering/X86/excessive-unrolling.ll | 6 +-
.../X86/hoist-load-of-baseptr.ll | 10 +-
.../PhaseOrdering/X86/pixel-splat.ll | 2 +-
.../X86/pr48844-br-to-switch-vectorization.ll | 2 +-
.../Transforms/PhaseOrdering/X86/pr50555.ll | 4 +-
.../PhaseOrdering/X86/speculation-vs-tbaa.ll | 2 +-
.../PhaseOrdering/X86/spurious-peeling.ll | 12 +-
.../test/Transforms/PhaseOrdering/X86/vdiv.ll | 12 +-
.../Transforms/PhaseOrdering/X86/vec-shift.ll | 8 +-
llvm/test/Transforms/PhaseOrdering/basic.ll | 8 +-
.../PhaseOrdering/loop-access-checks.ll | 6 +-
llvm/test/Transforms/PhaseOrdering/pr39282.ll | 4 +-
.../PhaseOrdering/simplifycfg-options.ll | 2 +-
.../PhaseOrdering/switch_with_geps.ll | 56 +-
.../SLPVectorizer/AArch64/gather-cost.ll | 12 +-
.../SLPVectorizer/AArch64/gather-reduce.ll | 8 +-
.../SLPVectorizer/AArch64/loadorder.ll | 32 +-
.../WebAssembly/no-vectorize-rotate.ll | 2 +-
.../SLPVectorizer/X86/operandorder.ll | 4 +-
llvm/test/Transforms/SLPVectorizer/X86/opt.ll | 14 +-
.../Transforms/SLPVectorizer/X86/pr46983.ll | 32 +-
.../SLPVectorizer/X86/pr47629-inseltpoison.ll | 272 +++---
.../Transforms/SLPVectorizer/X86/pr47629.ll | 272 +++---
.../SampleProfile/pseudo-probe-instcombine.ll | 10 +-
.../Transforms/Util/strip-gc-relocates.ll | 4 +-
175 files changed, 2388 insertions(+), 2440 deletions(-)
diff --git a/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c b/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
index 3922513e22469a..5422d993ff1575 100644
--- a/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
+++ b/clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
@@ -25,13 +25,13 @@ void test1(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 0
// CHECK-NEXT: store <16 x i8> [[TMP2]], ptr [[RESP:%.*]], align 16
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 1
-// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 1
+// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 16
// CHECK-NEXT: store <16 x i8> [[TMP3]], ptr [[TMP4]], align 16
// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 2
-// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 2
+// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 32
// CHECK-NEXT: store <16 x i8> [[TMP5]], ptr [[TMP6]], align 16
// CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } [[TMP1]], 3
-// CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 3
+// CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 48
// CHECK-NEXT: store <16 x i8> [[TMP7]], ptr [[TMP8]], align 16
// CHECK-NEXT: ret void
//
@@ -60,7 +60,7 @@ void test3(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, unsi
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 0
// CHECK-NEXT: store <16 x i8> [[TMP2]], ptr [[RESP:%.*]], align 16
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 1
-// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 1
+// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 16
// CHECK-NEXT: store <16 x i8> [[TMP3]], ptr [[TMP4]], align 16
// CHECK-NEXT: ret void
//
@@ -1072,7 +1072,7 @@ void test76(unsigned char *vqp, unsigned char *vpp, vector unsigned char vc, uns
// CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 0
// CHECK-NEXT: store <16 x i8> [[TMP2]], ptr [[RESP:%.*]], align 16
// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <16 x i8>, <16 x i8> } [[TMP1]], 1
-// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds <16 x i8>, ptr [[RESP]], i64 1
+// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[RESP]], i64 16
// CHECK-NEXT: store <16 x i8> [[TMP3]], ptr [[TMP4]], align 16
// CHECK-NEXT: ret void
//
diff --git a/clang/test/CodeGen/aarch64-ls64-inline-asm.c b/clang/test/CodeGen/aarch64-ls64-inline-asm.c
index ac2dbe1fa1b31a..744d6919b05ee4 100644
--- a/clang/test/CodeGen/aarch64-ls64-inline-asm.c
+++ b/clang/test/CodeGen/aarch64-ls64-inline-asm.c
@@ -16,8 +16,8 @@ void load(struct foo *output, void *addr)
// CHECK-LABEL: @store(
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP1:%.*]] = load i512, ptr [[INPUT:%.*]], align 8
-// CHECK-NEXT: tail call void asm sideeffect "st64b $0,[$1]", "r,r,~{memory}"(i512 [[TMP1]], ptr [[ADDR:%.*]]) #[[ATTR1]], !srcloc !3
+// CHECK-NEXT: [[TMP0:%.*]] = load i512, ptr [[INPUT:%.*]], align 8
+// CHECK-NEXT: tail call void asm sideeffect "st64b $0,[$1]", "r,r,~{memory}"(i512 [[TMP0]], ptr [[ADDR:%.*]]) #[[ATTR1]], !srcloc !3
// CHECK-NEXT: ret void
//
void store(const struct foo *input, void *addr)
@@ -29,25 +29,25 @@ void store(const struct foo *input, void *addr)
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[IN:%.*]], align 4, !tbaa [[TBAA4:![0-9]+]]
// CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP0]] to i64
-// CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 1
+// CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 4
// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4, !tbaa [[TBAA4]]
// CHECK-NEXT: [[CONV2:%.*]] = sext i32 [[TMP1]] to i64
-// CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 4
+// CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 16
// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4, !tbaa [[TBAA4]]
// CHECK-NEXT: [[CONV5:%.*]] = sext i32 [[TMP2]] to i64
-// CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 16
+// CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 64
// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX7]], align 4, !tbaa [[TBAA4]]
// CHECK-NEXT: [[CONV8:%.*]] = sext i32 [[TMP3]] to i64
-// CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 25
+// CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 100
// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX10]], align 4, !tbaa [[TBAA4]]
// CHECK-NEXT: [[CONV11:%.*]] = sext i32 [[TMP4]] to i64
-// CHECK-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 36
+// CHECK-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 144
// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX13]], align 4, !tbaa [[TBAA4]]
// CHECK-NEXT: [[CONV14:%.*]] = sext i32 [[TMP5]] to i64
-// CHECK-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 49
+// CHECK-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 196
// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX16]], align 4, !tbaa [[TBAA4]]
// CHECK-NEXT: [[CONV17:%.*]] = sext i32 [[TMP6]] to i64
-// CHECK-NEXT: [[ARRAYIDX19:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 64
+// CHECK-NEXT: [[ARRAYIDX19:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 256
// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX19]], align 4, !tbaa [[TBAA4]]
// CHECK-NEXT: [[CONV20:%.*]] = sext i32 [[TMP7]] to i64
// CHECK-NEXT: [[S_SROA_10_0_INSERT_EXT:%.*]] = zext i64 [[CONV20]] to i512
diff --git a/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c b/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c
index 22e2e0c2ff102d..323afb64591249 100644
--- a/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c
+++ b/clang/test/CodeGen/attr-arm-sve-vector-bits-bitcast.c
@@ -30,21 +30,21 @@ DEFINE_STRUCT(bool)
// CHECK-128-LABEL: @read_int64(
// CHECK-128-NEXT: entry:
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 16, !tbaa [[TBAA2:![0-9]+]]
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP0]], i64 0)
// CHECK-128-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]]
//
// CHECK-256-LABEL: @read_int64(
// CHECK-256-NEXT: entry:
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 16, !tbaa [[TBAA2:![0-9]+]]
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v4i64(<vscale x 2 x i64> undef, <4 x i64> [[TMP0]], i64 0)
// CHECK-256-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]]
//
// CHECK-512-LABEL: @read_int64(
// CHECK-512-NEXT: entry:
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i64>, ptr [[Y]], align 16, !tbaa [[TBAA2:![0-9]+]]
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.insert.nxv2i64.v8i64(<vscale x 2 x i64> undef, <8 x i64> [[TMP0]], i64 0)
// CHECK-512-NEXT: ret <vscale x 2 x i64> [[CAST_SCALABLE]]
@@ -56,21 +56,21 @@ svint64_t read_int64(struct struct_int64 *s) {
// CHECK-128-LABEL: @write_int64(
// CHECK-128-NEXT: entry:
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i64> @llvm.vector.extract.v2i64.nxv2i64(<vscale x 2 x i64> [[X:%.*]], i64 0)
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-128-NEXT: ret void
//
// CHECK-256-LABEL: @write_int64(
// CHECK-256-NEXT: entry:
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i64> @llvm.vector.extract.v4i64.nxv2i64(<vscale x 2 x i64> [[X:%.*]], i64 0)
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-256-NEXT: ret void
//
// CHECK-512-LABEL: @write_int64(
// CHECK-512-NEXT: entry:
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i64> @llvm.vector.extract.v8i64.nxv2i64(<vscale x 2 x i64> [[X:%.*]], i64 0)
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
// CHECK-512-NEXT: store <8 x i64> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-512-NEXT: ret void
//
@@ -84,21 +84,21 @@ void write_int64(struct struct_int64 *s, svint64_t x) {
// CHECK-128-LABEL: @read_float64(
// CHECK-128-NEXT: entry:
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v2f64(<vscale x 2 x double> undef, <2 x double> [[TMP0]], i64 0)
// CHECK-128-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]]
//
// CHECK-256-LABEL: @read_float64(
// CHECK-256-NEXT: entry:
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v4f64(<vscale x 2 x double> undef, <4 x double> [[TMP0]], i64 0)
// CHECK-256-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]]
//
// CHECK-512-LABEL: @read_float64(
// CHECK-512-NEXT: entry:
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x double>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x double> @llvm.vector.insert.nxv2f64.v8f64(<vscale x 2 x double> undef, <8 x double> [[TMP0]], i64 0)
// CHECK-512-NEXT: ret <vscale x 2 x double> [[CAST_SCALABLE]]
@@ -110,21 +110,21 @@ svfloat64_t read_float64(struct struct_float64 *s) {
// CHECK-128-LABEL: @write_float64(
// CHECK-128-NEXT: entry:
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x double> @llvm.vector.extract.v2f64.nxv2f64(<vscale x 2 x double> [[X:%.*]], i64 0)
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-128-NEXT: ret void
//
// CHECK-256-LABEL: @write_float64(
// CHECK-256-NEXT: entry:
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x double> @llvm.vector.extract.v4f64.nxv2f64(<vscale x 2 x double> [[X:%.*]], i64 0)
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-256-NEXT: ret void
//
// CHECK-512-LABEL: @write_float64(
// CHECK-512-NEXT: entry:
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x double> @llvm.vector.extract.v8f64.nxv2f64(<vscale x 2 x double> [[X:%.*]], i64 0)
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
// CHECK-512-NEXT: store <8 x double> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-512-NEXT: ret void
//
@@ -138,21 +138,21 @@ void write_float64(struct struct_float64 *s, svfloat64_t x) {
// CHECK-128-LABEL: @read_bfloat16(
// CHECK-128-NEXT: entry:
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v8bf16(<vscale x 8 x bfloat> undef, <8 x bfloat> [[TMP0]], i64 0)
// CHECK-128-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]]
//
// CHECK-256-LABEL: @read_bfloat16(
// CHECK-256-NEXT: entry:
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: [[TMP0:%.*]] = load <16 x bfloat>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v16bf16(<vscale x 8 x bfloat> undef, <16 x bfloat> [[TMP0]], i64 0)
// CHECK-256-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]]
//
// CHECK-512-LABEL: @read_bfloat16(
// CHECK-512-NEXT: entry:
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
// CHECK-512-NEXT: [[TMP0:%.*]] = load <32 x bfloat>, ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.insert.nxv8bf16.v32bf16(<vscale x 8 x bfloat> undef, <32 x bfloat> [[TMP0]], i64 0)
// CHECK-512-NEXT: ret <vscale x 8 x bfloat> [[CAST_SCALABLE]]
@@ -164,21 +164,21 @@ svbfloat16_t read_bfloat16(struct struct_bfloat16 *s) {
// CHECK-128-LABEL: @write_bfloat16(
// CHECK-128-NEXT: entry:
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x bfloat> @llvm.vector.extract.v8bf16.nxv8bf16(<vscale x 8 x bfloat> [[X:%.*]], i64 0)
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: store <8 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-128-NEXT: ret void
//
// CHECK-256-LABEL: @write_bfloat16(
// CHECK-256-NEXT: entry:
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <16 x bfloat> @llvm.vector.extract.v16bf16.nxv8bf16(<vscale x 8 x bfloat> [[X:%.*]], i64 0)
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: store <16 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-256-NEXT: ret void
//
// CHECK-512-LABEL: @write_bfloat16(
// CHECK-512-NEXT: entry:
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <32 x bfloat> @llvm.vector.extract.v32bf16.nxv8bf16(<vscale x 8 x bfloat> [[X:%.*]], i64 0)
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BFLOAT16:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 64
// CHECK-512-NEXT: store <32 x bfloat> [[CAST_FIXED]], ptr [[Y]], align 16, !tbaa [[TBAA2]]
// CHECK-512-NEXT: ret void
//
@@ -192,7 +192,7 @@ void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) {
// CHECK-128-LABEL: @read_bool(
// CHECK-128-NEXT: entry:
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 2
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i8>, ptr [[Y]], align 2, !tbaa [[TBAA2]]
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v2i8(<vscale x 2 x i8> undef, <2 x i8> [[TMP0]], i64 0)
// CHECK-128-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1>
@@ -200,7 +200,7 @@ void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) {
//
// CHECK-256-LABEL: @read_bool(
// CHECK-256-NEXT: entry:
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 4
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[Y]], align 2, !tbaa [[TBAA2]]
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v4i8(<vscale x 2 x i8> undef, <4 x i8> [[TMP0]], i64 0)
// CHECK-256-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1>
@@ -208,7 +208,7 @@ void write_bfloat16(struct struct_bfloat16 *s, svbfloat16_t x) {
//
// CHECK-512-LABEL: @read_bool(
// CHECK-512-NEXT: entry:
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
// CHECK-512-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[Y]], align 2, !tbaa [[TBAA2]]
// CHECK-512-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 2 x i8> @llvm.vector.insert.nxv2i8.v8i8(<vscale x 2 x i8> undef, <8 x i8> [[TMP0]], i64 0)
// CHECK-512-NEXT: [[TMP1:%.*]] = bitcast <vscale x 2 x i8> [[CAST_SCALABLE]] to <vscale x 16 x i1>
@@ -222,7 +222,7 @@ svbool_t read_bool(struct struct_bool *s) {
// CHECK-128-NEXT: entry:
// CHECK-128-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X:%.*]] to <vscale x 2 x i8>
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i8> @llvm.vector.extract.v2i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0)
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 2
// CHECK-128-NEXT: store <2 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[TBAA2]]
// CHECK-128-NEXT: ret void
//
@@ -230,7 +230,7 @@ svbool_t read_bool(struct struct_bool *s) {
// CHECK-256-NEXT: entry:
// CHECK-256-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X:%.*]] to <vscale x 2 x i8>
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i8> @llvm.vector.extract.v4i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0)
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 4
// CHECK-256-NEXT: store <4 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[TBAA2]]
// CHECK-256-NEXT: ret void
//
@@ -238,7 +238,7 @@ svbool_t read_bool(struct struct_bool *s) {
// CHECK-512-NEXT: entry:
// CHECK-512-NEXT: [[TMP0:%.*]] = bitcast <vscale x 16 x i1> [[X:%.*]] to <vscale x 2 x i8>
// CHECK-512-NEXT: [[CAST_FIXED:%.*]] = tail call <8 x i8> @llvm.vector.extract.v8i8.nxv2i8(<vscale x 2 x i8> [[TMP0]], i64 0)
-// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_BOOL:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-512-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
// CHECK-512-NEXT: store <8 x i8> [[CAST_FIXED]], ptr [[Y]], align 2, !tbaa [[TBAA2]]
// CHECK-512-NEXT: ret void
//
diff --git a/clang/test/CodeGen/attr-riscv-rvv-vector-bits-bitcast.c b/clang/test/CodeGen/attr-riscv-rvv-vector-bits-bitcast.c
index 886af083f1c009..6d290f9cb47bea 100644
--- a/clang/test/CodeGen/attr-riscv-rvv-vector-bits-bitcast.c
+++ b/clang/test/CodeGen/attr-riscv-rvv-vector-bits-bitcast.c
@@ -35,21 +35,21 @@ DEFINE_STRUCT(float64m1)
// CHECK-64-LABEL: @read_int64m1(
// CHECK-64-NEXT: entry:
-// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x i64>, ptr [[Y]], align 8, !tbaa [[TBAA4:![0-9]+]]
// CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v1i64(<vscale x 1 x i64> undef, <1 x i64> [[TMP0]], i64 0)
// CHECK-64-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
//
// CHECK-128-LABEL: @read_int64m1(
// CHECK-128-NEXT: entry:
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[Y]], align 8, !tbaa [[TBAA4:![0-9]+]]
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v2i64(<vscale x 1 x i64> undef, <2 x i64> [[TMP0]], i64 0)
// CHECK-128-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
//
// CHECK-256-LABEL: @read_int64m1(
// CHECK-256-NEXT: entry:
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x i64>, ptr [[Y]], align 8, !tbaa [[TBAA4:![0-9]+]]
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x i64> @llvm.vector.insert.nxv1i64.v4i64(<vscale x 1 x i64> undef, <4 x i64> [[TMP0]], i64 0)
// CHECK-256-NEXT: ret <vscale x 1 x i64> [[CAST_SCALABLE]]
@@ -61,21 +61,21 @@ vint64m1_t read_int64m1(struct struct_int64m1 *s) {
// CHECK-64-LABEL: @write_int64m1(
// CHECK-64-NEXT: entry:
// CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x i64> @llvm.vector.extract.v1i64.nxv1i64(<vscale x 1 x i64> [[X:%.*]], i64 0)
-// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
// CHECK-64-NEXT: store <1 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-64-NEXT: ret void
//
// CHECK-128-LABEL: @write_int64m1(
// CHECK-128-NEXT: entry:
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x i64> @llvm.vector.extract.v2i64.nxv1i64(<vscale x 1 x i64> [[X:%.*]], i64 0)
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: store <2 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-128-NEXT: ret void
//
// CHECK-256-LABEL: @write_int64m1(
// CHECK-256-NEXT: entry:
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x i64> @llvm.vector.extract.v4i64.nxv1i64(<vscale x 1 x i64> [[X:%.*]], i64 0)
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_INT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: store <4 x i64> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-256-NEXT: ret void
//
@@ -89,21 +89,21 @@ void write_int64m1(struct struct_int64m1 *s, vint64m1_t x) {
// CHECK-64-LABEL: @read_float64m1(
// CHECK-64-NEXT: entry:
-// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
// CHECK-64-NEXT: [[TMP0:%.*]] = load <1 x double>, ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-64-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v1f64(<vscale x 1 x double> undef, <1 x double> [[TMP0]], i64 0)
// CHECK-64-NEXT: ret <vscale x 1 x double> [[CAST_SCALABLE]]
//
// CHECK-128-LABEL: @read_float64m1(
// CHECK-128-NEXT: entry:
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-128-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v2f64(<vscale x 1 x double> undef, <2 x double> [[TMP0]], i64 0)
// CHECK-128-NEXT: ret <vscale x 1 x double> [[CAST_SCALABLE]]
//
// CHECK-256-LABEL: @read_float64m1(
// CHECK-256-NEXT: entry:
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: [[TMP0:%.*]] = load <4 x double>, ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-256-NEXT: [[CAST_SCALABLE:%.*]] = tail call <vscale x 1 x double> @llvm.vector.insert.nxv1f64.v4f64(<vscale x 1 x double> undef, <4 x double> [[TMP0]], i64 0)
// CHECK-256-NEXT: ret <vscale x 1 x double> [[CAST_SCALABLE]]
@@ -115,21 +115,21 @@ vfloat64m1_t read_float64m1(struct struct_float64m1 *s) {
// CHECK-64-LABEL: @write_float64m1(
// CHECK-64-NEXT: entry:
// CHECK-64-NEXT: [[CAST_FIXED:%.*]] = tail call <1 x double> @llvm.vector.extract.v1f64.nxv1f64(<vscale x 1 x double> [[X:%.*]], i64 0)
-// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-64-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 8
// CHECK-64-NEXT: store <1 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-64-NEXT: ret void
//
// CHECK-128-LABEL: @write_float64m1(
// CHECK-128-NEXT: entry:
// CHECK-128-NEXT: [[CAST_FIXED:%.*]] = tail call <2 x double> @llvm.vector.extract.v2f64.nxv1f64(<vscale x 1 x double> [[X:%.*]], i64 0)
-// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-128-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 16
// CHECK-128-NEXT: store <2 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-128-NEXT: ret void
//
// CHECK-256-LABEL: @write_float64m1(
// CHECK-256-NEXT: entry:
// CHECK-256-NEXT: [[CAST_FIXED:%.*]] = tail call <4 x double> @llvm.vector.extract.v4f64.nxv1f64(<vscale x 1 x double> [[X:%.*]], i64 0)
-// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_STRUCT_FLOAT64M1:%.*]], ptr [[S:%.*]], i64 0, i32 1
+// CHECK-256-NEXT: [[Y:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 32
// CHECK-256-NEXT: store <4 x double> [[CAST_FIXED]], ptr [[Y]], align 8, !tbaa [[TBAA4]]
// CHECK-256-NEXT: ret void
//
diff --git a/clang/test/CodeGen/cleanup-destslot-simple.c b/clang/test/CodeGen/cleanup-destslot-simple.c
index 075afbba18cdb9..a02841a5e2e2ef 100644
--- a/clang/test/CodeGen/cleanup-destslot-simple.c
+++ b/clang/test/CodeGen/cleanup-destslot-simple.c
@@ -101,8 +101,8 @@
// CHECK-KMSAN-NEXT: call void @__msan_warning(i32 [[TMP9]]) #[[ATTR3:[0-9]+]], !dbg [[DBG20]]
// CHECK-KMSAN-NEXT: br label [[TMP10]], !dbg [[DBG20]]
// CHECK-KMSAN: 10:
-// CHECK-KMSAN-NEXT: [[RETVAL_ORIGIN:%.*]] = getelementptr { [100 x i64], [100 x i64], [100 x i64], [100 x i64], i64, [200 x i32], i32, i32 }, ptr [[TMP0]], i64 0, i32 6
-// CHECK-KMSAN-NEXT: [[RETVAL_SHADOW:%.*]] = getelementptr { [100 x i64], [100 x i64], [100 x i64], [100 x i64], i64, [200 x i32], i32, i32 }, ptr [[TMP0]], i64 0, i32 1
+// CHECK-KMSAN-NEXT: [[RETVAL_ORIGIN:%.*]] = getelementptr i8, ptr [[TMP0]], i64 4008
+// CHECK-KMSAN-NEXT: [[RETVAL_SHADOW:%.*]] = getelementptr i8, ptr [[TMP0]], i64 800
// CHECK-KMSAN-NEXT: [[TMP11:%.*]] = load i32, ptr [[P_0_P_0_P_0_P_0_]], align 4, !dbg [[DBG20]], !tbaa [[TBAA11]]
// CHECK-KMSAN-NEXT: [[TMP12:%.*]] = call { ptr, ptr } @__msan_metadata_ptr_for_load_4(ptr nonnull [[P_0_P_0_P_0_P_0_]]) #[[ATTR2]], !dbg [[DBG20]]
// CHECK-KMSAN-NEXT: [[TMP13:%.*]] = extractvalue { ptr, ptr } [[TMP12]], 0, !dbg [[DBG20]]
diff --git a/clang/test/CodeGen/hexagon-brev-ld-ptr-incdec.c b/clang/test/CodeGen/hexagon-brev-ld-ptr-incdec.c
index c9b97da34d59e7..7802168de4d75a 100644
--- a/clang/test/CodeGen/hexagon-brev-ld-ptr-incdec.c
+++ b/clang/test/CodeGen/hexagon-brev-ld-ptr-incdec.c
@@ -6,9 +6,9 @@
// the return value will be the value in A[2]
// CHECK: @brev_ptr_inc
// CHECK-DAG: llvm.hexagon.L2.loadri.pbr
-// CHECK-DAG: getelementptr{{.*}}i32 1
-// CHECK-NOT: getelementptr{{.*}}i32 2
-// CHECK-NOT: getelementptr{{.*}}i32 1
+// CHECK-DAG: getelementptr inbounds i8, {{.*}}i32 4
+// CHECK-NOT: getelementptr inbounds i8, {{.*}}i32 8
+// CHECK-NOT: getelementptr inbounds i8, {{.*}}i32 4
int brev_ptr_inc(int A[], int B[]) {
int *p0 = &B[0];
int *p1 = &A[0];
diff --git a/clang/test/CodeGen/ms-intrinsics.c b/clang/test/CodeGen/ms-intrinsics.c
index ffab8b998d8bdf..5bb003d1f91fc0 100644
--- a/clang/test/CodeGen/ms-intrinsics.c
+++ b/clang/test/CodeGen/ms-intrinsics.c
@@ -156,7 +156,7 @@ unsigned char test_BitScanForward(unsigned long *Index, unsigned long Mask) {
// CHECK: [[RESULT:%[a-z0-9._]+]] = phi i8 [ 0, %[[ISZERO_LABEL:[a-z0-9._]+]] ], [ 1, %[[ISNOTZERO_LABEL]] ]
// CHECK: ret i8 [[RESULT]]
// CHECK: [[ISNOTZERO_LABEL]]:
-// CHECK: [[IDXGEP:%[a-z0-9._]+]] = getelementptr inbounds i32, ptr %Index, {{i64|i32}} 1
+// CHECK: [[IDXGEP:%[a-z0-9._]+]] = getelementptr inbounds i8, ptr %Index, {{i64|i32}} 4
// CHECK: [[INDEX:%[0-9]+]] = tail call i32 @llvm.cttz.i32(i32 %Mask, i1 true)
// CHECK: store i32 [[INDEX]], ptr [[IDXGEP]], align 4
// CHECK: br label %[[END_LABEL]]
@@ -171,7 +171,7 @@ unsigned char test_BitScanReverse(unsigned long *Index, unsigned long Mask) {
// CHECK: [[RESULT:%[a-z0-9._]+]] = phi i8 [ 0, %[[ISZERO_LABEL:[a-z0-9._]+]] ], [ 1, %[[ISNOTZERO_LABEL]] ]
// CHECK: ret i8 [[RESULT]]
// CHECK: [[ISNOTZERO_LABEL]]:
-// CHECK: [[IDXGEP:%[a-z0-9._]+]] = getelementptr inbounds i32, ptr %Index, {{i64|i32}} 1
+// CHECK: [[IDXGEP:%[a-z0-9._]+]] = getelementptr inbounds i8, ptr %Index, {{i64|i32}} 4
// CHECK: [[REVINDEX:%[0-9]+]] = tail call i32 @llvm.ctlz.i32(i32 %Mask, i1 true)
// CHECK: [[INDEX:%[0-9]+]] = xor i32 [[REVINDEX]], 31
// CHECK: store i32 [[INDEX]], ptr [[IDXGEP]], align 4
@@ -437,10 +437,10 @@ unsigned char test_InterlockedCompareExchange128(
++ExchangeLow, ++ComparandResult);
}
// CHECK-64: define{{.*}}i8 @test_InterlockedCompareExchange128(ptr{{[a-z_ ]*}}%Destination, i64{{[a-z_ ]*}}%ExchangeHigh, i64{{[a-z_ ]*}}%ExchangeLow, ptr{{[a-z_ ]*}}%ComparandResult){{.*}}{
-// CHECK-64: %incdec.ptr = getelementptr inbounds i64, ptr %Destination, i64 1
+// CHECK-64: %incdec.ptr = getelementptr inbounds i8, ptr %Destination, i64 8
// CHECK-64: %inc = add nsw i64 %ExchangeHigh, 1
// CHECK-64: %inc1 = add nsw i64 %ExchangeLow, 1
-// CHECK-64: %incdec.ptr2 = getelementptr inbounds i64, ptr %ComparandResult, i64 1
+// CHECK-64: %incdec.ptr2 = getelementptr inbounds i8, ptr %ComparandResult, i64 8
// CHECK-64: [[EH:%[0-9]+]] = zext i64 %inc to i128
// CHECK-64: [[EL:%[0-9]+]] = zext i64 %inc1 to i128
// CHECK-64: [[EHS:%[0-9]+]] = shl nuw i128 [[EH]], 64
@@ -486,7 +486,7 @@ short test_InterlockedIncrement16(short volatile *Addend) {
return _InterlockedIncrement16(++Addend);
}
// CHECK: define{{.*}}i16 @test_InterlockedIncrement16(ptr{{[a-z_ ]*}}%Addend){{.*}}{
-// CHECK: %incdec.ptr = getelementptr inbounds i16, ptr %Addend, {{i64|i32}} 1
+// CHECK: %incdec.ptr = getelementptr inbounds i8, ptr %Addend, {{i64|i32}} 2
// CHECK: [[TMP:%[0-9]+]] = atomicrmw add ptr %incdec.ptr, i16 1 seq_cst, align 2
// CHECK: [[RESULT:%[0-9]+]] = add i16 [[TMP]], 1
// CHECK: ret i16 [[RESULT]]
@@ -496,7 +496,7 @@ long test_InterlockedIncrement(long volatile *Addend) {
return _InterlockedIncrement(++Addend);
}
// CHECK: define{{.*}}i32 @test_InterlockedIncrement(ptr{{[a-z_ ]*}}%Addend){{.*}}{
-// CHECK: %incdec.ptr = getelementptr inbounds i32, ptr %Addend, {{i64|i32}} 1
+// CHECK: %incdec.ptr = getelementptr inbounds i8, ptr %Addend, {{i64|i32}} 4
// CHECK: [[TMP:%[0-9]+]] = atomicrmw add ptr %incdec.ptr, i32 1 seq_cst, align 4
// CHECK: [[RESULT:%[0-9]+]] = add i32 [[TMP]], 1
// CHECK: ret i32 [[RESULT]]
diff --git a/clang/test/CodeGen/nofpclass.c b/clang/test/CodeGen/nofpclass.c
index daaf24e77b09d7..dd90d02f7759b6 100644
--- a/clang/test/CodeGen/nofpclass.c
+++ b/clang/test/CodeGen/nofpclass.c
@@ -925,9 +925,9 @@ _Complex _Float16 defined_complex_func_f16_ret(_Complex _Float16 c) {
// CLFINITEONLY-NEXT: [[INDIRECT_ARG_TEMP:%.*]] = alloca { double, double }, align 8
// CLFINITEONLY-NEXT: [[CONV:%.*]] = fpext float [[F32]] to double
// CLFINITEONLY-NEXT: [[CF16_REAL:%.*]] = load half, ptr [[CF16]], align 8
-// CLFINITEONLY-NEXT: [[CF16_IMAGP:%.*]] = getelementptr inbounds { half, half }, ptr [[CF16]], i64 0, i32 1
+// CLFINITEONLY-NEXT: [[CF16_IMAGP:%.*]] = getelementptr inbounds i8, ptr [[CF16]], i64 2
// CLFINITEONLY-NEXT: [[CF16_IMAG:%.*]] = load half, ptr [[CF16_IMAGP]], align 2
-// CLFINITEONLY-NEXT: [[INDIRECT_ARG_TEMP_IMAGP:%.*]] = getelementptr inbounds { double, double }, ptr [[INDIRECT_ARG_TEMP]], i64 0, i32 1
+// CLFINITEONLY-NEXT: [[INDIRECT_ARG_TEMP_IMAGP:%.*]] = getelementptr inbounds i8, ptr [[INDIRECT_ARG_TEMP]], i64 8
// CLFINITEONLY-NEXT: store double [[CF64_COERCE0]], ptr [[INDIRECT_ARG_TEMP]], align 8
// CLFINITEONLY-NEXT: store double [[CF64_COERCE1]], ptr [[INDIRECT_ARG_TEMP_IMAGP]], align 8
// CLFINITEONLY-NEXT: [[COERCE5_SROA_0_0_VEC_INSERT:%.*]] = insertelement <2 x half> poison, half [[CF16_REAL]], i64 0
@@ -1176,9 +1176,9 @@ float call_variadic(float f32, double f64, _Float16 f16,
// CLFINITEONLY-NEXT: [[INDIRECT_ARG_TEMP:%.*]] = alloca { double, double }, align 8
// CLFINITEONLY-NEXT: [[CONV:%.*]] = fpext float [[F32]] to double
// CLFINITEONLY-NEXT: [[CF16_REAL:%.*]] = load half, ptr [[CF16]], align 8
-// CLFINITEONLY-NEXT: [[CF16_IMAGP:%.*]] = getelementptr inbounds { half, half }, ptr [[CF16]], i64 0, i32 1
+// CLFINITEONLY-NEXT: [[CF16_IMAGP:%.*]] = getelementptr inbounds i8, ptr [[CF16]], i64 2
// CLFINITEONLY-NEXT: [[CF16_IMAG:%.*]] = load half, ptr [[CF16_IMAGP]], align 2
-// CLFINITEONLY-NEXT: [[INDIRECT_ARG_TEMP_IMAGP:%.*]] = getelementptr inbounds { double, double }, ptr [[INDIRECT_ARG_TEMP]], i64 0, i32 1
+// CLFINITEONLY-NEXT: [[INDIRECT_ARG_TEMP_IMAGP:%.*]] = getelementptr inbounds i8, ptr [[INDIRECT_ARG_TEMP]], i64 8
// CLFINITEONLY-NEXT: store double [[CF64_COERCE0]], ptr [[INDIRECT_ARG_TEMP]], align 8
// CLFINITEONLY-NEXT: store double [[CF64_COERCE1]], ptr [[INDIRECT_ARG_TEMP_IMAGP]], align 8
// CLFINITEONLY-NEXT: [[COERCE5_SROA_0_0_VEC_INSERT:%.*]] = insertelement <2 x half> poison, half [[CF16_REAL]], i64 0
diff --git a/clang/test/CodeGen/union-tbaa1.c b/clang/test/CodeGen/union-tbaa1.c
index f6f10b3e5b1e12..a5faa8269aed69 100644
--- a/clang/test/CodeGen/union-tbaa1.c
+++ b/clang/test/CodeGen/union-tbaa1.c
@@ -27,10 +27,10 @@ void bar(vect32 p[][2]);
// CHECK-NEXT: store i32 [[TMP3]], ptr [[VEC]], align 4, !tbaa [[TBAA2]]
// CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[INDEX]], align 4, !tbaa [[TBAA2]]
// CHECK-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [4 x [2 x %union.vect32]], ptr [[TMP]], i32 0, i32 [[TMP4]], i32 1
-// CHECK-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds [2 x i16], ptr [[ARRAYIDX14]], i32 0, i32 1
+// CHECK-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds i8, ptr [[ARRAYIDX14]], i32 2
// CHECK-NEXT: [[TMP5:%.*]] = load i16, ptr [[ARRAYIDX15]], align 2, !tbaa [[TBAA6]]
// CHECK-NEXT: [[CONV16:%.*]] = zext i16 [[TMP5]] to i32
-// CHECK-NEXT: [[ARRAYIDX17:%.*]] = getelementptr inbounds i32, ptr [[VEC]], i32 1
+// CHECK-NEXT: [[ARRAYIDX17:%.*]] = getelementptr inbounds i8, ptr [[VEC]], i32 4
// CHECK-NEXT: store i32 [[CONV16]], ptr [[ARRAYIDX17]], align 4, !tbaa [[TBAA2]]
// CHECK-NEXT: call void @bar(ptr noundef nonnull [[TMP]]) #[[ATTR3]]
// CHECK-NEXT: call void @llvm.lifetime.end.p0(i64 32, ptr nonnull [[TMP]]) #[[ATTR3]]
diff --git a/clang/test/CodeGenCXX/RelativeVTablesABI/dynamic-cast.cpp b/clang/test/CodeGenCXX/RelativeVTablesABI/dynamic-cast.cpp
index 20b9557bcc1a59..4fef80f051ac98 100644
--- a/clang/test/CodeGenCXX/RelativeVTablesABI/dynamic-cast.cpp
+++ b/clang/test/CodeGenCXX/RelativeVTablesABI/dynamic-cast.cpp
@@ -33,7 +33,7 @@
// CHECK-NEXT: br i1 [[isnull]], label %[[dynamic_cast_end:[a-z0-9._]+]], label %[[dynamic_cast_notnull:[a-z0-9._]+]]
// CHECK: [[dynamic_cast_notnull]]:
// CHECK-DAG: [[vtable:%[a-z0-9]+]] = load ptr, ptr %b, align 8
-// CHECK-DAG: [[offset_ptr:%.+]] = getelementptr inbounds i32, ptr [[vtable]], i64 -2
+// CHECK-DAG: [[offset_ptr:%.+]] = getelementptr inbounds i8, ptr [[vtable]], i64 -8
// CHECK-DAG: [[offset_to_top:%.+]] = load i32, ptr [[offset_ptr]], align 4
// CHECK-DAG: [[offset_to_top2:%.+]] = sext i32 [[offset_to_top]] to i64
// CHECK-DAG: [[casted:%.+]] = getelementptr inbounds i8, ptr %b, i64 [[offset_to_top2]]
diff --git a/clang/test/CodeGenCXX/RelativeVTablesABI/type-info.cpp b/clang/test/CodeGenCXX/RelativeVTablesABI/type-info.cpp
index bc10461187b758..83c8956f9ff163 100644
--- a/clang/test/CodeGenCXX/RelativeVTablesABI/type-info.cpp
+++ b/clang/test/CodeGenCXX/RelativeVTablesABI/type-info.cpp
@@ -2,8 +2,6 @@
// RUN: %clang_cc1 %s -triple=aarch64-unknown-fuchsia -O3 -S -o - -emit-llvm -fcxx-exceptions -fexceptions | FileCheck %s
-// CHECK: %"class.std::type_info" = type { ptr, ptr }
-
// CHECK: $_ZTI1A.rtti_proxy = comdat any
// CHECK: $_ZTI1B.rtti_proxy = comdat any
@@ -37,7 +35,7 @@
// CHECK-NEXT: [[vtable:%[a-z0-9]+]] = load ptr, ptr %a
// CHECK-NEXT: [[type_info_ptr:%[0-9]+]] = tail call ptr @llvm.load.relative.i32(ptr [[vtable]], i32 -4)
// CHECK-NEXT: [[type_info_ptr2:%[0-9]+]] = load ptr, ptr [[type_info_ptr]], align 8
-// CHECK-NEXT: [[name_ptr:%[a-z0-9._]+]] = getelementptr inbounds %"class.std::type_info", ptr [[type_info_ptr2]], i64 0, i32 1
+// CHECK-NEXT: [[name_ptr:%[a-z0-9._]+]] = getelementptr inbounds i8, ptr [[type_info_ptr2]], i64 8
// CHECK-NEXT: [[name:%[0-9]+]] = load ptr, ptr [[name_ptr]], align 8
// CHECK-NEXT: [[eq:%[a-z0-9.]+]] = icmp eq ptr [[name]], @_ZTS1B
// CHECK-NEXT: ret i1 [[eq]]
diff --git a/clang/test/CodeGenCXX/microsoft-abi-dynamic-cast.cpp b/clang/test/CodeGenCXX/microsoft-abi-dynamic-cast.cpp
index 5c4b86744287ec..4b046acf0d8789 100644
--- a/clang/test/CodeGenCXX/microsoft-abi-dynamic-cast.cpp
+++ b/clang/test/CodeGenCXX/microsoft-abi-dynamic-cast.cpp
@@ -18,7 +18,7 @@ T* test1(V* x) { return &dynamic_cast<T&>(*x); }
T* test2(A* x) { return &dynamic_cast<T&>(*x); }
// CHECK-LABEL: define dso_local noundef ptr @"?test2@@YAPAUT@@PAUA@@@Z"(ptr noundef %x)
// CHECK: [[VBTBL:%.*]] = load ptr, ptr %x, align 4
-// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i32, ptr [[VBTBL]], i32 1
+// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i8, ptr [[VBTBL]], i32 4
// CHECK-NEXT: [[VBOFFS:%.*]] = load i32, ptr [[VBOFFP]], align 4
// CHECK-NEXT: [[ADJ:%.*]] = getelementptr inbounds i8, ptr %x, i32 [[VBOFFS]]
// CHECK-NEXT: [[CALL:%.*]] = tail call ptr @__RTDynamicCast(ptr nonnull [[ADJ]], i32 [[VBOFFS]], ptr nonnull @"??_R0?AUA@@@8", ptr nonnull @"??_R0?AUT@@@8", i32 1)
@@ -28,7 +28,7 @@ T* test3(B* x) { return &dynamic_cast<T&>(*x); }
// CHECK-LABEL: define dso_local noundef ptr @"?test3@@YAPAUT@@PAUB@@@Z"(ptr noundef %x)
// CHECK: [[VBPTR:%.*]] = getelementptr inbounds i8, ptr %x, i32 4
// CHECK-NEXT: [[VBTBL:%.*]] = load ptr, ptr [[VBPTR:%.*]], align 4
-// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i32, ptr [[VBTBL]], i32 1
+// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i8, ptr [[VBTBL]], i32 4
// CHECK-NEXT: [[VBOFFS:%.*]] = load i32, ptr [[VBOFFP]], align 4
// CHECK-NEXT: [[DELTA:%.*]] = add nsw i32 [[VBOFFS]], 4
// CHECK-NEXT: [[ADJ:%.*]] = getelementptr inbounds i8, ptr %x, i32 [[DELTA]]
@@ -45,7 +45,7 @@ T* test5(A* x) { return dynamic_cast<T*>(x); }
// CHECK: [[CHECK:%.*]] = icmp eq ptr %x, null
// CHECK-NEXT: br i1 [[CHECK]]
// CHECK: [[VBTBL:%.*]] = load ptr, ptr %x, align 4
-// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i32, ptr [[VBTBL]], i32 1
+// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i8, ptr [[VBTBL]], i32 4
// CHECK-NEXT: [[VBOFFS:%.*]] = load i32, ptr [[VBOFFP]], align 4
// CHECK-NEXT: [[ADJ:%.*]] = getelementptr inbounds i8, ptr %x, i32 [[VBOFFS]]
// CHECK-NEXT: [[CALL:%.*]] = tail call ptr @__RTDynamicCast(ptr nonnull [[ADJ]], i32 [[VBOFFS]], ptr {{.*}}@"??_R0?AUA@@@8", ptr {{.*}}@"??_R0?AUT@@@8", i32 0)
@@ -59,7 +59,7 @@ T* test6(B* x) { return dynamic_cast<T*>(x); }
// CHECK-NEXT: br i1 [[CHECK]]
// CHECK: [[VBPTR:%.*]] = getelementptr inbounds i8, ptr %x, i32 4
// CHECK-NEXT: [[VBTBL:%.*]] = load ptr, ptr [[VBPTR]], align 4
-// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i32, ptr [[VBTBL]], i32 1
+// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i8, ptr [[VBTBL]], i32 4
// CHECK-NEXT: [[VBOFFS:%.*]] = load i32, ptr [[VBOFFP]], align 4
// CHECK-NEXT: [[DELTA:%.*]] = add nsw i32 [[VBOFFS]], 4
// CHECK-NEXT: [[ADJ:%.*]] = getelementptr inbounds i8, ptr %x, i32 [[DELTA]]
@@ -78,7 +78,7 @@ void* test8(A* x) { return dynamic_cast<void*>(x); }
// CHECK: [[CHECK:%.*]] = icmp eq ptr %x, null
// CHECK-NEXT: br i1 [[CHECK]]
// CHECK: [[VBTBL:%.*]] = load ptr, ptr %x, align 4
-// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i32, ptr [[VBTBL]], i32 1
+// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i8, ptr [[VBTBL]], i32 4
// CHECK-NEXT: [[VBOFFS:%.*]] = load i32, ptr [[VBOFFP]], align 4
// CHECK-NEXT: [[ADJ:%.*]] = getelementptr inbounds i8, ptr %x, i32 [[VBOFFS]]
// CHECK-NEXT: [[RES:%.*]] = tail call ptr @__RTCastToVoid(ptr nonnull [[ADJ]])
@@ -92,7 +92,7 @@ void* test9(B* x) { return dynamic_cast<void*>(x); }
// CHECK-NEXT: br i1 [[CHECK]]
// CHECK: [[VBPTR:%.*]] = getelementptr inbounds i8, ptr %x, i32 4
// CHECK-NEXT: [[VBTBL:%.*]] = load ptr, ptr [[VBPTR]], align 4
-// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i32, ptr [[VBTBL]], i32 1
+// CHECK-NEXT: [[VBOFFP:%.*]] = getelementptr inbounds i8, ptr [[VBTBL]], i32 4
// CHECK-NEXT: [[VBOFFS:%.*]] = load i32, ptr [[VBOFFP]], align 4
// CHECK-NEXT: [[BASE:%.*]] = getelementptr i8, ptr %x, i32 [[VBOFFS]]
// CHECK-NEXT: [[ADJ:%.*]] = getelementptr i8, ptr [[BASE]], i32 4
diff --git a/clang/test/CodeGenCXX/microsoft-abi-typeid.cpp b/clang/test/CodeGenCXX/microsoft-abi-typeid.cpp
index 5b3c812c078467..31d7d23fd6fc5c 100644
--- a/clang/test/CodeGenCXX/microsoft-abi-typeid.cpp
+++ b/clang/test/CodeGenCXX/microsoft-abi-typeid.cpp
@@ -31,7 +31,7 @@ const std::type_info* test3_typeid() { return &typeid(*fn()); }
// CHECK: call ptr @__RTtypeid(ptr null)
// CHECK-NEXT: unreachable
// CHECK: [[VBTBL:%.*]] = load ptr, ptr [[CALL]], align 4
-// CHECK-NEXT: [[VBSLOT:%.*]] = getelementptr inbounds i32, ptr [[VBTBL]], i32 1
+// CHECK-NEXT: [[VBSLOT:%.*]] = getelementptr inbounds i8, ptr [[VBTBL]], i32 4
// CHECK-NEXT: [[VBASE_OFFS:%.*]] = load i32, ptr [[VBSLOT]], align 4
// CHECK-NEXT: [[ADJ:%.*]] = getelementptr inbounds i8, ptr [[CALL]], i32 [[VBASE_OFFS]]
// CHECK-NEXT: [[RT:%.*]] = tail call ptr @__RTtypeid(ptr nonnull [[ADJ]])
diff --git a/clang/test/CodeGenObjC/arc-foreach.m b/clang/test/CodeGenObjC/arc-foreach.m
index 64099e04d6d03d..71edc5161303c6 100644
--- a/clang/test/CodeGenObjC/arc-foreach.m
+++ b/clang/test/CodeGenObjC/arc-foreach.m
@@ -65,7 +65,7 @@ void test0(NSArray *array) {
// CHECK-LP64-NEXT: call void @llvm.objc.storeStrong(ptr [[CAPTURED]], ptr null)
// CHECK-LP64-NOT: call void (...) @llvm.objc.clang.arc.use(
-// CHECK-LP64-OPT: [[D0:%.*]] = getelementptr inbounds [[BLOCK_T]], ptr [[BLOCK]], i64 0, i32 5
+// CHECK-LP64-OPT: [[D0:%.*]] = getelementptr inbounds i8, ptr [[BLOCK]], i64 32
// CHECK-LP64-OPT: [[CAPTURE:%.*]] = load ptr, ptr [[D0]]
// CHECK-LP64-OPT: call void (...) @llvm.objc.clang.arc.use(ptr [[CAPTURE]])
@@ -189,7 +189,7 @@ - (void) foo2 {
// CHECK-LP64-OPT-LABEL: define internal void @"\01-[I1 foo2]"(
// CHECK-LP64-OPT: ptr %self
// CHECK-LP64-OPT: [[BLOCK:%.*]] = alloca [[BLOCK_T:<{.*}>]],
-// CHECK-LP64-OPT: [[T0:%.*]] = getelementptr inbounds [[BLOCK_T]], ptr [[BLOCK]], i64 0, i32 5
+// CHECK-LP64-OPT: [[T0:%.*]] = getelementptr inbounds i8, ptr [[BLOCK]], i64 32
// CHECK-LP64: call void @llvm.objc.storeStrong(ptr [[BC]], ptr null)
// CHECK-LP64-NOT: call void (...) @llvm.objc.clang.arc.use(ptr [[BC]])
diff --git a/clang/test/CodeGenObjCXX/arc-cxx11-init-list.mm b/clang/test/CodeGenObjCXX/arc-cxx11-init-list.mm
index c2d4f8685d7efb..fcf47c7faa9ce4 100644
--- a/clang/test/CodeGenObjCXX/arc-cxx11-init-list.mm
+++ b/clang/test/CodeGenObjCXX/arc-cxx11-init-list.mm
@@ -42,7 +42,7 @@ + (instancetype) new;
// CHECK: define{{.*}} void @_Z4foo1v(ptr {{.*}} %[[AGG_RESULT:.*]])
// CHECK: store ptr @[[REFTMP]], ptr %[[AGG_RESULT]]
-// CHECK: %[[SIZE:.*]] = getelementptr inbounds %"class.std::initializer_list.0", ptr %[[AGG_RESULT]], i32 0, i32 1
+// CHECK: %[[SIZE:.*]] = getelementptr inbounds i8, ptr %[[AGG_RESULT]], i32 4
// CHECK: store i32 2, ptr %[[SIZE]]
// CHECK: ret void
diff --git a/clang/test/Headers/__clang_hip_math.hip b/clang/test/Headers/__clang_hip_math.hip
index fa13ddba425bde..5230c360d0bef5 100644
--- a/clang/test/Headers/__clang_hip_math.hip
+++ b/clang/test/Headers/__clang_hip_math.hip
@@ -2829,7 +2829,7 @@ extern "C" __device__ double test_normcdfinv(double x) {
// DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// DEFAULT-NEXT: [[MUL_I:%.*]] = fmul contract float [[TMP0]], [[TMP0]]
// DEFAULT-NEXT: [[ADD_I]] = fadd contract float [[__R_0_I4]], [[MUL_I]]
-// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds float, ptr [[__A_ADDR_0_I3]], i64 1
+// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 4
// DEFAULT-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5NORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP20:![0-9]+]]
// DEFAULT: _ZL5normfiPKf.exit:
@@ -2849,7 +2849,7 @@ extern "C" __device__ double test_normcdfinv(double x) {
// FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// FINITEONLY-NEXT: [[MUL_I:%.*]] = fmul nnan ninf contract float [[TMP0]], [[TMP0]]
// FINITEONLY-NEXT: [[ADD_I]] = fadd nnan ninf contract float [[__R_0_I4]], [[MUL_I]]
-// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds float, ptr [[__A_ADDR_0_I3]], i64 1
+// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 4
// FINITEONLY-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5NORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP20:![0-9]+]]
// FINITEONLY: _ZL5normfiPKf.exit:
@@ -2869,7 +2869,7 @@ extern "C" __device__ double test_normcdfinv(double x) {
// APPROX-NEXT: [[TMP0:%.*]] = load float, ptr [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// APPROX-NEXT: [[MUL_I:%.*]] = fmul contract float [[TMP0]], [[TMP0]]
// APPROX-NEXT: [[ADD_I]] = fadd contract float [[__R_0_I4]], [[MUL_I]]
-// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds float, ptr [[__A_ADDR_0_I3]], i64 1
+// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 4
// APPROX-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5NORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP20:![0-9]+]]
// APPROX: _ZL5normfiPKf.exit:
@@ -2893,7 +2893,7 @@ extern "C" __device__ float test_normf(int x, const float *y) {
// DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// DEFAULT-NEXT: [[MUL_I:%.*]] = fmul contract double [[TMP0]], [[TMP0]]
// DEFAULT-NEXT: [[ADD_I]] = fadd contract double [[__R_0_I4]], [[MUL_I]]
-// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds double, ptr [[__A_ADDR_0_I3]], i64 1
+// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 8
// DEFAULT-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL4NORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP21:![0-9]+]]
// DEFAULT: _ZL4normiPKd.exit:
@@ -2913,7 +2913,7 @@ extern "C" __device__ float test_normf(int x, const float *y) {
// FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// FINITEONLY-NEXT: [[MUL_I:%.*]] = fmul nnan ninf contract double [[TMP0]], [[TMP0]]
// FINITEONLY-NEXT: [[ADD_I]] = fadd nnan ninf contract double [[__R_0_I4]], [[MUL_I]]
-// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds double, ptr [[__A_ADDR_0_I3]], i64 1
+// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 8
// FINITEONLY-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL4NORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP21:![0-9]+]]
// FINITEONLY: _ZL4normiPKd.exit:
@@ -2933,7 +2933,7 @@ extern "C" __device__ float test_normf(int x, const float *y) {
// APPROX-NEXT: [[TMP0:%.*]] = load double, ptr [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// APPROX-NEXT: [[MUL_I:%.*]] = fmul contract double [[TMP0]], [[TMP0]]
// APPROX-NEXT: [[ADD_I]] = fadd contract double [[__R_0_I4]], [[MUL_I]]
-// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds double, ptr [[__A_ADDR_0_I3]], i64 1
+// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 8
// APPROX-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL4NORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP21:![0-9]+]]
// APPROX: _ZL4normiPKd.exit:
@@ -3253,7 +3253,7 @@ extern "C" __device__ double test_rint(double x) {
// DEFAULT-NEXT: [[TMP0:%.*]] = load float, ptr [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// DEFAULT-NEXT: [[MUL_I:%.*]] = fmul contract float [[TMP0]], [[TMP0]]
// DEFAULT-NEXT: [[ADD_I]] = fadd contract float [[__R_0_I4]], [[MUL_I]]
-// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds float, ptr [[__A_ADDR_0_I3]], i64 1
+// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 4
// DEFAULT-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
// DEFAULT: _ZL6rnormfiPKf.exit:
@@ -3273,7 +3273,7 @@ extern "C" __device__ double test_rint(double x) {
// FINITEONLY-NEXT: [[TMP0:%.*]] = load float, ptr [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// FINITEONLY-NEXT: [[MUL_I:%.*]] = fmul nnan ninf contract float [[TMP0]], [[TMP0]]
// FINITEONLY-NEXT: [[ADD_I]] = fadd nnan ninf contract float [[__R_0_I4]], [[MUL_I]]
-// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds float, ptr [[__A_ADDR_0_I3]], i64 1
+// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 4
// FINITEONLY-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
// FINITEONLY: _ZL6rnormfiPKf.exit:
@@ -3293,7 +3293,7 @@ extern "C" __device__ double test_rint(double x) {
// APPROX-NEXT: [[TMP0:%.*]] = load float, ptr [[__A_ADDR_0_I3]], align 4, !tbaa [[TBAA16]]
// APPROX-NEXT: [[MUL_I:%.*]] = fmul contract float [[TMP0]], [[TMP0]]
// APPROX-NEXT: [[ADD_I]] = fadd contract float [[__R_0_I4]], [[MUL_I]]
-// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds float, ptr [[__A_ADDR_0_I3]], i64 1
+// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 4
// APPROX-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL6RNORMFIPKF_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP22:![0-9]+]]
// APPROX: _ZL6rnormfiPKf.exit:
@@ -3317,7 +3317,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// DEFAULT-NEXT: [[TMP0:%.*]] = load double, ptr [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// DEFAULT-NEXT: [[MUL_I:%.*]] = fmul contract double [[TMP0]], [[TMP0]]
// DEFAULT-NEXT: [[ADD_I]] = fadd contract double [[__R_0_I4]], [[MUL_I]]
-// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds double, ptr [[__A_ADDR_0_I3]], i64 1
+// DEFAULT-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 8
// DEFAULT-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// DEFAULT-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
// DEFAULT: _ZL5rnormiPKd.exit:
@@ -3337,7 +3337,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// FINITEONLY-NEXT: [[TMP0:%.*]] = load double, ptr [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// FINITEONLY-NEXT: [[MUL_I:%.*]] = fmul nnan ninf contract double [[TMP0]], [[TMP0]]
// FINITEONLY-NEXT: [[ADD_I]] = fadd nnan ninf contract double [[__R_0_I4]], [[MUL_I]]
-// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds double, ptr [[__A_ADDR_0_I3]], i64 1
+// FINITEONLY-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 8
// FINITEONLY-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// FINITEONLY-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
// FINITEONLY: _ZL5rnormiPKd.exit:
@@ -3357,7 +3357,7 @@ extern "C" __device__ float test_rnormf(int x, const float* y) {
// APPROX-NEXT: [[TMP0:%.*]] = load double, ptr [[__A_ADDR_0_I3]], align 8, !tbaa [[TBAA18]]
// APPROX-NEXT: [[MUL_I:%.*]] = fmul contract double [[TMP0]], [[TMP0]]
// APPROX-NEXT: [[ADD_I]] = fadd contract double [[__R_0_I4]], [[MUL_I]]
-// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds double, ptr [[__A_ADDR_0_I3]], i64 1
+// APPROX-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__A_ADDR_0_I3]], i64 8
// APPROX-NEXT: [[TOBOOL_NOT_I:%.*]] = icmp eq i32 [[DEC_I]], 0
// APPROX-NEXT: br i1 [[TOBOOL_NOT_I]], label [[_ZL5RNORMIPKD_EXIT]], label [[WHILE_BODY_I]], !llvm.loop [[LOOP23:![0-9]+]]
// APPROX: _ZL5rnormiPKd.exit:
diff --git a/clang/test/OpenMP/bug57757.cpp b/clang/test/OpenMP/bug57757.cpp
index 4aa5c2639c51aa..e1f646e2b141a0 100644
--- a/clang/test/OpenMP/bug57757.cpp
+++ b/clang/test/OpenMP/bug57757.cpp
@@ -19,9 +19,9 @@ void foo() {
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = tail call i32 @__kmpc_global_thread_num(ptr nonnull @[[GLOB1:[0-9]+]])
// CHECK-NEXT: [[TMP1:%.*]] = tail call ptr @__kmpc_omp_task_alloc(ptr nonnull @[[GLOB1]], i32 [[TMP0]], i32 0, i64 56, i64 1, ptr nonnull @.omp_task_entry.)
-// CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP1]], i64 0, i32 1
+// CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
// CHECK-NEXT: store ptr @_Z3barif, ptr [[TMP2]], align 8, !tbaa [[TBAA3:![0-9]+]]
-// CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP1]], i64 0, i32 2
+// CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
// CHECK-NEXT: store i32 0, ptr [[TMP3]], align 8, !tbaa [[TBAA12:![0-9]+]]
// CHECK-NEXT: [[TMP4:%.*]] = tail call i32 @__kmpc_omp_task(ptr nonnull @[[GLOB1]], i32 [[TMP0]], ptr [[TMP1]])
// CHECK-NEXT: ret void
@@ -30,7 +30,7 @@ void foo() {
// CHECK-LABEL: define {{[^@]+}}@.omp_task_entry.
// CHECK-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK-NEXT: entry:
-// CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP1]], i64 0, i32 2
+// CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
// CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META13:![0-9]+]])
// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4, !tbaa [[TBAA16:![0-9]+]], !alias.scope [[META13]], !noalias [[META17:![0-9]+]]
// CHECK-NEXT: switch i32 [[TMP3]], label [[DOTOMP_OUTLINED__EXIT:%.*]] [
@@ -42,9 +42,9 @@ void foo() {
// CHECK-NEXT: [[TMP4:%.*]] = tail call i32 @__kmpc_omp_task(ptr nonnull @[[GLOB1]], i32 [[TMP0]], ptr [[TMP1]]), !noalias [[META13]]
// CHECK-NEXT: br label [[DOTOMP_OUTLINED__EXIT]]
// CHECK: .untied.next..i:
-// CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP1]], i64 0, i32 1
-// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP1]], i64 0, i32 1, i32 2
-// CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP1]], i64 0, i32 1, i32 1
+// CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
+// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 52
+// CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 48
// CHECK-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP5]], align 8, !tbaa [[TBAA19:![0-9]+]], !noalias [[META13]]
// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA16]], !noalias [[META13]]
// CHECK-NEXT: [[TMP10:%.*]] = load float, ptr [[TMP6]], align 4, !tbaa [[TBAA20:![0-9]+]], !noalias [[META13]]
diff --git a/flang/test/HLFIR/no-block-merging.fir b/flang/test/HLFIR/no-block-merging.fir
index 987cf8bf01e8a9..75803bf0108847 100644
--- a/flang/test/HLFIR/no-block-merging.fir
+++ b/flang/test/HLFIR/no-block-merging.fir
@@ -27,7 +27,7 @@ func.func @no_shape_merge(%cdt: i1, %from: !fir.ref<!fir.array<?xf64>>, %to : !f
// Note: block merging happens in the output below, but after FIR codegen.
// CHECK-LABEL: define void @no_shape_merge(
-// CHECK: %[[GEP:.*]] = getelementptr double, ptr %{{.*}}
+// CHECK: %[[GEP:.*]] = getelementptr i8, ptr %{{.*}}
// CHECK: %[[LOAD:.*]] = load double, ptr %[[GEP]]
// CHECK: store double %[[LOAD]], ptr %{{.*}}
// CHECK: ret void
diff --git a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
index 351fc3b0174fc7..c83b9d2d4505ef 100644
--- a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
@@ -2352,6 +2352,15 @@ Instruction *InstCombinerImpl::visitGetElementPtrInst(GetElementPtrInst &GEP) {
if (MadeChange)
return &GEP;
+ // Canonicalize constant GEPs to i8 type.
+ if (!GEPEltType->isIntegerTy(8) && GEP.hasAllConstantIndices()) {
+ APInt Offset(DL.getIndexTypeSizeInBits(GEPType), 0);
+ if (GEP.accumulateConstantOffset(DL, Offset))
+ return replaceInstUsesWith(
+ GEP, Builder.CreateGEP(Builder.getInt8Ty(), PtrOp,
+ Builder.getInt(Offset), "", GEP.isInBounds()));
+ }
+
// Check to see if the inputs to the PHI node are getelementptr instructions.
if (auto *PN = dyn_cast<PHINode>(PtrOp)) {
auto *Op1 = dyn_cast<GetElementPtrInst>(PN->getOperand(0));
diff --git a/llvm/test/Analysis/BasicAA/featuretest.ll b/llvm/test/Analysis/BasicAA/featuretest.ll
index f556c95747a19c..5ef94cd8c7fdaf 100644
--- a/llvm/test/Analysis/BasicAA/featuretest.ll
+++ b/llvm/test/Analysis/BasicAA/featuretest.ll
@@ -59,7 +59,7 @@ define i32 @constant_array_index_test() {
; CHECK-LABEL: @constant_array_index_test(
; CHECK-NEXT: [[ARRAY1:%.*]] = alloca [100 x i32], align 4
; CHECK-NEXT: call void @external(ptr nonnull [[ARRAY1]])
-; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds i32, ptr [[ARRAY1]], i64 6
+; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds i8, ptr [[ARRAY1]], i64 24
; CHECK-NEXT: store i32 1, ptr [[P2]], align 4
; CHECK-NEXT: ret i32 0
;
@@ -80,12 +80,12 @@ define i32 @constant_array_index_test() {
; they cannot alias.
define i32 @gep_distance_test(ptr %A) {
; NO_ASSUME-LABEL: @gep_distance_test(
-; NO_ASSUME-NEXT: [[B:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 2
+; NO_ASSUME-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 8
; NO_ASSUME-NEXT: store i32 7, ptr [[B]], align 4
; NO_ASSUME-NEXT: ret i32 0
;
; USE_ASSUME-LABEL: @gep_distance_test(
-; USE_ASSUME-NEXT: [[B:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 2
+; USE_ASSUME-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 8
; USE_ASSUME-NEXT: store i32 7, ptr [[B]], align 4
; USE_ASSUME-NEXT: call void @llvm.assume(i1 true) [ "dereferenceable"(ptr [[A]], i64 4), "nonnull"(ptr [[A]]), "align"(ptr [[A]], i64 4) ]
; USE_ASSUME-NEXT: ret i32 0
diff --git a/llvm/test/CodeGen/AMDGPU/vector-alloca-bitcast.ll b/llvm/test/CodeGen/AMDGPU/vector-alloca-bitcast.ll
index f779c9a468f479..769db215cf0349 100644
--- a/llvm/test/CodeGen/AMDGPU/vector-alloca-bitcast.ll
+++ b/llvm/test/CodeGen/AMDGPU/vector-alloca-bitcast.ll
@@ -369,11 +369,11 @@ define amdgpu_kernel void @vector_bitcast_from_alloca_array(ptr addrspace(1) %ou
; OPT-LABEL: @vector_bitcast_to_array_from_alloca_array(
; OPT-NOT: alloca
; OPT-NEXT: store i32 1, ptr addrspace(1) %out, align 4
-; OPT-NEXT: %out.repack1 = getelementptr inbounds [4 x i32], ptr addrspace(1) %out, i64 0, i64 1
+; OPT-NEXT: %out.repack1 = getelementptr inbounds i8, ptr addrspace(1) %out, i64 4
; OPT-NEXT: store i32 2, ptr addrspace(1) %out.repack1, align 4
-; OPT-NEXT: %out.repack2 = getelementptr inbounds [4 x i32], ptr addrspace(1) %out, i64 0, i64 2
+; OPT-NEXT: %out.repack2 = getelementptr inbounds i8, ptr addrspace(1) %out, i64 8
; OPT-NEXT: store i32 3, ptr addrspace(1) %out.repack2, align 4
-; OPT-NEXT: %out.repack3 = getelementptr inbounds [4 x i32], ptr addrspace(1) %out, i64 0, i64 3
+; OPT-NEXT: %out.repack3 = getelementptr inbounds i8, ptr addrspace(1) %out, i64 12
; OPT-NEXT: store i32 4, ptr addrspace(1) %out.repack3, align 4
; GCN-LABEL: {{^}}vector_bitcast_to_array_from_alloca_array:
@@ -395,11 +395,11 @@ define amdgpu_kernel void @vector_bitcast_to_array_from_alloca_array(ptr addrspa
; OPT-LABEL: @vector_bitcast_to_struct_from_alloca_array(
; OPT-NOT: alloca
; OPT-NEXT: store i32 1, ptr addrspace(1) %out, align 4
-; OPT-NEXT: %out.repack1 = getelementptr inbounds %struct.v4, ptr addrspace(1) %out, i64 0, i32 1
+; OPT-NEXT: %out.repack1 = getelementptr inbounds i8, ptr addrspace(1) %out, i64 4
; OPT-NEXT: store i32 2, ptr addrspace(1) %out.repack1, align 4
-; OPT-NEXT: %out.repack2 = getelementptr inbounds %struct.v4, ptr addrspace(1) %out, i64 0, i32 2
+; OPT-NEXT: %out.repack2 = getelementptr inbounds i8, ptr addrspace(1) %out, i64 8
; OPT-NEXT: store i32 3, ptr addrspace(1) %out.repack2, align 4
-; OPT-NEXT: %out.repack3 = getelementptr inbounds %struct.v4, ptr addrspace(1) %out, i64 0, i32 3
+; OPT-NEXT: %out.repack3 = getelementptr inbounds i8, ptr addrspace(1) %out, i64 12
; OPT-NEXT: store i32 4, ptr addrspace(1) %out.repack3, align 4
; GCN-LABEL: {{^}}vector_bitcast_to_struct_from_alloca_array:
diff --git a/llvm/test/CodeGen/BPF/preserve-static-offset/load-inline.ll b/llvm/test/CodeGen/BPF/preserve-static-offset/load-inline.ll
index 9149b350dd8956..b43c73035b332f 100644
--- a/llvm/test/CodeGen/BPF/preserve-static-offset/load-inline.ll
+++ b/llvm/test/CodeGen/BPF/preserve-static-offset/load-inline.ll
@@ -57,8 +57,8 @@ entry:
; CHECK: define dso_local void @quux(ptr nocapture noundef readonly %[[p:.*]])
; CHECK: %[[bb_i1:.*]] = tail call i32 (ptr, i1, i8, i8, i8, i1, ...)
; CHECK-SAME: @llvm.bpf.getelementptr.and.load.i32
-; CHECK-SAME: (ptr readonly elementtype(%struct.foo) %[[p]],
-; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 0, i32 immarg 1, i32 immarg 1)
+; CHECK-SAME: (ptr readonly elementtype(i8) %[[p]],
+; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 8)
; CHECK-SAME: #[[v2:.*]], !tbaa
; CHECK-NEXT: tail call void @consume(i32 noundef %[[bb_i1]])
; CHECK: attributes #[[v2]] = { memory(argmem: read) }
diff --git a/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll-inline.ll b/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll-inline.ll
index 78172cd17dca48..1f0be06755b442 100644
--- a/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll-inline.ll
+++ b/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll-inline.ll
@@ -75,8 +75,8 @@ entry:
; CHECK: define dso_local void @quux(ptr nocapture noundef readonly %[[p:.*]])
; CHECK: %[[v1:.*]] = tail call i32 (ptr, i1, i8, i8, i8, i1, ...)
; CHECK-SAME: @llvm.bpf.getelementptr.and.load.i32
-; CHECK-SAME: (ptr readonly elementtype(%struct.foo) %[[p]],
-; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 0, i32 immarg 1, i64 immarg 1)
+; CHECK-SAME: (ptr readonly elementtype(i8) %[[p]],
+; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 8)
; CHECK: tail call void @consume(i32 noundef %[[v1]])
; CHECK: tail call void @consume(i32 noundef %[[v1]])
diff --git a/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll.ll b/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll.ll
index 7c3303342bb6db..2409fbc8b7e851 100644
--- a/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll.ll
+++ b/llvm/test/CodeGen/BPF/preserve-static-offset/load-unroll.ll
@@ -53,14 +53,14 @@ while.end: ; preds = %while.cond
; CHECK: define dso_local void @bar(ptr nocapture noundef readonly %[[p:.*]])
; CHECK: %[[v1:.*]] = tail call i32 (ptr, i1, i8, i8, i8, i1, ...)
; CHECK-SAME: @llvm.bpf.getelementptr.and.load.i32
-; CHECK-SAME: (ptr readonly elementtype(%struct.foo) %[[p]],
-; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 0, i32 immarg 1, i64 immarg 0)
+; CHECK-SAME: (ptr readonly elementtype(i8) %[[p]],
+; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 4)
; CHECK-SAME: #[[attrs:.*]], !tbaa
; CHECK-NEXT: tail call void @consume(i32 noundef %[[v1]])
; CHECK-NEXT: %[[v2:.*]] = tail call i32 (ptr, i1, i8, i8, i8, i1, ...)
; CHECK-SAME: @llvm.bpf.getelementptr.and.load.i32
-; CHECK-SAME: (ptr readonly elementtype(%struct.foo) %[[p]],
-; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 0, i32 immarg 1, i64 immarg 1)
+; CHECK-SAME: (ptr readonly elementtype(i8) %[[p]],
+; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 8)
; CHECK-SAME: #[[attrs]], !tbaa
; CHECK-NEXT: tail call void @consume(i32 noundef %[[v2]])
; CHECK: attributes #[[attrs]] = { memory(argmem: read) }
diff --git a/llvm/test/CodeGen/BPF/preserve-static-offset/store-unroll-inline.ll b/llvm/test/CodeGen/BPF/preserve-static-offset/store-unroll-inline.ll
index 161aded79eac3c..3c6280ca75b64d 100644
--- a/llvm/test/CodeGen/BPF/preserve-static-offset/store-unroll-inline.ll
+++ b/llvm/test/CodeGen/BPF/preserve-static-offset/store-unroll-inline.ll
@@ -75,8 +75,8 @@ entry:
; CHECK-NEXT: tail call void (i32, ptr, i1, i8, i8, i8, i1, ...)
; CHECK-SAME: @llvm.bpf.getelementptr.and.store.i32
; CHECK-SAME: (i32 1,
-; CHECK-SAME: ptr writeonly elementtype(%struct.foo) %[[p]],
-; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 0, i32 immarg 1)
+; CHECK-SAME: ptr writeonly elementtype(i8) %[[p]],
+; CHECK-SAME: i1 false, i8 0, i8 1, i8 2, i1 true, i64 immarg 4)
; CHECK-NEXT: ret void
; Function Attrs: nocallback nofree nosync nounwind speculatable willreturn memory(none)
diff --git a/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll b/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll
index 24c5e1c2f789a5..73279844e91b33 100644
--- a/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll
+++ b/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll
@@ -13,12 +13,12 @@ define <64 x i16> @f0(ptr %a0, i32 %a1) #0 {
; CHECK-LABEL: @f0(
; CHECK-NEXT: b0:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i16, ptr [[A0:%.*]], i32 [[A1:%.*]]
-; CHECK-NEXT: [[V1:%.*]] = getelementptr i16, ptr [[TMP0]], i32 64
+; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
; CHECK-NEXT: [[ITP:%.*]] = inttoptr i32 [[AND]] to ptr
; CHECK-NEXT: [[PTI1:%.*]] = ptrtoint ptr [[V1]] to i32
-; CHECK-NEXT: [[ALD15:%.*]] = load <32 x i32>, ptr [[ITP]], align 128, !tbaa [[TBAA0:![0-9]+]]
+; CHECK-NEXT: [[ALD14:%.*]] = load <32 x i32>, ptr [[ITP]], align 128, !tbaa [[TBAA0:![0-9]+]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[ITP]], i32 128
; CHECK-NEXT: [[ALD2:%.*]] = load <128 x i8>, ptr [[GEP]], align 128, !tbaa [[TBAA0]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[ITP]], i32 256
@@ -26,12 +26,12 @@ define <64 x i16> @f0(ptr %a0, i32 %a1) #0 {
; CHECK-NEXT: [[ISZ:%.*]] = icmp ne i32 [[AND4]], 0
; CHECK-NEXT: [[CUP:%.*]] = call <32 x i32> @llvm.hexagon.V6.vL32b.pred.ai.128B(i1 [[ISZ]], ptr [[GEP3]], i32 0), !tbaa [[TBAA0]]
; CHECK-NEXT: [[CST5:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
-; CHECK-NEXT: [[CUP7:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CST5]], <32 x i32> [[ALD15]], i32 [[PTI1]])
-; CHECK-NEXT: [[CST13:%.*]] = bitcast <32 x i32> [[CUP7]] to <64 x i16>
-; CHECK-NEXT: [[CST10:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
-; CHECK-NEXT: [[CUP11:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CUP]], <32 x i32> [[CST10]], i32 [[PTI1]])
-; CHECK-NEXT: [[CST14:%.*]] = bitcast <32 x i32> [[CUP11]] to <64 x i16>
-; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST13]], [[CST14]]
+; CHECK-NEXT: [[CUP7:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CST5]], <32 x i32> [[ALD14]], i32 [[PTI1]])
+; CHECK-NEXT: [[CST12:%.*]] = bitcast <32 x i32> [[CUP7]] to <64 x i16>
+; CHECK-NEXT: [[CST9:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
+; CHECK-NEXT: [[CUP10:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CUP]], <32 x i32> [[CST9]], i32 [[PTI1]])
+; CHECK-NEXT: [[CST13:%.*]] = bitcast <32 x i32> [[CUP10]] to <64 x i16>
+; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST12]], [[CST13]]
; CHECK-NEXT: ret <64 x i16> [[V8]]
;
b0:
@@ -51,12 +51,12 @@ define <64 x i16> @f1(ptr %a0, i32 %a1) #0 {
; CHECK-LABEL: @f1(
; CHECK-NEXT: b0:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i16, ptr [[A0:%.*]], i32 [[A1:%.*]]
-; CHECK-NEXT: [[V1:%.*]] = getelementptr i16, ptr [[TMP0]], i32 64
+; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
; CHECK-NEXT: [[ITP:%.*]] = inttoptr i32 [[AND]] to ptr
; CHECK-NEXT: [[PTI1:%.*]] = ptrtoint ptr [[V1]] to i32
-; CHECK-NEXT: [[ALD15:%.*]] = load <32 x i32>, ptr [[ITP]], align 128, !tbaa [[TBAA0]]
+; CHECK-NEXT: [[ALD14:%.*]] = load <32 x i32>, ptr [[ITP]], align 128, !tbaa [[TBAA0]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[ITP]], i32 128
; CHECK-NEXT: [[ALD2:%.*]] = load <128 x i8>, ptr [[GEP]], align 128
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[ITP]], i32 256
@@ -64,12 +64,12 @@ define <64 x i16> @f1(ptr %a0, i32 %a1) #0 {
; CHECK-NEXT: [[ISZ:%.*]] = icmp ne i32 [[AND4]], 0
; CHECK-NEXT: [[CUP:%.*]] = call <32 x i32> @llvm.hexagon.V6.vL32b.pred.ai.128B(i1 [[ISZ]], ptr [[GEP3]], i32 0)
; CHECK-NEXT: [[CST5:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
-; CHECK-NEXT: [[CUP7:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CST5]], <32 x i32> [[ALD15]], i32 [[PTI1]])
-; CHECK-NEXT: [[CST13:%.*]] = bitcast <32 x i32> [[CUP7]] to <64 x i16>
-; CHECK-NEXT: [[CST10:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
-; CHECK-NEXT: [[CUP11:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CUP]], <32 x i32> [[CST10]], i32 [[PTI1]])
-; CHECK-NEXT: [[CST14:%.*]] = bitcast <32 x i32> [[CUP11]] to <64 x i16>
-; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST13]], [[CST14]]
+; CHECK-NEXT: [[CUP7:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CST5]], <32 x i32> [[ALD14]], i32 [[PTI1]])
+; CHECK-NEXT: [[CST12:%.*]] = bitcast <32 x i32> [[CUP7]] to <64 x i16>
+; CHECK-NEXT: [[CST9:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
+; CHECK-NEXT: [[CUP10:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CUP]], <32 x i32> [[CST9]], i32 [[PTI1]])
+; CHECK-NEXT: [[CST13:%.*]] = bitcast <32 x i32> [[CUP10]] to <64 x i16>
+; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST12]], [[CST13]]
; CHECK-NEXT: ret <64 x i16> [[V8]]
;
b0:
@@ -89,12 +89,12 @@ define <64 x i16> @f2(ptr %a0, i32 %a1) #0 {
; CHECK-LABEL: @f2(
; CHECK-NEXT: b0:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i16, ptr [[A0:%.*]], i32 [[A1:%.*]]
-; CHECK-NEXT: [[V1:%.*]] = getelementptr i16, ptr [[TMP0]], i32 64
+; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
; CHECK-NEXT: [[ITP:%.*]] = inttoptr i32 [[AND]] to ptr
; CHECK-NEXT: [[PTI1:%.*]] = ptrtoint ptr [[V1]] to i32
-; CHECK-NEXT: [[ALD15:%.*]] = load <32 x i32>, ptr [[ITP]], align 128, !tbaa [[TBAA0]]
+; CHECK-NEXT: [[ALD14:%.*]] = load <32 x i32>, ptr [[ITP]], align 128, !tbaa [[TBAA0]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[ITP]], i32 128
; CHECK-NEXT: [[ALD2:%.*]] = load <128 x i8>, ptr [[GEP]], align 128
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[ITP]], i32 256
@@ -102,12 +102,12 @@ define <64 x i16> @f2(ptr %a0, i32 %a1) #0 {
; CHECK-NEXT: [[ISZ:%.*]] = icmp ne i32 [[AND4]], 0
; CHECK-NEXT: [[CUP:%.*]] = call <32 x i32> @llvm.hexagon.V6.vL32b.pred.ai.128B(i1 [[ISZ]], ptr [[GEP3]], i32 0), !tbaa [[TBAA3:![0-9]+]]
; CHECK-NEXT: [[CST5:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
-; CHECK-NEXT: [[CUP7:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CST5]], <32 x i32> [[ALD15]], i32 [[PTI1]])
-; CHECK-NEXT: [[CST13:%.*]] = bitcast <32 x i32> [[CUP7]] to <64 x i16>
-; CHECK-NEXT: [[CST10:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
-; CHECK-NEXT: [[CUP11:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CUP]], <32 x i32> [[CST10]], i32 [[PTI1]])
-; CHECK-NEXT: [[CST14:%.*]] = bitcast <32 x i32> [[CUP11]] to <64 x i16>
-; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST13]], [[CST14]]
+; CHECK-NEXT: [[CUP7:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CST5]], <32 x i32> [[ALD14]], i32 [[PTI1]])
+; CHECK-NEXT: [[CST12:%.*]] = bitcast <32 x i32> [[CUP7]] to <64 x i16>
+; CHECK-NEXT: [[CST9:%.*]] = bitcast <128 x i8> [[ALD2]] to <32 x i32>
+; CHECK-NEXT: [[CUP10:%.*]] = call <32 x i32> @llvm.hexagon.V6.valignb.128B(<32 x i32> [[CUP]], <32 x i32> [[CST9]], i32 [[PTI1]])
+; CHECK-NEXT: [[CST13:%.*]] = bitcast <32 x i32> [[CUP10]] to <64 x i16>
+; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST12]], [[CST13]]
; CHECK-NEXT: ret <64 x i16> [[V8]]
;
b0:
@@ -127,7 +127,7 @@ define void @f3(ptr %a0, i32 %a1, <64 x i16> %a2, <64 x i16> %a3) #0 {
; CHECK-LABEL: @f3(
; CHECK-NEXT: b0:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i16, ptr [[A0:%.*]], i32 [[A1:%.*]]
-; CHECK-NEXT: [[V1:%.*]] = getelementptr i16, ptr [[TMP0]], i32 64
+; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
; CHECK-NEXT: [[ITP:%.*]] = inttoptr i32 [[AND]] to ptr
@@ -180,7 +180,7 @@ define void @f4(ptr %a0, i32 %a1, <64 x i16> %a2, <64 x i16> %a3) #0 {
; CHECK-LABEL: @f4(
; CHECK-NEXT: b0:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i16, ptr [[A0:%.*]], i32 [[A1:%.*]]
-; CHECK-NEXT: [[V1:%.*]] = getelementptr i16, ptr [[TMP0]], i32 64
+; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
; CHECK-NEXT: [[ITP:%.*]] = inttoptr i32 [[AND]] to ptr
@@ -233,7 +233,7 @@ define void @f5(ptr %a0, i32 %a1, <64 x i16> %a2, <64 x i16> %a3) #0 {
; CHECK-LABEL: @f5(
; CHECK-NEXT: b0:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i16, ptr [[A0:%.*]], i32 [[A1:%.*]]
-; CHECK-NEXT: [[V1:%.*]] = getelementptr i16, ptr [[TMP0]], i32 64
+; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
; CHECK-NEXT: [[ITP:%.*]] = inttoptr i32 [[AND]] to ptr
diff --git a/llvm/test/Transforms/Coroutines/coro-async.ll b/llvm/test/Transforms/Coroutines/coro-async.ll
index f796ce3dbb0d82..3740c3d1d83871 100644
--- a/llvm/test/Transforms/Coroutines/coro-async.ll
+++ b/llvm/test/Transforms/Coroutines/coro-async.ll
@@ -132,7 +132,7 @@ define void @my_async_function_pa(ptr %ctxt, ptr %task, ptr %actor) {
; CHECK: [[CALLEE_CTXT:%.*]] = tail call ptr @llvm.coro.async.context.alloc(ptr %task, ptr nonnull @my_other_async_function_fp)
; CHECK: [[CALLEE_CTXT_SPILL:%.*]] = getelementptr inbounds i8, ptr %async.ctxt, i64 160
; CHECK: store ptr [[CALLEE_CTXT]], ptr [[CALLEE_CTXT_SPILL]]
-; CHECK: [[TYPED_RETURN_TO_CALLER_ADDR:%.*]] = getelementptr inbounds %async.ctxt, ptr [[CALLEE_CTXT]], i64 0, i32 1
+; CHECK: [[TYPED_RETURN_TO_CALLER_ADDR:%.*]] = getelementptr inbounds i8, ptr [[CALLEE_CTXT]], i64 8
; CHECK: store ptr @my_async_functionTQ0_, ptr [[TYPED_RETURN_TO_CALLER_ADDR]]
; CHECK: store ptr %async.ctxt, ptr [[CALLEE_CTXT]]
; Make sure the spill is underaligned to the max context alignment (16).
diff --git a/llvm/test/Transforms/Coroutines/coro-retcon-alloca-opaque-ptr.ll b/llvm/test/Transforms/Coroutines/coro-retcon-alloca-opaque-ptr.ll
index 112ffd215b1be4..2dabc72a510148 100644
--- a/llvm/test/Transforms/Coroutines/coro-retcon-alloca-opaque-ptr.ll
+++ b/llvm/test/Transforms/Coroutines/coro-retcon-alloca-opaque-ptr.ll
@@ -8,7 +8,7 @@ declare {ptr, ptr, i32} @prototype_f(ptr, i1)
define {ptr, ptr, i32} @f(ptr %buffer, i32 %n, { i32 } %dummy) {
; CHECK-LABEL: @f(
; CHECK-NEXT: coro.return:
-; CHECK-NEXT: [[N_VAL_SPILL_ADDR:%.*]] = getelementptr inbounds [[F_FRAME:%.*]], ptr [[BUFFER:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[N_VAL_SPILL_ADDR:%.*]] = getelementptr inbounds i8, ptr [[BUFFER:%.*]], i64 8
; CHECK-NEXT: store i32 [[N:%.*]], ptr [[N_VAL_SPILL_ADDR]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = tail call ptr @allocate(i32 [[N]])
; CHECK-NEXT: store ptr [[TMP0]], ptr [[BUFFER]], align 8
diff --git a/llvm/test/Transforms/Coroutines/coro-retcon-alloca.ll b/llvm/test/Transforms/Coroutines/coro-retcon-alloca.ll
index fc009520fe4fa0..9b00d7929684fe 100644
--- a/llvm/test/Transforms/Coroutines/coro-retcon-alloca.ll
+++ b/llvm/test/Transforms/Coroutines/coro-retcon-alloca.ll
@@ -7,7 +7,7 @@ declare {ptr, ptr, i32} @prototype_f(ptr, i1)
define {ptr, ptr, i32} @f(ptr %buffer, i32 %n) {
; CHECK-LABEL: @f(
; CHECK-NEXT: coro.return:
-; CHECK-NEXT: [[N_VAL_SPILL_ADDR:%.*]] = getelementptr inbounds [[F_FRAME:%.*]], ptr [[BUFFER:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[N_VAL_SPILL_ADDR:%.*]] = getelementptr inbounds i8, ptr [[BUFFER:%.*]], i64 8
; CHECK-NEXT: store i32 [[N:%.*]], ptr [[N_VAL_SPILL_ADDR]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = tail call ptr @allocate(i32 [[N]])
; CHECK-NEXT: store ptr [[TMP0]], ptr [[BUFFER]], align 8
diff --git a/llvm/test/Transforms/Coroutines/coro-retcon-once-value.ll b/llvm/test/Transforms/Coroutines/coro-retcon-once-value.ll
index b67abfb051f350..2ee515f0a1a91e 100644
--- a/llvm/test/Transforms/Coroutines/coro-retcon-once-value.ll
+++ b/llvm/test/Transforms/Coroutines/coro-retcon-once-value.ll
@@ -159,7 +159,7 @@ declare void @print(i32)
; CHECK-NEXT: PostSpill:
; CHECK-NEXT: [[TMP0:%.*]] = tail call ptr @allocate(i32 16)
; CHECK-NEXT: store ptr [[TMP0]], ptr [[BUFFER:%.*]], align 8
-; CHECK-NEXT: [[VAL_SPILL_ADDR:%.*]] = getelementptr inbounds [[G_FRAME:%.*]], ptr [[TMP0]], i64 0, i32 1
+; CHECK-NEXT: [[VAL_SPILL_ADDR:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; CHECK-NEXT: store i32 [[VAL:%.*]], ptr [[VAL_SPILL_ADDR]], align 4
; CHECK-NEXT: store ptr [[ARRAY:%.*]], ptr [[TMP0]], align 8
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr [[ARRAY]], align 4
@@ -180,7 +180,7 @@ declare void @print(i32)
; CHECK-NEXT: store i32 0, ptr [[ARRAY_RELOAD]], align 4
; CHECK-NEXT: br label [[COROEND]]
; CHECK: CoroEnd:
-; CHECK-NEXT: [[VAL_RELOAD_ADDR:%.*]] = getelementptr inbounds [[G_FRAME:%.*]], ptr [[TMP2]], i64 0, i32 1
+; CHECK-NEXT: [[VAL_RELOAD_ADDR:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 8
; CHECK-NEXT: [[VAL_RELOAD:%.*]] = load i32, ptr [[VAL_RELOAD_ADDR]], align 4
; CHECK-NEXT: [[NEW_VAL:%.*]] = add i32 [[VAL_RELOAD]], 123
; CHECK-NEXT: tail call void @deallocate(ptr [[TMP2]])
@@ -198,7 +198,7 @@ declare void @print(i32)
; CHECK-NEXT: store i32 10, ptr [[ARRAY_RELOAD]], align 4
; CHECK-NEXT: br label [[COROEND]]
; CHECK: CoroEnd:
-; CHECK-NEXT: [[VAL_RELOAD_ADDR:%.*]] = getelementptr inbounds [[G_FRAME:%.*]], ptr [[TMP2]], i64 0, i32 1
+; CHECK-NEXT: [[VAL_RELOAD_ADDR:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 8
; CHECK-NEXT: [[VAL_RELOAD:%.*]] = load i32, ptr [[VAL_RELOAD_ADDR]], align 4
; CHECK-NEXT: [[NEW_VAL:%.*]] = add i32 [[VAL_RELOAD]], 123
; CHECK-NEXT: tail call void @deallocate(ptr [[TMP2]])
diff --git a/llvm/test/Transforms/Coroutines/coro-retcon-resume-values.ll b/llvm/test/Transforms/Coroutines/coro-retcon-resume-values.ll
index f4333a12bb34a8..4c10cb3805c7c3 100644
--- a/llvm/test/Transforms/Coroutines/coro-retcon-resume-values.ll
+++ b/llvm/test/Transforms/Coroutines/coro-retcon-resume-values.ll
@@ -38,12 +38,12 @@ define i32 @main() {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = tail call ptr @allocate(i32 12)
; CHECK-NEXT: store i32 1, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[N_VAL3_SPILL_ADDR_I:%.*]] = getelementptr inbounds [[F_FRAME:%.*]], ptr [[TMP0]], i64 0, i32 1
+; CHECK-NEXT: [[N_VAL3_SPILL_ADDR_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 4
; CHECK-NEXT: store i32 1, ptr [[N_VAL3_SPILL_ADDR_I]], align 4, !noalias !0
-; CHECK-NEXT: [[INPUT_SPILL_ADDR_I:%.*]] = getelementptr inbounds [[F_FRAME]], ptr [[TMP0]], i64 0, i32 2
+; CHECK-NEXT: [[INPUT_SPILL_ADDR_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; CHECK-NEXT: store i32 2, ptr [[INPUT_SPILL_ADDR_I]], align 4, !noalias !0
-; CHECK-NEXT: [[INPUT_RELOAD_ADDR13_I:%.*]] = getelementptr inbounds [[F_FRAME]], ptr [[TMP0]], i64 0, i32 2
-; CHECK-NEXT: [[N_VAL3_RELOAD_ADDR11_I:%.*]] = getelementptr inbounds [[F_FRAME]], ptr [[TMP0]], i64 0, i32 1
+; CHECK-NEXT: [[INPUT_RELOAD_ADDR13_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
+; CHECK-NEXT: [[N_VAL3_RELOAD_ADDR11_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 4
; CHECK-NEXT: store i32 3, ptr [[N_VAL3_RELOAD_ADDR11_I]], align 4, !noalias !3
; CHECK-NEXT: store i32 4, ptr [[INPUT_RELOAD_ADDR13_I]], align 4, !noalias !3
; CHECK-NEXT: tail call void @print(i32 7), !noalias !6
diff --git a/llvm/test/Transforms/Coroutines/coro-swifterror.ll b/llvm/test/Transforms/Coroutines/coro-swifterror.ll
index f192dbac3197e3..de28ed56081682 100644
--- a/llvm/test/Transforms/Coroutines/coro-swifterror.ll
+++ b/llvm/test/Transforms/Coroutines/coro-swifterror.ll
@@ -50,7 +50,7 @@ define ptr @g(ptr %buffer, i32 %n) {
; CHECK-NEXT: store ptr null, ptr [[TMP0]], align 4
; CHECK-NEXT: call void @maybeThrow(ptr nonnull swifterror [[TMP0]])
; CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[DOTSPILL_ADDR:%.*]] = getelementptr inbounds [[G_FRAME:%.*]], ptr [[BUFFER]], i32 0, i32 1
+; CHECK-NEXT: [[DOTSPILL_ADDR:%.*]] = getelementptr inbounds i8, ptr [[BUFFER]], i32 4
; CHECK-NEXT: store ptr [[TMP1]], ptr [[DOTSPILL_ADDR]], align 4
; CHECK-NEXT: call void @logError(ptr [[TMP1]])
; CHECK-NEXT: ret ptr @g.resume.0
diff --git a/llvm/test/Transforms/InstCombine/2007-03-25-BadShiftMask.ll b/llvm/test/Transforms/InstCombine/2007-03-25-BadShiftMask.ll
index ffc05a40865d3c..27005f38f29f11 100644
--- a/llvm/test/Transforms/InstCombine/2007-03-25-BadShiftMask.ll
+++ b/llvm/test/Transforms/InstCombine/2007-03-25-BadShiftMask.ll
@@ -10,7 +10,7 @@ define i32 @main() {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[U:%.*]] = alloca [[STRUCT__1ANON:%.*]], align 8
; CHECK-NEXT: store double 0x7FF0000000000000, ptr [[U]], align 8
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT__0ANON:%.*]], ptr [[U]], i64 0, i32 1
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 4
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = and i32 [[TMP6]], 2146435072
; CHECK-NEXT: [[DOTNOT:%.*]] = icmp eq i32 [[TMP0]], 2146435072
diff --git a/llvm/test/Transforms/InstCombine/2009-01-08-AlignAlloca.ll b/llvm/test/Transforms/InstCombine/2009-01-08-AlignAlloca.ll
index dee71b2290acfc..c31f6d4b4a1fe5 100644
--- a/llvm/test/Transforms/InstCombine/2009-01-08-AlignAlloca.ll
+++ b/llvm/test/Transforms/InstCombine/2009-01-08-AlignAlloca.ll
@@ -13,7 +13,7 @@ define i32 @bar(i64 %key_token2) nounwind {
; CHECK-NEXT: [[IOSPEC:%.*]] = alloca [[STRUCT_KEY:%.*]], align 8
; CHECK-NEXT: [[RET:%.*]] = alloca i32, align 4
; CHECK-NEXT: store i32 0, ptr [[IOSPEC]], align 4
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds { i32, i32 }, ptr [[IOSPEC]], i32 0, i32 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[IOSPEC]], i32 4
; CHECK-NEXT: store i32 0, ptr [[TMP0]], align 4
; CHECK-NEXT: store i64 [[KEY_TOKEN2:%.*]], ptr [[IOSPEC]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = call i32 (...) @foo(ptr nonnull byval([[STRUCT_KEY]]) align 4 [[IOSPEC]], ptr nonnull [[RET]]) #[[ATTR0:[0-9]+]]
diff --git a/llvm/test/Transforms/InstCombine/2009-02-20-InstCombine-SROA.ll b/llvm/test/Transforms/InstCombine/2009-02-20-InstCombine-SROA.ll
index 9f69d67f61e115..b532c815567389 100644
--- a/llvm/test/Transforms/InstCombine/2009-02-20-InstCombine-SROA.ll
+++ b/llvm/test/Transforms/InstCombine/2009-02-20-InstCombine-SROA.ll
@@ -24,7 +24,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: [[__LAST_ADDR_I_I:%.*]] = alloca %"struct.__gnu_cxx::__normal_iterator<int*,std::vector<int, std::allocator<int> > >", align 8
; IC-NEXT: [[TMP0:%.*]] = alloca i32, align 4
; IC-NEXT: store i32 42, ptr [[TMP0]], align 4
-; IC-NEXT: [[TMP1:%.*]] = getelementptr %"struct.std::_Vector_base<int,std::allocator<int> >::_Vector_impl", ptr [[X:%.*]], i32 0, i32 1
+; IC-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[X:%.*]], i32 4
; IC-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP1]], align 4
; IC-NEXT: [[TMP3:%.*]] = load ptr, ptr [[X]], align 4
; IC-NEXT: store ptr [[TMP3]], ptr [[__FIRST_ADDR_I_I]], align 4
@@ -45,7 +45,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT:%.*]]
; IC: bb2.i.i:
; IC-NEXT: [[TMP13:%.*]] = load ptr, ptr [[__FIRST_ADDR_I_I]], align 4
-; IC-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP13]], i32 1
+; IC-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i32 4
; IC-NEXT: store ptr [[TMP14]], ptr [[__FIRST_ADDR_I_I]], align 4
; IC-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; IC-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -56,7 +56,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC: bb5.i.i:
; IC-NEXT: [[TMP19:%.*]] = load ptr, ptr [[__FIRST_ADDR_I_I]], align 4
-; IC-NEXT: [[TMP20:%.*]] = getelementptr i32, ptr [[TMP19]], i32 1
+; IC-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i32 4
; IC-NEXT: store ptr [[TMP20]], ptr [[__FIRST_ADDR_I_I]], align 4
; IC-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; IC-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -67,7 +67,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC: bb8.i.i:
; IC-NEXT: [[TMP25:%.*]] = load ptr, ptr [[__FIRST_ADDR_I_I]], align 4
-; IC-NEXT: [[TMP26:%.*]] = getelementptr i32, ptr [[TMP25]], i32 1
+; IC-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[TMP25]], i32 4
; IC-NEXT: store ptr [[TMP26]], ptr [[__FIRST_ADDR_I_I]], align 4
; IC-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4
; IC-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP0]], align 4
@@ -78,7 +78,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC: bb11.i.i:
; IC-NEXT: [[TMP31:%.*]] = load ptr, ptr [[__FIRST_ADDR_I_I]], align 4
-; IC-NEXT: [[TMP32:%.*]] = getelementptr i32, ptr [[TMP31]], i32 1
+; IC-NEXT: [[TMP32:%.*]] = getelementptr i8, ptr [[TMP31]], i32 4
; IC-NEXT: store ptr [[TMP32]], ptr [[__FIRST_ADDR_I_I]], align 4
; IC-NEXT: [[TMP33:%.*]] = add nsw i32 [[__TRIP_COUNT_0_I_I:%.*]], -1
; IC-NEXT: br label [[BB12_I_I]]
@@ -109,7 +109,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC: bb17.i.i:
; IC-NEXT: [[TMP46:%.*]] = load ptr, ptr [[__FIRST_ADDR_I_I]], align 4
-; IC-NEXT: [[TMP47:%.*]] = getelementptr i32, ptr [[TMP46]], i32 1
+; IC-NEXT: [[TMP47:%.*]] = getelementptr i8, ptr [[TMP46]], i32 4
; IC-NEXT: store ptr [[TMP47]], ptr [[__FIRST_ADDR_I_I]], align 4
; IC-NEXT: br label [[BB18_I_I]]
; IC: bb18.i.i:
@@ -123,7 +123,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC: bb21.i.i:
; IC-NEXT: [[TMP53:%.*]] = load ptr, ptr [[__FIRST_ADDR_I_I]], align 4
-; IC-NEXT: [[TMP54:%.*]] = getelementptr i32, ptr [[TMP53]], i32 1
+; IC-NEXT: [[TMP54:%.*]] = getelementptr i8, ptr [[TMP53]], i32 4
; IC-NEXT: store ptr [[TMP54]], ptr [[__FIRST_ADDR_I_I]], align 4
; IC-NEXT: br label [[BB22_I_I]]
; IC: bb22.i.i:
@@ -137,7 +137,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC: bb25.i.i:
; IC-NEXT: [[TMP60:%.*]] = load ptr, ptr [[__FIRST_ADDR_I_I]], align 4
-; IC-NEXT: [[TMP61:%.*]] = getelementptr i32, ptr [[TMP60]], i32 1
+; IC-NEXT: [[TMP61:%.*]] = getelementptr i8, ptr [[TMP60]], i32 4
; IC-NEXT: store ptr [[TMP61]], ptr [[__FIRST_ADDR_I_I]], align 4
; IC-NEXT: br label [[BB26_I_I]]
; IC: bb26.i.i:
@@ -151,7 +151,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
;
; IC_SROA-LABEL: @_Z3fooRSt6vectorIiSaIiEE(
; IC_SROA-NEXT: entry:
-; IC_SROA-NEXT: [[TMP0:%.*]] = getelementptr %"struct.std::_Vector_base<int,std::allocator<int> >::_Vector_impl", ptr [[X:%.*]], i32 0, i32 1
+; IC_SROA-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[X:%.*]], i32 4
; IC_SROA-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 4
; IC_SROA-NEXT: [[TMP2:%.*]] = load ptr, ptr [[X]], align 4
; IC_SROA-NEXT: [[TMP3:%.*]] = ptrtoint ptr [[TMP1]] to i32
@@ -166,28 +166,28 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC_SROA: bb1.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT:%.*]]
; IC_SROA: bb2.i.i:
-; IC_SROA-NEXT: [[TMP9:%.*]] = getelementptr i32, ptr [[__FIRST_ADDR_I_I_SROA_0_0]], i32 1
+; IC_SROA-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[__FIRST_ADDR_I_I_SROA_0_0]], i32 4
; IC_SROA-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
; IC_SROA-NEXT: [[TMP11:%.*]] = icmp eq i32 [[TMP10]], 42
; IC_SROA-NEXT: br i1 [[TMP11]], label [[BB4_I_I:%.*]], label [[BB5_I_I:%.*]]
; IC_SROA: bb4.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC_SROA: bb5.i.i:
-; IC_SROA-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP9]], i32 1
+; IC_SROA-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP9]], i32 4
; IC_SROA-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; IC_SROA-NEXT: [[TMP14:%.*]] = icmp eq i32 [[TMP13]], 42
; IC_SROA-NEXT: br i1 [[TMP14]], label [[BB7_I_I:%.*]], label [[BB8_I_I:%.*]]
; IC_SROA: bb7.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC_SROA: bb8.i.i:
-; IC_SROA-NEXT: [[TMP15:%.*]] = getelementptr i32, ptr [[TMP12]], i32 1
+; IC_SROA-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[TMP12]], i32 4
; IC_SROA-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; IC_SROA-NEXT: [[TMP17:%.*]] = icmp eq i32 [[TMP16]], 42
; IC_SROA-NEXT: br i1 [[TMP17]], label [[BB10_I_I:%.*]], label [[BB11_I_I:%.*]]
; IC_SROA: bb10.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC_SROA: bb11.i.i:
-; IC_SROA-NEXT: [[TMP18:%.*]] = getelementptr i32, ptr [[TMP15]], i32 1
+; IC_SROA-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[TMP15]], i32 4
; IC_SROA-NEXT: [[TMP19:%.*]] = add nsw i32 [[__TRIP_COUNT_0_I_I:%.*]], -1
; IC_SROA-NEXT: br label [[BB12_I_I]]
; IC_SROA: bb12.i.i:
@@ -212,7 +212,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC_SROA: bb16.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC_SROA: bb17.i.i:
-; IC_SROA-NEXT: [[TMP27:%.*]] = getelementptr i32, ptr [[__FIRST_ADDR_I_I_SROA_0_0]], i32 1
+; IC_SROA-NEXT: [[TMP27:%.*]] = getelementptr i8, ptr [[__FIRST_ADDR_I_I_SROA_0_0]], i32 4
; IC_SROA-NEXT: br label [[BB18_I_I]]
; IC_SROA: bb18.i.i:
; IC_SROA-NEXT: [[__FIRST_ADDR_I_I_SROA_0_1:%.*]] = phi ptr [ [[TMP27]], [[BB17_I_I]] ], [ [[__FIRST_ADDR_I_I_SROA_0_0]], [[BB13_I_I]] ]
@@ -222,7 +222,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC_SROA: bb20.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC_SROA: bb21.i.i:
-; IC_SROA-NEXT: [[TMP30:%.*]] = getelementptr i32, ptr [[__FIRST_ADDR_I_I_SROA_0_1]], i32 1
+; IC_SROA-NEXT: [[TMP30:%.*]] = getelementptr i8, ptr [[__FIRST_ADDR_I_I_SROA_0_1]], i32 4
; IC_SROA-NEXT: br label [[BB22_I_I]]
; IC_SROA: bb22.i.i:
; IC_SROA-NEXT: [[__FIRST_ADDR_I_I_SROA_0_2:%.*]] = phi ptr [ [[TMP30]], [[BB21_I_I]] ], [ [[__FIRST_ADDR_I_I_SROA_0_0]], [[BB13_I_I]] ]
@@ -232,7 +232,7 @@ define ptr @_Z3fooRSt6vectorIiSaIiEE(ptr %X) {
; IC_SROA: bb24.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
; IC_SROA: bb25.i.i:
-; IC_SROA-NEXT: [[TMP33:%.*]] = getelementptr i32, ptr [[__FIRST_ADDR_I_I_SROA_0_2]], i32 1
+; IC_SROA-NEXT: [[TMP33:%.*]] = getelementptr i8, ptr [[__FIRST_ADDR_I_I_SROA_0_2]], i32 4
; IC_SROA-NEXT: br label [[BB26_I_I]]
; IC_SROA: bb26.i.i:
; IC_SROA-NEXT: br label [[_ZST4FINDIN9__GNU_CXX17__NORMAL_ITERATORIPIST6VECTORIISAIIEEEEIET_S7_S7_RKT0__EXIT]]
diff --git a/llvm/test/Transforms/InstCombine/X86/x86-addsub-inseltpoison.ll b/llvm/test/Transforms/InstCombine/X86/x86-addsub-inseltpoison.ll
index ad4088c63e7613..f03b633c581d6e 100644
--- a/llvm/test/Transforms/InstCombine/X86/x86-addsub-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/X86/x86-addsub-inseltpoison.ll
@@ -146,7 +146,7 @@ define void @PR46277(float %0, float %1, float %2, float %3, <4 x float> %4, ptr
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <4 x float> [[TMP7]], float [[TMP1:%.*]], i64 1
; CHECK-NEXT: [[TMP9:%.*]] = tail call <4 x float> @llvm.x86.sse3.addsub.ps(<4 x float> [[TMP8]], <4 x float> [[TMP4:%.*]])
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x float> [[TMP9]], i64 0
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[TMP5:%.*]], i64 1
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[TMP5:%.*]], i64 4
; CHECK-NEXT: store float [[TMP10]], ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x float> [[TMP9]], i64 1
; CHECK-NEXT: store float [[TMP12]], ptr [[TMP11]], align 4
diff --git a/llvm/test/Transforms/InstCombine/X86/x86-addsub.ll b/llvm/test/Transforms/InstCombine/X86/x86-addsub.ll
index c788e4ea81efc6..9b0387e51eead7 100644
--- a/llvm/test/Transforms/InstCombine/X86/x86-addsub.ll
+++ b/llvm/test/Transforms/InstCombine/X86/x86-addsub.ll
@@ -146,7 +146,7 @@ define void @PR46277(float %0, float %1, float %2, float %3, <4 x float> %4, ptr
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <4 x float> [[TMP7]], float [[TMP1:%.*]], i64 1
; CHECK-NEXT: [[TMP9:%.*]] = tail call <4 x float> @llvm.x86.sse3.addsub.ps(<4 x float> [[TMP8]], <4 x float> [[TMP4:%.*]])
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x float> [[TMP9]], i64 0
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[TMP5:%.*]], i64 1
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[TMP5:%.*]], i64 4
; CHECK-NEXT: store float [[TMP10]], ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x float> [[TMP9]], i64 1
; CHECK-NEXT: store float [[TMP12]], ptr [[TMP11]], align 4
diff --git a/llvm/test/Transforms/InstCombine/add3.ll b/llvm/test/Transforms/InstCombine/add3.ll
index ba9ce8e64438d2..87e12088703946 100644
--- a/llvm/test/Transforms/InstCombine/add3.ll
+++ b/llvm/test/Transforms/InstCombine/add3.ll
@@ -15,7 +15,7 @@ define void @test2(i32 %.val24) {
; CHECK-NEXT: store i32 1, ptr [[TMP1]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = add i32 [[DOTVAL24]], -16
; CHECK-NEXT: [[TMP3:%.*]] = inttoptr i32 [[TMP2]] to ptr
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[TMP3]], i32 1
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[TMP3]], i32 4
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP6:%.*]] = tail call i32 @callee(i32 [[TMP5]])
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/InstCombine/array.ll b/llvm/test/Transforms/InstCombine/array.ll
index 396a7aa340f6bf..236821d8ba4c02 100644
--- a/llvm/test/Transforms/InstCombine/array.ll
+++ b/llvm/test/Transforms/InstCombine/array.ll
@@ -7,7 +7,7 @@ define void @test(ptr %ptr, i32 %a, i32 %b) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[A]] to i64
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[TMP0]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[TMP1]], i64 10
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[TMP1]], i64 40
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret void
;
@@ -79,7 +79,7 @@ define void @test_zext_nneg(ptr %ptr, i32 %a, i32 %b) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[A]] to i64
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[TMP0]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[TMP1]], i64 10
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[TMP1]], i64 40
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/assume.ll b/llvm/test/Transforms/InstCombine/assume.ll
index c7609564685fed..927f0a86b0a252 100644
--- a/llvm/test/Transforms/InstCombine/assume.ll
+++ b/llvm/test/Transforms/InstCombine/assume.ll
@@ -774,7 +774,7 @@ exit:
define void @canonicalize_assume(ptr %0) {
; DEFAULT-LABEL: @canonicalize_assume(
-; DEFAULT-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP0:%.*]], i64 2
+; DEFAULT-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP0:%.*]], i64 8
; DEFAULT-NEXT: call void @llvm.assume(i1 true) [ "align"(ptr [[TMP2]], i64 16) ]
; DEFAULT-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/cast_phi.ll b/llvm/test/Transforms/InstCombine/cast_phi.ll
index 5b4425b4d83082..68847e73ac5d29 100644
--- a/llvm/test/Transforms/InstCombine/cast_phi.ll
+++ b/llvm/test/Transforms/InstCombine/cast_phi.ll
@@ -16,9 +16,9 @@ define void @MainKernel(i32 %iNumSteps, i32 %tid, i32 %base) {
; CHECK-NEXT: [[CMP7:%.*]] = icmp eq i32 [[TID]], 0
; CHECK-NEXT: br i1 [[CMP7]], label [[DOTBB1:%.*]], label [[DOTBB2:%.*]]
; CHECK: .bb1:
-; CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [258 x float], ptr [[CALLA]], i64 0, i64 256
+; CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8, ptr [[CALLA]], i64 1024
; CHECK-NEXT: store float [[CONV_I]], ptr [[ARRAYIDX10]], align 4
-; CHECK-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [258 x float], ptr [[CALLB]], i64 0, i64 256
+; CHECK-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, ptr [[CALLB]], i64 1024
; CHECK-NEXT: store float 0.000000e+00, ptr [[ARRAYIDX11]], align 4
; CHECK-NEXT: br label [[DOTBB2]]
; CHECK: .bb2:
diff --git a/llvm/test/Transforms/InstCombine/catchswitch-phi.ll b/llvm/test/Transforms/InstCombine/catchswitch-phi.ll
index 9a12e8bebabf64..038847609b0f90 100644
--- a/llvm/test/Transforms/InstCombine/catchswitch-phi.ll
+++ b/llvm/test/Transforms/InstCombine/catchswitch-phi.ll
@@ -22,11 +22,11 @@ define void @test0(i1 %c1) personality ptr @__gxx_wasm_personality_v0 {
; CHECK-NEXT: [[TMP0:%.*]] = alloca [[STRUCT_BLAM:%.*]], align 4
; CHECK-NEXT: br i1 [[C1:%.*]], label [[BB1:%.*]], label [[BB2:%.*]]
; CHECK: bb1:
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_BLAM]], ptr [[TMP0]], i32 0, i32 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i32 4
; CHECK-NEXT: invoke void @foo()
; CHECK-NEXT: to label [[BB3:%.*]] unwind label [[BB4:%.*]]
; CHECK: bb2:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_BLAM]], ptr [[TMP0]], i32 0, i32 1
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i32 4
; CHECK-NEXT: invoke void @foo()
; CHECK-NEXT: to label [[BB3]] unwind label [[BB4]]
; CHECK: bb3:
diff --git a/llvm/test/Transforms/InstCombine/compare-alloca.ll b/llvm/test/Transforms/InstCombine/compare-alloca.ll
index 9ca8dc866e5658..c2639d39bc6566 100644
--- a/llvm/test/Transforms/InstCombine/compare-alloca.ll
+++ b/llvm/test/Transforms/InstCombine/compare-alloca.ll
@@ -75,7 +75,7 @@ define i1 @alloca_argument_compare_escaped_through_store(ptr %arg, ptr %ptr) {
; CHECK-LABEL: @alloca_argument_compare_escaped_through_store(
; CHECK-NEXT: [[ALLOC:%.*]] = alloca i64, align 8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[ALLOC]], [[ARG:%.*]]
-; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i64, ptr [[ALLOC]], i32 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr [[ALLOC]], i32 8
; CHECK-NEXT: store ptr [[P]], ptr [[PTR:%.*]], align 4
; CHECK-NEXT: ret i1 [[CMP]]
;
diff --git a/llvm/test/Transforms/InstCombine/extractvalue.ll b/llvm/test/Transforms/InstCombine/extractvalue.ll
index fce85596cf5f26..0b7c6498390631 100644
--- a/llvm/test/Transforms/InstCombine/extractvalue.ll
+++ b/llvm/test/Transforms/InstCombine/extractvalue.ll
@@ -51,7 +51,7 @@ define i32 @foo(i32 %a, i32 %b) {
; The new load should be in the same spot as the old load.
define i32 @extract2gep(ptr %pair, ptr %P) {
; CHECK-LABEL: @extract2gep(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds { i16, i32 }, ptr [[PAIR:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[PAIR:%.*]], i64 4
; CHECK-NEXT: [[E:%.*]] = load i32, ptr [[TMP1]], align 4
; CHECK-NEXT: store i32 0, ptr [[P:%.*]], align 4
; CHECK-NEXT: br label [[LOOP:%.*]]
@@ -82,7 +82,7 @@ end:
; to a 3-index inbounds gep + smaller load.
define i16 @doubleextract2gep(ptr %arg) {
; CHECK-LABEL: @doubleextract2gep(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds { i16, { i32, i16 } }, ptr [[ARG:%.*]], i64 0, i32 1, i32 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[ARG:%.*]], i64 8
; CHECK-NEXT: [[E2:%.*]] = load i16, ptr [[TMP1]], align 2
; CHECK-NEXT: ret i16 [[E2]]
;
diff --git a/llvm/test/Transforms/InstCombine/gep-addrspace.ll b/llvm/test/Transforms/InstCombine/gep-addrspace.ll
index 64813264e97668..789a31e5939466 100644
--- a/llvm/test/Transforms/InstCombine/gep-addrspace.ll
+++ b/llvm/test/Transforms/InstCombine/gep-addrspace.ll
@@ -45,7 +45,7 @@ define { i8, i8 } @inbounds_after_addrspacecast() {
; CHECK-NEXT: [[T0:%.*]] = alloca i16, align 2
; CHECK-NEXT: call void @escape_alloca(ptr nonnull [[T0]])
; CHECK-NEXT: [[T1:%.*]] = addrspacecast ptr [[T0]] to ptr addrspace(11)
-; CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds [2 x i8], ptr addrspace(11) [[T1]], i64 0, i64 1
+; CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i8, ptr addrspace(11) [[T1]], i64 1
; CHECK-NEXT: [[T3:%.*]] = load i8, ptr addrspace(11) [[T2]], align 1
; CHECK-NEXT: [[INSERT:%.*]] = insertvalue { i8, i8 } zeroinitializer, i8 [[T3]], 1
; CHECK-NEXT: ret { i8, i8 } [[INSERT]]
diff --git a/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll b/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll
index 2d5721b8c3656d..849793c7a85aff 100644
--- a/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll
+++ b/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll
@@ -13,7 +13,7 @@ declare void @use(i1)
; result = (((ptr) p + a) + b) + 1
define ptr @basic(ptr %p, i64 %a, i64 %b) {
; CHECK-LABEL: @basic(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[A:%.*]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[B:%.*]]
; CHECK-NEXT: ret ptr [[TMP3]]
@@ -50,9 +50,9 @@ define ptr @partialConstant2(ptr %p, i64 %a, i64 %b) {
; result = ((ptr) p + a) + 3
define ptr @merge(ptr %p, i64 %a) {
; CHECK-LABEL: @merge(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[A:%.*]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 8
; CHECK-NEXT: ret ptr [[TMP3]]
;
%1 = getelementptr inbounds i32, ptr %p, i64 1
@@ -67,12 +67,12 @@ define ptr @merge(ptr %p, i64 %a) {
; result = (ptr) ((ptr) ((ptr) ptr + a) + (a * b)) + 9
define ptr @nested(ptr %p, i64 %a, i64 %b) {
; CHECK-LABEL: @nested(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds <3 x i32>, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 16
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[A:%.*]]
; CHECK-NEXT: [[TMP3:%.*]] = mul i64 [[A]], [[B:%.*]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds <5 x i32>, ptr [[TMP2]], i64 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 128
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i16, ptr [[TMP4]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds <4 x i32>, ptr [[TMP5]], i64 1
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i64 16
; CHECK-NEXT: ret ptr [[TMP6]]
;
%1 = getelementptr inbounds <3 x i32>, ptr %p, i64 1
@@ -87,7 +87,7 @@ define ptr @nested(ptr %p, i64 %a, i64 %b) {
; It is valid to swap if the source operand of the first GEP has multiple uses.
define ptr @multipleUses1(ptr %p) {
; CHECK-LABEL: @multipleUses1(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[P]] to i64
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[TMP3]]
@@ -101,7 +101,7 @@ define ptr @multipleUses1(ptr %p) {
; It is valid to swap if the second GEP has multiple uses.
define ptr @multipleUses2(ptr %p, i64 %a) {
; CHECK-LABEL: @multipleUses2(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[A:%.*]]
; CHECK-NEXT: call void @use(ptr nonnull [[TMP2]])
; CHECK-NEXT: ret ptr [[TMP2]]
@@ -115,7 +115,7 @@ define ptr @multipleUses2(ptr %p, i64 %a) {
; Negative test. It is not valid to swap if the first GEP has multiple uses.
define ptr @multipleUses3(ptr %p) {
; CHECK-LABEL: @multipleUses3(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[TMP1]] to i64
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[TMP3]]
diff --git a/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll b/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
index d0e5919b09f1ce..99cdb6bc760b46 100644
--- a/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
+++ b/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
@@ -94,7 +94,7 @@ define void @PR37005(ptr %base, ptr %in) {
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[E1:%.*]] = getelementptr inbounds ptr, ptr [[IN:%.*]], i64 undef
-; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds ptr, ptr [[E1]], i64 6
+; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds i8, ptr [[E1]], i64 48
; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds ptr, ptr [[E2]], <2 x i64> <i64 0, i64 1>
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint <2 x ptr> [[E4]] to <2 x i64>
; CHECK-NEXT: [[TMP0:%.*]] = lshr <2 x i64> [[PI1]], <i64 14, i64 14>
@@ -126,7 +126,7 @@ define void @PR37005_2(ptr %base, ptr %in) {
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[E1:%.*]] = getelementptr inbounds ptr, ptr [[IN:%.*]], i64 undef
-; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds ptr, ptr [[E1]], i64 6
+; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds i8, ptr [[E1]], i64 48
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint ptr [[E2]] to i64
; CHECK-NEXT: [[TMP0:%.*]] = lshr i64 [[PI1]], 14
; CHECK-NEXT: [[SL1:%.*]] = and i64 [[TMP0]], 1125899906842496
@@ -156,7 +156,7 @@ define void @PR37005_3(<2 x ptr> %base, ptr %in) {
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[E1:%.*]] = getelementptr inbounds ptr, ptr [[IN:%.*]], i64 undef
-; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds ptr, ptr [[E1]], i64 6
+; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds i8, ptr [[E1]], i64 48
; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds ptr, ptr [[E2]], <2 x i64> <i64 0, i64 1>
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint <2 x ptr> [[E4]] to <2 x i64>
; CHECK-NEXT: [[TMP0:%.*]] = lshr <2 x i64> [[PI1]], <i64 14, i64 14>
diff --git a/llvm/test/Transforms/InstCombine/gep-custom-dl.ll b/llvm/test/Transforms/InstCombine/gep-custom-dl.ll
index 7be15e38d44a7c..d9449e05612c38 100644
--- a/llvm/test/Transforms/InstCombine/gep-custom-dl.ll
+++ b/llvm/test/Transforms/InstCombine/gep-custom-dl.ll
@@ -13,7 +13,7 @@ target datalayout = "e-m:m-p:40:64:64:32-i32:32-i16:16-i8:8-n32"
; Test that two array indexing geps fold
define ptr @test1(ptr %I) {
; CHECK-LABEL: @test1(
-; CHECK-NEXT: [[B:%.*]] = getelementptr i32, ptr [[I:%.*]], i32 21
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[I:%.*]], i32 84
; CHECK-NEXT: ret ptr [[B]]
;
%A = getelementptr i32, ptr %I, i8 17
@@ -24,7 +24,7 @@ define ptr @test1(ptr %I) {
; Test that two getelementptr insts fold
define ptr @test2(ptr %I) {
; CHECK-LABEL: @test2(
-; CHECK-NEXT: [[A:%.*]] = getelementptr { i32 }, ptr [[I:%.*]], i32 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i32 4
; CHECK-NEXT: ret ptr [[A]]
;
%A = getelementptr { i32 }, ptr %I, i32 1
diff --git a/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll b/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll
index bcba8eadb1f966..509d19c8c00c7e 100644
--- a/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll
+++ b/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll
@@ -12,7 +12,7 @@ target datalayout = "i24:8:8"
; result = (ptr) p + 3
define ptr @mergeBasic(ptr %p) {
; CHECK-LABEL: @mergeBasic(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 12
; CHECK-NEXT: ret ptr [[TMP1]]
;
%1 = getelementptr inbounds i32, ptr %p, i64 1
@@ -58,7 +58,7 @@ define ptr @zeroSum(ptr %p) {
; result = (ptr) ((ptr) p + 1) + 17
define ptr @array1(ptr %p) {
; CHECK-LABEL: @array1(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [20 x i8], ptr [[P:%.*]], i64 1, i64 17
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 37
; CHECK-NEXT: ret ptr [[TMP1]]
;
%1 = getelementptr inbounds [20 x i8], ptr %p, i64 1, i64 1
@@ -93,7 +93,7 @@ define ptr @struct1(ptr %p) {
; result = &((struct.A*) p - 1).member1
define ptr @struct2(ptr %p) {
; CHECK-LABEL: @struct2(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [[STRUCT_A:%.*]], ptr [[P:%.*]], i64 -1, i32 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 -4
; CHECK-NEXT: ret ptr [[TMP1]]
;
%1 = getelementptr inbounds %struct.A, ptr %p, i64 0, i32 1
@@ -104,7 +104,7 @@ define ptr @struct2(ptr %p) {
; result = (ptr) &((struct.B) p)[0].member2.member0 + 7
define ptr @structStruct(ptr %p) {
; CHECK-LABEL: @structStruct(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_B:%.*]], ptr [[P:%.*]], i64 0, i32 2, i32 0, i64 7
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 15
; CHECK-NEXT: ret ptr [[TMP1]]
;
%1 = getelementptr inbounds %struct.B, ptr %p, i64 0, i32 2, i32 0, i64 3
@@ -144,7 +144,7 @@ define ptr @notDivisible(ptr %p) {
; or divisible by the other's size.
define ptr @partialConstant2(ptr %p, i64 %a) {
; CHECK-LABEL: @partialConstant2(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [4 x i64], ptr [[TMP1]], i64 [[A:%.*]], i64 2
; CHECK-NEXT: ret ptr [[TMP2]]
;
@@ -157,7 +157,7 @@ define ptr @partialConstant2(ptr %p, i64 %a) {
; first GEP by the second GEP.
define ptr @partialConstant3(ptr %p) {
; CHECK-LABEL: @partialConstant3(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[TMP1]] to i64
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [4 x i64], ptr [[TMP1]], i64 [[TMP2]], i64 2
; CHECK-NEXT: ret ptr [[TMP3]]
@@ -200,7 +200,7 @@ define ptr @partialConstantMemberAliasing2(ptr %p, i64 %a) {
define ptr @partialConstantMemberAliasing3(ptr %p, i64 %a) {
; CHECK-LABEL: @partialConstantMemberAliasing3(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[A:%.*]], i32 2
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 1
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 4
; CHECK-NEXT: ret ptr [[TMP2]]
;
%1 = getelementptr inbounds %struct.C, ptr %p, i64 %a, i32 2
diff --git a/llvm/test/Transforms/InstCombine/gep-vector-indices.ll b/llvm/test/Transforms/InstCombine/gep-vector-indices.ll
index d9f7e4ff9f701b..e9534e45ec141d 100644
--- a/llvm/test/Transforms/InstCombine/gep-vector-indices.ll
+++ b/llvm/test/Transforms/InstCombine/gep-vector-indices.ll
@@ -3,7 +3,7 @@
define ptr @vector_splat_indices_v2i64_ext0(ptr %a) {
; CHECK-LABEL: @vector_splat_indices_v2i64_ext0(
-; CHECK-NEXT: [[RES:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 4
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 16
; CHECK-NEXT: ret ptr [[RES]]
;
%gep = getelementptr i32, ptr %a, <2 x i64> <i64 4, i64 4>
@@ -13,7 +13,7 @@ define ptr @vector_splat_indices_v2i64_ext0(ptr %a) {
define ptr @vector_splat_indices_nxv2i64_ext0(ptr %a) {
; CHECK-LABEL: @vector_splat_indices_nxv2i64_ext0(
-; CHECK-NEXT: [[RES:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 4
+; CHECK-NEXT: [[RES:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 16
; CHECK-NEXT: ret ptr [[RES]]
;
%tmp = insertelement <vscale x 2 x i64> poison, i64 4, i32 0
@@ -74,7 +74,7 @@ define ptr @vector_splat_ptrs_nxv2i64_ext0(ptr %a, i64 %index) {
define ptr @vector_struct1_splat_indices_v4i64_ext1(ptr %a) {
; CHECK-LABEL: @vector_struct1_splat_indices_v4i64_ext1(
-; CHECK-NEXT: [[RES:%.*]] = getelementptr { float, float }, ptr [[A:%.*]], i64 4, i32 0
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 32
; CHECK-NEXT: ret ptr [[RES]]
;
%gep = getelementptr {float, float}, ptr %a, <4 x i32> <i32 4, i32 4, i32 4, i32 4>, i32 0
@@ -85,7 +85,7 @@ define ptr @vector_struct1_splat_indices_v4i64_ext1(ptr %a) {
define ptr @vector_struct2_splat_indices_v4i64_ext1(ptr %a) {
; CHECK-LABEL: @vector_struct2_splat_indices_v4i64_ext1(
-; CHECK-NEXT: [[RES:%.*]] = getelementptr { float, [8 x float] }, ptr [[A:%.*]], i64 2, i32 1, i64 4
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 92
; CHECK-NEXT: ret ptr [[RES]]
;
%gep = getelementptr {float, [8 x float]}, ptr %a, i32 2, i32 1, <4 x i32> <i32 4, i32 4, i32 4, i32 4>
diff --git a/llvm/test/Transforms/InstCombine/gep-vector.ll b/llvm/test/Transforms/InstCombine/gep-vector.ll
index 9ebb245325b148..3465c620b70fe5 100644
--- a/llvm/test/Transforms/InstCombine/gep-vector.ll
+++ b/llvm/test/Transforms/InstCombine/gep-vector.ll
@@ -136,7 +136,7 @@ define ptr @test_accumulate_constant_offset_vscale_nonzero(<vscale x 16 x i1> %p
define ptr @test_accumulate_constant_offset_vscale_zero(<vscale x 16 x i1> %pg, ptr %base) {
; CHECK-LABEL: @test_accumulate_constant_offset_vscale_zero(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr <vscale x 16 x i8>, ptr [[BASE:%.*]], i64 0, i64 4
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 4
; CHECK-NEXT: ret ptr [[GEP]]
;
%gep = getelementptr <vscale x 16 x i8>, ptr %base, i64 0, i64 4
diff --git a/llvm/test/Transforms/InstCombine/gepphigep.ll b/llvm/test/Transforms/InstCombine/gepphigep.ll
index bc2d8201a36a3d..93266c67543963 100644
--- a/llvm/test/Transforms/InstCombine/gepphigep.ll
+++ b/llvm/test/Transforms/InstCombine/gepphigep.ll
@@ -54,7 +54,7 @@ define i32 @test2(ptr %dm, i1 %tmp4, i64 %tmp9, i64 %tmp19) {
; CHECK-NEXT: store i32 0, ptr [[TMP10]], align 4
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT2]], ptr [[TMP1]], i64 [[TMP19:%.*]]
; CHECK-NEXT: store i32 0, ptr [[TMP20]], align 4
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT2]], ptr [[TMP1]], i64 [[TMP9]], i32 1
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i64 4
; CHECK-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4
; CHECK-NEXT: ret i32 [[TMP25]]
;
diff --git a/llvm/test/Transforms/InstCombine/getelementptr.ll b/llvm/test/Transforms/InstCombine/getelementptr.ll
index 373b7f5f2fc0a5..f1b7489480a691 100644
--- a/llvm/test/Transforms/InstCombine/getelementptr.ll
+++ b/llvm/test/Transforms/InstCombine/getelementptr.ll
@@ -42,7 +42,7 @@ define ptr @test2(ptr %I) {
; Test that two array indexing geps fold
define ptr @test3(ptr %I) {
; CHECK-LABEL: @test3(
-; CHECK-NEXT: [[B:%.*]] = getelementptr i32, ptr [[I:%.*]], i64 21
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 84
; CHECK-NEXT: ret ptr [[B]]
;
%A = getelementptr i32, ptr %I, i64 17
@@ -53,7 +53,7 @@ define ptr @test3(ptr %I) {
; Test that two getelementptr insts fold
define ptr @test4(ptr %I) {
; CHECK-LABEL: @test4(
-; CHECK-NEXT: [[A:%.*]] = getelementptr { i32 }, ptr [[I:%.*]], i64 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 4
; CHECK-NEXT: ret ptr [[A]]
;
%A = getelementptr { i32 }, ptr %I, i64 1
@@ -152,8 +152,8 @@ define i1 @test10(ptr %x, ptr %y) {
define i1 @test10_addrspacecast(ptr %x, ptr addrspace(3) %y) {
; CHECK-LABEL: @test10_addrspacecast(
-; CHECK-NEXT: [[T1:%.*]] = getelementptr { i32, i32 }, ptr [[X:%.*]], i64 0, i32 1
-; CHECK-NEXT: [[T3:%.*]] = getelementptr { i32, i32 }, ptr addrspace(3) [[Y:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[T1:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 4
+; CHECK-NEXT: [[T3:%.*]] = getelementptr i8, ptr addrspace(3) [[Y:%.*]], i64 4
; CHECK-NEXT: [[T3_C:%.*]] = addrspacecast ptr addrspace(3) [[T3]] to ptr
; CHECK-NEXT: [[T4:%.*]] = icmp eq ptr [[T1]], [[T3_C]]
; CHECK-NEXT: ret i1 [[T4]]
@@ -180,7 +180,7 @@ define i1 @test11(ptr %X) {
define i32 @test12(ptr %a) {
; CHECK-LABEL: @test12(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[G3:%.*]] = getelementptr [[STRUCT_A:%.*]], ptr [[A:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[G3:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 8
; CHECK-NEXT: store i32 10, ptr [[G3]], align 4
; CHECK-NEXT: ret i32 10
;
@@ -586,7 +586,7 @@ define i32 @test27(ptr %to, ptr %from) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[FROM_ADDR:%.*]] = alloca ptr, align 8
; CHECK-NEXT: [[T344:%.*]] = load ptr, ptr [[FROM_ADDR]], align 8
-; CHECK-NEXT: [[T348:%.*]] = getelementptr [[STRUCT_SIGINFO_T:%.*]], ptr [[T344]], i64 0, i32 3, i32 0, i32 3
+; CHECK-NEXT: [[T348:%.*]] = getelementptr i8, ptr [[T344]], i64 24
; CHECK-NEXT: [[T351:%.*]] = load i32, ptr [[T348]], align 8
; CHECK-NEXT: [[T360:%.*]] = call i32 asm sideeffect "...", "=r,ir,*m,i,0,~{dirflag},~{fpsr},~{flags}"(i32 [[T351]], ptr elementtype([[STRUCT___LARGE_STRUCT:%.*]]) null, i32 -14, i32 0) #[[ATTR0:[0-9]+]]
; CHECK-NEXT: unreachable
@@ -613,14 +613,15 @@ define i32 @test28() nounwind {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[ORIENTATIONS:%.*]] = alloca [1 x [1 x %struct.x]], align 8
; CHECK-NEXT: [[T3:%.*]] = call i32 @puts(ptr noundef nonnull dereferenceable(1) @.str) #[[ATTR0]]
+; CHECK-NEXT: [[T45:%.*]] = getelementptr inbounds i8, ptr [[ORIENTATIONS]], i64 1
; CHECK-NEXT: br label [[BB10:%.*]]
; CHECK: bb10:
; CHECK-NEXT: [[INDVAR:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INDVAR_NEXT:%.*]], [[BB10]] ]
; CHECK-NEXT: [[T12_REC:%.*]] = xor i32 [[INDVAR]], -1
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[T12_REC]] to i64
-; CHECK-NEXT: [[T12:%.*]] = getelementptr inbounds [1 x [1 x %struct.x]], ptr [[ORIENTATIONS]], i64 1, i64 0, i64 [[TMP0]]
+; CHECK-NEXT: [[T12:%.*]] = getelementptr inbounds [[STRUCT_X:%.*]], ptr [[T45]], i64 [[TMP0]]
; CHECK-NEXT: [[T16:%.*]] = call i32 (ptr, ...) @printf(ptr noundef nonnull dereferenceable(1) @.str1, ptr nonnull [[T12]]) #[[ATTR0]]
-; CHECK-NEXT: [[T84:%.*]] = icmp eq i32 [[INDVAR]], 0
+; CHECK-NEXT: [[T84:%.*]] = icmp eq ptr [[T12]], [[ORIENTATIONS]]
; CHECK-NEXT: [[INDVAR_NEXT]] = add i32 [[INDVAR]], 1
; CHECK-NEXT: br i1 [[T84]], label [[BB17:%.*]], label [[BB10]]
; CHECK: bb17:
@@ -722,9 +723,9 @@ define ptr @test32(ptr %v) {
; CHECK-LABEL: @test32(
; CHECK-NEXT: [[A:%.*]] = alloca [4 x ptr], align 16
; CHECK-NEXT: store ptr null, ptr [[A]], align 8
-; CHECK-NEXT: [[D:%.*]] = getelementptr inbounds { [16 x i8] }, ptr [[A]], i64 0, i32 0, i64 8
+; CHECK-NEXT: [[D:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 8
; CHECK-NEXT: store ptr [[V:%.*]], ptr [[D]], align 8
-; CHECK-NEXT: [[F:%.*]] = getelementptr inbounds [4 x ptr], ptr [[A]], i64 0, i64 2
+; CHECK-NEXT: [[F:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; CHECK-NEXT: [[G:%.*]] = load ptr, ptr [[F]], align 8
; CHECK-NEXT: ret ptr [[G]]
;
@@ -743,7 +744,7 @@ define ptr @test32(ptr %v) {
define ptr @test33(ptr %A) {
; CHECK-LABEL: @test33(
-; CHECK-NEXT: [[C:%.*]] = getelementptr [[STRUCT_ANON:%.*]], ptr [[A:%.*]], i64 0, i32 2
+; CHECK-NEXT: [[C:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 4
; CHECK-NEXT: ret ptr [[C]]
;
%C = getelementptr %struct.anon, ptr %A, i32 0, i32 2
@@ -752,7 +753,7 @@ define ptr @test33(ptr %A) {
define ptr addrspace(1) @test33_as1(ptr addrspace(1) %A) {
; CHECK-LABEL: @test33_as1(
-; CHECK-NEXT: [[C:%.*]] = getelementptr [[STRUCT_ANON:%.*]], ptr addrspace(1) [[A:%.*]], i16 0, i32 2
+; CHECK-NEXT: [[C:%.*]] = getelementptr i8, ptr addrspace(1) [[A:%.*]], i16 4
; CHECK-NEXT: ret ptr addrspace(1) [[C]]
;
%C = getelementptr %struct.anon, ptr addrspace(1) %A, i32 0, i32 2
@@ -761,7 +762,7 @@ define ptr addrspace(1) @test33_as1(ptr addrspace(1) %A) {
define ptr addrspace(1) @test33_array_as1(ptr addrspace(1) %A) {
; CHECK-LABEL: @test33_array_as1(
-; CHECK-NEXT: [[C:%.*]] = getelementptr [5 x i32], ptr addrspace(1) [[A:%.*]], i16 0, i16 2
+; CHECK-NEXT: [[C:%.*]] = getelementptr i8, ptr addrspace(1) [[A:%.*]], i16 8
; CHECK-NEXT: ret ptr addrspace(1) [[C]]
;
%C = getelementptr [5 x i32], ptr addrspace(1) %A, i32 0, i32 2
@@ -771,7 +772,7 @@ define ptr addrspace(1) @test33_array_as1(ptr addrspace(1) %A) {
; Make sure the GEP indices use the right pointer sized integer
define ptr addrspace(1) @test33_array_struct_as1(ptr addrspace(1) %A) {
; CHECK-LABEL: @test33_array_struct_as1(
-; CHECK-NEXT: [[C:%.*]] = getelementptr [20 x i32], ptr addrspace(1) [[A:%.*]], i16 0, i16 2
+; CHECK-NEXT: [[C:%.*]] = getelementptr i8, ptr addrspace(1) [[A:%.*]], i16 8
; CHECK-NEXT: ret ptr addrspace(1) [[C]]
;
%C = getelementptr [20 x i32], ptr addrspace(1) %A, i32 0, i32 2
@@ -781,7 +782,7 @@ define ptr addrspace(1) @test33_array_struct_as1(ptr addrspace(1) %A) {
define ptr addrspace(1) @test33_addrspacecast(ptr %A) {
; CHECK-LABEL: @test33_addrspacecast(
; CHECK-NEXT: [[B:%.*]] = addrspacecast ptr [[A:%.*]] to ptr addrspace(1)
-; CHECK-NEXT: [[C:%.*]] = getelementptr [[STRUCT_ANON:%.*]], ptr addrspace(1) [[B]], i16 0, i32 2
+; CHECK-NEXT: [[C:%.*]] = getelementptr i8, ptr addrspace(1) [[B]], i16 4
; CHECK-NEXT: ret ptr addrspace(1) [[C]]
;
%B = addrspacecast ptr %A to ptr addrspace(1)
@@ -861,7 +862,7 @@ declare void @pr10322_f3(ptr)
define void @pr10322_f1(ptr %foo) {
; CHECK-LABEL: @pr10322_f1(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [[PR10322_T:%.*]], ptr [[FOO:%.*]], i64 2
+; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, ptr [[FOO:%.*]], i64 16
; CHECK-NEXT: call void @pr10322_f2(ptr nonnull [[ARRAYIDX8]]) #[[ATTR0]]
; CHECK-NEXT: call void @pr10322_f3(ptr nonnull [[ARRAYIDX8]]) #[[ATTR0]]
; CHECK-NEXT: ret void
@@ -881,7 +882,7 @@ entry:
define void @three_gep_f(ptr %x) {
; CHECK-LABEL: @three_gep_f(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr [[THREE_GEP_T2:%.*]], ptr [[X:%.*]], i64 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 8
; CHECK-NEXT: call void @three_gep_h(ptr [[GEP1]])
; CHECK-NEXT: call void @three_gep_g(ptr [[GEP1]])
; CHECK-NEXT: ret void
@@ -901,7 +902,7 @@ declare void @three_gep_h(ptr)
define void @test39(ptr %arg, i8 %arg1) nounwind {
; CHECK-LABEL: @test39(
-; CHECK-NEXT: [[T:%.*]] = getelementptr inbounds [[STRUCT_HAM:%.*]], ptr [[ARG:%.*]], i64 0, i32 2
+; CHECK-NEXT: [[T:%.*]] = getelementptr inbounds i8, ptr [[ARG:%.*]], i64 16
; CHECK-NEXT: [[T2:%.*]] = load ptr, ptr [[T]], align 8
; CHECK-NEXT: [[T4:%.*]] = getelementptr inbounds i8, ptr [[T2]], i64 -8
; CHECK-NEXT: store i8 [[ARG1:%.*]], ptr [[T4]], align 8
@@ -1152,7 +1153,7 @@ define ptr @test48(ptr %I, i64 %C, i64 %D) {
define ptr @test49(ptr %I, i64 %C) {
; CHECK-LABEL: @test49(
-; CHECK-NEXT: [[B:%.*]] = getelementptr i32, ptr [[I:%.*]], i64 -1
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 -4
; CHECK-NEXT: ret ptr [[B]]
;
%notC = xor i64 -1, %C
@@ -1219,7 +1220,7 @@ define ptr @test_nzgep_zgep(ptr %base, i64 %idx) {
define ptr @test_gep_inbounds_of_gep(ptr %base) {
; CHECK-LABEL: @test_gep_inbounds_of_gep(
-; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 8
+; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 32
; CHECK-NEXT: ret ptr [[PTR2]]
;
%ptr1 = getelementptr i32, ptr %base, i64 4
@@ -1308,7 +1309,7 @@ define i32 @test_gep_bitcast_malloc(ptr %a) {
; CHECK-LABEL: @test_gep_bitcast_malloc(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[CALL:%.*]] = call noalias dereferenceable_or_null(16) ptr @malloc(i64 16)
-; CHECK-NEXT: [[G3:%.*]] = getelementptr [[STRUCT_A:%.*]], ptr [[CALL]], i64 0, i32 2
+; CHECK-NEXT: [[G3:%.*]] = getelementptr i8, ptr [[CALL]], i64 12
; CHECK-NEXT: [[A_C:%.*]] = load i32, ptr [[G3]], align 4
; CHECK-NEXT: ret i32 [[A_C]]
;
@@ -1321,9 +1322,9 @@ entry:
define ptr @gep_of_gep_multiuse_const_and_const(ptr %p, i64 %idx) {
; CHECK-LABEL: @gep_of_gep_multiuse_const_and_const(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr { i32, i32 }, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 8
; CHECK-NEXT: call void @use(ptr [[GEP1]])
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr { i32, i32 }, ptr [[P]], i64 1, i32 1
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[P]], i64 12
; CHECK-NEXT: ret ptr [[GEP2]]
;
%gep1 = getelementptr { i32, i32 }, ptr %p, i64 1
@@ -1336,7 +1337,7 @@ define ptr @gep_of_gep_multiuse_var_and_const(ptr %p, i64 %idx) {
; CHECK-LABEL: @gep_of_gep_multiuse_var_and_const(
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr { i32, i32 }, ptr [[P:%.*]], i64 [[IDX:%.*]]
; CHECK-NEXT: call void @use(ptr [[GEP1]])
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr { i32, i32 }, ptr [[P]], i64 [[IDX]], i32 1
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[GEP1]], i64 4
; CHECK-NEXT: ret ptr [[GEP2]]
;
%gep1 = getelementptr { i32, i32 }, ptr %p, i64 %idx
diff --git a/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll b/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll
index 330825e1055eaa..491f2146716753 100644
--- a/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll
@@ -7,7 +7,7 @@ declare i32 @test58_d(i64 )
define i1 @test59(ptr %foo) {
; CHECK-LABEL: @test59(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[FOO:%.*]], i32 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[FOO:%.*]], i32 8
; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[GEP1]] to i40
; CHECK-NEXT: [[USE:%.*]] = zext i40 [[TMP1]] to i64
; CHECK-NEXT: [[CALL:%.*]] = call i32 @test58_d(i64 [[USE]])
@@ -23,7 +23,7 @@ define i1 @test59(ptr %foo) {
define i1 @test59_as1(ptr addrspace(1) %foo) {
; CHECK-LABEL: @test59_as1(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[FOO:%.*]], i16 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[FOO:%.*]], i16 8
; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr addrspace(1) [[GEP1]] to i16
; CHECK-NEXT: [[USE:%.*]] = zext i16 [[TMP1]] to i64
; CHECK-NEXT: [[CALL:%.*]] = call i32 @test58_d(i64 [[USE]])
diff --git a/llvm/test/Transforms/InstCombine/icmp-gep.ll b/llvm/test/Transforms/InstCombine/icmp-gep.ll
index 99c784d15eb30a..2a9dbe8f20dd35 100644
--- a/llvm/test/Transforms/InstCombine/icmp-gep.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-gep.ll
@@ -243,7 +243,7 @@ declare i32 @test58_d(i64)
define i1 @test59(ptr %foo) {
; CHECK-LABEL: @test59(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[FOO:%.*]], i64 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[FOO:%.*]], i64 8
; CHECK-NEXT: [[USE:%.*]] = ptrtoint ptr [[GEP1]] to i64
; CHECK-NEXT: [[CALL:%.*]] = call i32 @test58_d(i64 [[USE]])
; CHECK-NEXT: ret i1 true
@@ -258,7 +258,7 @@ define i1 @test59(ptr %foo) {
define i1 @test59_as1(ptr addrspace(1) %foo) {
; CHECK-LABEL: @test59_as1(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[FOO:%.*]], i16 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[FOO:%.*]], i16 8
; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr addrspace(1) [[GEP1]] to i16
; CHECK-NEXT: [[USE:%.*]] = zext i16 [[TMP1]] to i64
; CHECK-NEXT: [[CALL:%.*]] = call i32 @test58_d(i64 [[USE]])
@@ -416,7 +416,7 @@ define i1 @test60_extra_use(ptr %foo, i64 %i, i64 %j) {
define i1 @test60_extra_use_const_operands_inbounds(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test60_extra_use_const_operands_inbounds(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[FOO:%.*]], i64 1
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[FOO:%.*]], i64 4
; CHECK-NEXT: call void @use(ptr nonnull [[GEP1]])
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[J:%.*]], 2
; CHECK-NEXT: ret i1 [[CMP]]
@@ -430,7 +430,7 @@ define i1 @test60_extra_use_const_operands_inbounds(ptr %foo, i64 %i, i64 %j) {
define i1 @test60_extra_use_const_operands_no_inbounds(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test60_extra_use_const_operands_no_inbounds(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[FOO:%.*]], i64 1
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[FOO:%.*]], i64 4
; CHECK-NEXT: call void @use(ptr [[GEP1]])
; CHECK-NEXT: [[GEP2_IDX_MASK:%.*]] = and i64 [[J:%.*]], 9223372036854775807
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[GEP2_IDX_MASK]], 2
diff --git a/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll b/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll
index 61cf727fb6c8c8..2b5b3fce705354 100644
--- a/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll
+++ b/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll
@@ -73,8 +73,8 @@ define ptr @test3_no_inbounds1(ptr %A, i32 %Offset) {
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb:
; CHECK-NEXT: [[RHS:%.*]] = phi ptr [ [[RHS_NEXT:%.*]], [[BB]] ], [ [[TMP]], [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[LHS:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 100
-; CHECK-NEXT: [[RHS_NEXT]] = getelementptr inbounds i32, ptr [[RHS]], i32 1
+; CHECK-NEXT: [[LHS:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 400
+; CHECK-NEXT: [[RHS_NEXT]] = getelementptr inbounds i8, ptr [[RHS]], i32 4
; CHECK-NEXT: [[COND:%.*]] = icmp ult ptr [[LHS]], [[RHS]]
; CHECK-NEXT: br i1 [[COND]], label [[BB2:%.*]], label [[BB]]
; CHECK: bb2:
@@ -102,8 +102,8 @@ define ptr @test3_no_inbounds2(ptr %A, i32 %Offset) {
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb:
; CHECK-NEXT: [[RHS:%.*]] = phi ptr [ [[RHS_NEXT:%.*]], [[BB]] ], [ [[TMP]], [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[LHS:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 100
-; CHECK-NEXT: [[RHS_NEXT]] = getelementptr i32, ptr [[RHS]], i32 1
+; CHECK-NEXT: [[LHS:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 400
+; CHECK-NEXT: [[RHS_NEXT]] = getelementptr i8, ptr [[RHS]], i32 4
; CHECK-NEXT: [[COND:%.*]] = icmp ult ptr [[LHS]], [[RHS]]
; CHECK-NEXT: br i1 [[COND]], label [[BB2:%.*]], label [[BB]]
; CHECK: bb2:
@@ -131,8 +131,8 @@ define ptr @test3_no_inbounds3(ptr %A, i32 %Offset) {
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb:
; CHECK-NEXT: [[RHS:%.*]] = phi ptr [ [[RHS_NEXT:%.*]], [[BB]] ], [ [[TMP]], [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[LHS:%.*]] = getelementptr i32, ptr [[A]], i32 100
-; CHECK-NEXT: [[RHS_NEXT]] = getelementptr inbounds i32, ptr [[RHS]], i32 1
+; CHECK-NEXT: [[LHS:%.*]] = getelementptr i8, ptr [[A]], i32 400
+; CHECK-NEXT: [[RHS_NEXT]] = getelementptr inbounds i8, ptr [[RHS]], i32 4
; CHECK-NEXT: [[COND:%.*]] = icmp ult ptr [[LHS]], [[RHS]]
; CHECK-NEXT: br i1 [[COND]], label [[BB2:%.*]], label [[BB]]
; CHECK: bb2:
@@ -318,7 +318,7 @@ define i1 @test8(ptr %in, i64 %offset) {
; CHECK-NEXT: [[GEPI8:%.*]] = getelementptr inbounds i8, ptr [[CASTI8]], i32 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = trunc i64 [[LD]] to i32
; CHECK-NEXT: [[PTRCAST:%.*]] = inttoptr i32 [[TMP2]] to ptr
-; CHECK-NEXT: [[GEPI32:%.*]] = getelementptr inbounds ptr, ptr [[PTRCAST]], i32 1
+; CHECK-NEXT: [[GEPI32:%.*]] = getelementptr inbounds i8, ptr [[PTRCAST]], i32 4
; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[GEPI32]], [[GEPI8]]
; CHECK-NEXT: ret i1 [[CMP]]
;
@@ -335,7 +335,7 @@ entry:
define void @test_zero_offset_cycle(ptr %arg) {
; CHECK-LABEL: @test_zero_offset_cycle(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds { i64, i64 }, ptr [[ARG:%.*]], i32 0, i32 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[ARG:%.*]], i32 8
; CHECK-NEXT: [[GEP_INT:%.*]] = ptrtoint ptr [[GEP]] to i32
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
diff --git a/llvm/test/Transforms/InstCombine/intptr1.ll b/llvm/test/Transforms/InstCombine/intptr1.ll
index fcd97b0cf8b0b2..2bf255341f4bbf 100644
--- a/llvm/test/Transforms/InstCombine/intptr1.ll
+++ b/llvm/test/Transforms/InstCombine/intptr1.ll
@@ -16,8 +16,8 @@ define void @test1(ptr %a, ptr readnone %a_end, ptr %b.i64) {
; CHECK-NEXT: [[I1:%.*]] = load float, ptr [[B_ADDR_02_PTR]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[I1]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[ADD]] = getelementptr inbounds float, ptr [[B_ADDR_02_PTR]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[ADD]] = getelementptr inbounds i8, ptr [[B_ADDR_02_PTR]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
; CHECK: for.end:
@@ -68,8 +68,8 @@ define void @test1_neg(ptr %a, ptr readnone %a_end, ptr %b.i64) {
; CHECK-NEXT: [[I1:%.*]] = load float, ptr [[A]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[I1]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[ADD]] = getelementptr inbounds float, ptr [[A]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[ADD]] = getelementptr inbounds i8, ptr [[A]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
; CHECK: for.end:
@@ -122,8 +122,8 @@ define void @test2(ptr %a, ptr readnone %a_end, ptr %b.float) {
; CHECK-NEXT: [[I1:%.*]] = load float, ptr [[B_ADDR_02_PTR]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[I1]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[ADD]] = getelementptr inbounds float, ptr [[B_ADDR_02_PTR]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[ADD]] = getelementptr inbounds i8, ptr [[B_ADDR_02_PTR]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
; CHECK: for.end:
@@ -172,8 +172,8 @@ define void @test3(ptr %a, ptr readnone %a_end, ptr %b.i8p) {
; CHECK-NEXT: [[I1:%.*]] = load float, ptr [[B_ADDR_02_PTR]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[I1]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[ADD]] = getelementptr inbounds float, ptr [[B_ADDR_02_PTR]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[ADD]] = getelementptr inbounds i8, ptr [[B_ADDR_02_PTR]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
; CHECK: for.end:
@@ -221,8 +221,8 @@ define void @test4(ptr %a, ptr readnone %a_end, ptr %b.float) {
; CHECK-NEXT: [[I1:%.*]] = load float, ptr [[B_ADDR_02_IN]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[I1]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[ADD]] = getelementptr inbounds float, ptr [[B_ADDR_02_IN]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[ADD]] = getelementptr inbounds i8, ptr [[B_ADDR_02_IN]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
; CHECK: for.end:
diff --git a/llvm/test/Transforms/InstCombine/intptr2.ll b/llvm/test/Transforms/InstCombine/intptr2.ll
index 6ace82bc8e09e3..c256171142b915 100644
--- a/llvm/test/Transforms/InstCombine/intptr2.ll
+++ b/llvm/test/Transforms/InstCombine/intptr2.ll
@@ -15,8 +15,8 @@ define void @test1(ptr %a, ptr readnone %a_end, ptr %b.i) {
; CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[B_ADDR_02_IN]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[TMP1]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[ADD]] = getelementptr inbounds float, ptr [[B_ADDR_02_IN]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[ADD]] = getelementptr inbounds i8, ptr [[B_ADDR_02_IN]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
; CHECK: for.end:
diff --git a/llvm/test/Transforms/InstCombine/intptr3.ll b/llvm/test/Transforms/InstCombine/intptr3.ll
index 0937d6d377b1e8..3bb3c7dcb4e1a0 100644
--- a/llvm/test/Transforms/InstCombine/intptr3.ll
+++ b/llvm/test/Transforms/InstCombine/intptr3.ll
@@ -15,8 +15,8 @@ define void @test(ptr %a, ptr readnone %a_end, i64 %b) unnamed_addr {
; CHECK-NEXT: [[L:%.*]] = load float, ptr [[B_ADDR_FLOAT]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[L]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[B_ADDR_FLOAT_INC]] = getelementptr inbounds float, ptr [[B_ADDR_FLOAT]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[B_ADDR_FLOAT_INC]] = getelementptr inbounds i8, ptr [[B_ADDR_FLOAT]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]]
; CHECK: for.end:
diff --git a/llvm/test/Transforms/InstCombine/intptr4.ll b/llvm/test/Transforms/InstCombine/intptr4.ll
index 612fef0c1837d7..5998501d82f4fa 100644
--- a/llvm/test/Transforms/InstCombine/intptr4.ll
+++ b/llvm/test/Transforms/InstCombine/intptr4.ll
@@ -22,8 +22,8 @@ define void @test(ptr %a, ptr readnone %a_end, i64 %b, ptr %bf) unnamed_addr {
; CHECK-NEXT: [[L:%.*]] = load float, ptr [[B_ADDR_FLOAT]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[L]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[B_ADDR_FLOAT_INC]] = getelementptr inbounds float, ptr [[B_ADDR_I64_PTR]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[B_ADDR_FLOAT_INC]] = getelementptr inbounds i8, ptr [[B_ADDR_I64_PTR]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK: for.end:
diff --git a/llvm/test/Transforms/InstCombine/intptr5.ll b/llvm/test/Transforms/InstCombine/intptr5.ll
index 1f493d6a3a1359..f992886208f671 100644
--- a/llvm/test/Transforms/InstCombine/intptr5.ll
+++ b/llvm/test/Transforms/InstCombine/intptr5.ll
@@ -24,8 +24,8 @@ define void @test(ptr %a, ptr readnone %a_end, i64 %b, ptr %bf) unnamed_addr {
; CHECK-NEXT: [[L:%.*]] = load float, ptr [[B_ADDR_FLOAT]], align 4
; CHECK-NEXT: [[MUL_I:%.*]] = fmul float [[L]], 4.200000e+01
; CHECK-NEXT: store float [[MUL_I]], ptr [[A_ADDR_03]], align 4
-; CHECK-NEXT: [[B_ADDR_FLOAT_INC]] = getelementptr inbounds float, ptr [[B_ADDR_I64_PTR]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_03]], i64 1
+; CHECK-NEXT: [[B_ADDR_FLOAT_INC]] = getelementptr inbounds i8, ptr [[B_ADDR_I64_PTR]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_03]], i64 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[INCDEC_PTR]], [[A_END]]
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK: for.end:
diff --git a/llvm/test/Transforms/InstCombine/intptr7.ll b/llvm/test/Transforms/InstCombine/intptr7.ll
index 78e0b59ee0e238..18fdd208e99a02 100644
--- a/llvm/test/Transforms/InstCombine/intptr7.ll
+++ b/llvm/test/Transforms/InstCombine/intptr7.ll
@@ -6,7 +6,7 @@ define void @matching_phi(i64 %a, ptr %b, i1 %cond) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 [[COND:%.*]], label [[BB2:%.*]], label [[BB1:%.*]]
; CHECK: bb1:
-; CHECK-NEXT: [[ADDB:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 2
+; CHECK-NEXT: [[ADDB:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 8
; CHECK-NEXT: br label [[BB3:%.*]]
; CHECK: bb2:
; CHECK-NEXT: [[ADD_INT:%.*]] = add i64 [[A:%.*]], 1
@@ -48,7 +48,7 @@ define void @no_matching_phi(i64 %a, ptr %b, i1 %cond) {
; CHECK-LABEL: @no_matching_phi(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[ADD_INT:%.*]] = add i64 [[A:%.*]], 1
-; CHECK-NEXT: [[ADDB:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 2
+; CHECK-NEXT: [[ADDB:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 8
; CHECK-NEXT: br i1 [[COND:%.*]], label [[B:%.*]], label [[A:%.*]]
; CHECK: A:
; CHECK-NEXT: [[TMP0:%.*]] = inttoptr i64 [[ADD_INT]] to ptr
diff --git a/llvm/test/Transforms/InstCombine/load-store-forward.ll b/llvm/test/Transforms/InstCombine/load-store-forward.ll
index 6be5f6ed42d538..aa593b95b36018 100644
--- a/llvm/test/Transforms/InstCombine/load-store-forward.ll
+++ b/llvm/test/Transforms/InstCombine/load-store-forward.ll
@@ -63,7 +63,7 @@ define i32 @vec_store_load_first_constexpr(ptr %p) {
define i32 @vec_store_load_second(ptr %p) {
; CHECK-LABEL: @vec_store_load_second(
; CHECK-NEXT: store <2 x i32> <i32 1, i32 2>, ptr [[P:%.*]], align 8
-; CHECK-NEXT: [[P3:%.*]] = getelementptr i32, ptr [[P]], i64 1
+; CHECK-NEXT: [[P3:%.*]] = getelementptr i8, ptr [[P]], i64 4
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr [[P3]], align 4
; CHECK-NEXT: ret i32 [[LOAD]]
;
diff --git a/llvm/test/Transforms/InstCombine/load.ll b/llvm/test/Transforms/InstCombine/load.ll
index fb6c50d5fd8fca..7d53c8ee35684b 100644
--- a/llvm/test/Transforms/InstCombine/load.ll
+++ b/llvm/test/Transforms/InstCombine/load.ll
@@ -136,7 +136,7 @@ C: ; preds = %F, %T
define double @test11(ptr %p) {
; CHECK-LABEL: @test11(
-; CHECK-NEXT: [[T0:%.*]] = getelementptr double, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[T0:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 8
; CHECK-NEXT: store double 2.000000e+00, ptr [[T0]], align 8
; CHECK-NEXT: ret double 2.000000e+00
;
diff --git a/llvm/test/Transforms/InstCombine/loadstore-metadata.ll b/llvm/test/Transforms/InstCombine/loadstore-metadata.ll
index e7ad7600a3f5fb..b9a96937e57c77 100644
--- a/llvm/test/Transforms/InstCombine/loadstore-metadata.ll
+++ b/llvm/test/Transforms/InstCombine/loadstore-metadata.ll
@@ -152,7 +152,7 @@ define void @test_load_cast_combine_nonnull(ptr %ptr) {
; CHECK-LABEL: @test_load_cast_combine_nonnull(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P:%.*]] = load ptr, ptr [[PTR:%.*]], align 8, !nonnull !6
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr ptr, ptr [[PTR]], i64 42
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[PTR]], i64 336
; CHECK-NEXT: store ptr [[P]], ptr [[GEP]], align 8
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memchr-5.ll b/llvm/test/Transforms/InstCombine/memchr-5.ll
index 40c6b28a30136d..010c83719b8b61 100644
--- a/llvm/test/Transforms/InstCombine/memchr-5.ll
+++ b/llvm/test/Transforms/InstCombine/memchr-5.ll
@@ -16,37 +16,37 @@ declare ptr @memchr(ptr, i32, i64)
define void @fold_memchr_a(ptr %pcmp) {
; BE-LABEL: @fold_memchr_a(
; BE-NEXT: store i64 0, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; BE-NEXT: store i64 1, ptr [[PSTOR1]], align 4
-; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; BE-NEXT: store i64 2, ptr [[PSTOR2]], align 4
-; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; BE-NEXT: store i64 3, ptr [[PSTOR3]], align 4
-; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; BE-NEXT: store i64 13, ptr [[PSTOR4]], align 4
-; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i64, ptr [[PCMP]], i64 6
+; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 48
; BE-NEXT: store i64 14, ptr [[PSTOR6]], align 4
-; BE-NEXT: [[PSTOR7:%.*]] = getelementptr i64, ptr [[PCMP]], i64 7
+; BE-NEXT: [[PSTOR7:%.*]] = getelementptr i8, ptr [[PCMP]], i64 56
; BE-NEXT: store i64 15, ptr [[PSTOR7]], align 4
-; BE-NEXT: [[PSTOR8:%.*]] = getelementptr i64, ptr [[PCMP]], i64 8
+; BE-NEXT: [[PSTOR8:%.*]] = getelementptr i8, ptr [[PCMP]], i64 64
; BE-NEXT: store i64 0, ptr [[PSTOR8]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memchr_a(
; LE-NEXT: store i64 3, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; LE-NEXT: store i64 2, ptr [[PSTOR1]], align 4
-; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; LE-NEXT: store i64 1, ptr [[PSTOR2]], align 4
-; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; LE-NEXT: store i64 0, ptr [[PSTOR3]], align 4
-; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; LE-NEXT: store i64 14, ptr [[PSTOR4]], align 4
-; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i64, ptr [[PCMP]], i64 6
+; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 48
; LE-NEXT: store i64 13, ptr [[PSTOR6]], align 4
-; LE-NEXT: [[PSTOR7:%.*]] = getelementptr i64, ptr [[PCMP]], i64 7
+; LE-NEXT: [[PSTOR7:%.*]] = getelementptr i8, ptr [[PCMP]], i64 56
; LE-NEXT: store i64 12, ptr [[PSTOR7]], align 4
-; LE-NEXT: [[PSTOR8:%.*]] = getelementptr i64, ptr [[PCMP]], i64 8
+; LE-NEXT: [[PSTOR8:%.*]] = getelementptr i8, ptr [[PCMP]], i64 64
; LE-NEXT: store i64 0, ptr [[PSTOR8]], align 4
; LE-NEXT: ret void
;
@@ -123,29 +123,29 @@ define void @fold_memchr_a(ptr %pcmp) {
define void @fold_memchr_a_p1(ptr %pcmp) {
; BE-LABEL: @fold_memchr_a_p1(
; BE-NEXT: store i64 0, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; BE-NEXT: store i64 1, ptr [[PSTOR1]], align 4
-; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; BE-NEXT: store i64 2, ptr [[PSTOR2]], align 4
-; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; BE-NEXT: store i64 3, ptr [[PSTOR3]], align 4
-; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; BE-NEXT: store i64 0, ptr [[PSTOR4]], align 4
-; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i64, ptr [[PCMP]], i64 5
+; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 40
; BE-NEXT: store i64 0, ptr [[PSTOR5]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memchr_a_p1(
; LE-NEXT: store i64 3, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; LE-NEXT: store i64 2, ptr [[PSTOR1]], align 4
-; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; LE-NEXT: store i64 1, ptr [[PSTOR2]], align 4
-; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; LE-NEXT: store i64 0, ptr [[PSTOR3]], align 4
-; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; LE-NEXT: store i64 0, ptr [[PSTOR4]], align 4
-; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i64, ptr [[PCMP]], i64 5
+; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 40
; LE-NEXT: store i64 0, ptr [[PSTOR5]], align 4
; LE-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memchr-9.ll b/llvm/test/Transforms/InstCombine/memchr-9.ll
index e2c2e6839817cd..fe80c282eed54b 100644
--- a/llvm/test/Transforms/InstCombine/memchr-9.ll
+++ b/llvm/test/Transforms/InstCombine/memchr-9.ll
@@ -19,23 +19,23 @@ declare ptr @memchr(ptr, i32, i64)
define void @fold_memchr_A_pIb_cst_cst(ptr %pchr) {
; CHECK-LABEL: @fold_memchr_A_pIb_cst_cst(
; CHECK-NEXT: store ptr @a, ptr [[PCHR:%.*]], align 8
-; CHECK-NEXT: [[PST_0_1_1:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 1
+; CHECK-NEXT: [[PST_0_1_1:%.*]] = getelementptr i8, ptr [[PCHR]], i64 8
; CHECK-NEXT: store ptr null, ptr [[PST_0_1_1]], align 8
-; CHECK-NEXT: [[PST_0_4_4:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 2
+; CHECK-NEXT: [[PST_0_4_4:%.*]] = getelementptr i8, ptr [[PCHR]], i64 16
; CHECK-NEXT: store ptr null, ptr [[PST_0_4_4]], align 8
-; CHECK-NEXT: [[PST_1_0_1:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 3
+; CHECK-NEXT: [[PST_1_0_1:%.*]] = getelementptr i8, ptr [[PCHR]], i64 24
; CHECK-NEXT: store ptr getelementptr (i8, ptr @a, i64 1), ptr [[PST_1_0_1]], align 8
-; CHECK-NEXT: [[PST_1_0_3:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 4
+; CHECK-NEXT: [[PST_1_0_3:%.*]] = getelementptr i8, ptr [[PCHR]], i64 32
; CHECK-NEXT: store ptr getelementptr (i8, ptr @a, i64 1), ptr [[PST_1_0_3]], align 8
-; CHECK-NEXT: [[PST_1_1_1:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 5
+; CHECK-NEXT: [[PST_1_1_1:%.*]] = getelementptr i8, ptr [[PCHR]], i64 40
; CHECK-NEXT: store ptr null, ptr [[PST_1_1_1]], align 8
-; CHECK-NEXT: [[PST_1_1_2:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 6
+; CHECK-NEXT: [[PST_1_1_2:%.*]] = getelementptr i8, ptr [[PCHR]], i64 48
; CHECK-NEXT: store ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 0, i64 1), ptr [[PST_1_1_2]], align 8
-; CHECK-NEXT: [[PST_1_3_3:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 7
+; CHECK-NEXT: [[PST_1_3_3:%.*]] = getelementptr i8, ptr [[PCHR]], i64 56
; CHECK-NEXT: store ptr null, ptr [[PST_1_3_3]], align 8
-; CHECK-NEXT: [[PST_1_3_4:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 8
+; CHECK-NEXT: [[PST_1_3_4:%.*]] = getelementptr i8, ptr [[PCHR]], i64 64
; CHECK-NEXT: store ptr null, ptr [[PST_1_3_4]], align 8
-; CHECK-NEXT: [[PST_1_3_6:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 10
+; CHECK-NEXT: [[PST_1_3_6:%.*]] = getelementptr i8, ptr [[PCHR]], i64 80
; CHECK-NEXT: store ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 1, i64 1), ptr [[PST_1_3_6]], align 8
; CHECK-NEXT: ret void
;
@@ -108,45 +108,45 @@ define void @fold_memchr_A_pIb_cst_N(i64 %N, ptr %pchr) {
; CHECK-NEXT: [[MEMCHR_CMP:%.*]] = icmp eq i64 [[N:%.*]], 0
; CHECK-NEXT: [[CHR_0_0_N:%.*]] = select i1 [[MEMCHR_CMP]], ptr null, ptr @a
; CHECK-NEXT: store ptr [[CHR_0_0_N]], ptr [[PCHR:%.*]], align 8
-; CHECK-NEXT: [[PST_0_1_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 1
+; CHECK-NEXT: [[PST_0_1_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 8
; CHECK-NEXT: [[MEMCHR_CMP1:%.*]] = icmp ult i64 [[N]], 3
; CHECK-NEXT: [[CHR_0_1_N:%.*]] = select i1 [[MEMCHR_CMP1]], ptr null, ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 0, i64 1)
; CHECK-NEXT: store ptr [[CHR_0_1_N]], ptr [[PST_0_1_N]], align 8
-; CHECK-NEXT: [[PST_0_4_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 2
+; CHECK-NEXT: [[PST_0_4_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 16
; CHECK-NEXT: store ptr null, ptr [[PST_0_4_N]], align 8
-; CHECK-NEXT: [[PST_1_0_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 3
+; CHECK-NEXT: [[PST_1_0_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 24
; CHECK-NEXT: [[MEMCHR_CMP2:%.*]] = icmp eq i64 [[N]], 0
; CHECK-NEXT: [[CHR_1_0_N:%.*]] = select i1 [[MEMCHR_CMP2]], ptr null, ptr getelementptr (i8, ptr @a, i64 1)
; CHECK-NEXT: store ptr [[CHR_1_0_N]], ptr [[PST_1_0_N]], align 8
-; CHECK-NEXT: [[PST_1_1_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 4
+; CHECK-NEXT: [[PST_1_1_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 32
; CHECK-NEXT: [[MEMCHR_CMP3:%.*]] = icmp ult i64 [[N]], 2
; CHECK-NEXT: [[CHR_1_1_N:%.*]] = select i1 [[MEMCHR_CMP3]], ptr null, ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 0, i64 1)
; CHECK-NEXT: store ptr [[CHR_1_1_N]], ptr [[PST_1_1_N]], align 8
-; CHECK-NEXT: [[PST_1_2_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 5
+; CHECK-NEXT: [[PST_1_2_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 40
; CHECK-NEXT: [[MEMCHR_CMP4:%.*]] = icmp ult i64 [[N]], 4
; CHECK-NEXT: [[CHR_1_2_N:%.*]] = select i1 [[MEMCHR_CMP4]], ptr null, ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 1, i64 0)
; CHECK-NEXT: store ptr [[CHR_1_2_N]], ptr [[PST_1_2_N]], align 8
-; CHECK-NEXT: [[PST_1_3_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 6
+; CHECK-NEXT: [[PST_1_3_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 48
; CHECK-NEXT: [[MEMCHR_CMP5:%.*]] = icmp ult i64 [[N]], 6
; CHECK-NEXT: [[CHR_1_3_N:%.*]] = select i1 [[MEMCHR_CMP5]], ptr null, ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 1, i64 1)
; CHECK-NEXT: store ptr [[CHR_1_3_N]], ptr [[PST_1_3_N]], align 8
-; CHECK-NEXT: [[PST_1_4_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 7
+; CHECK-NEXT: [[PST_1_4_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 56
; CHECK-NEXT: store ptr null, ptr [[PST_1_4_N]], align 8
-; CHECK-NEXT: [[PST_2_0_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 8
+; CHECK-NEXT: [[PST_2_0_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 64
; CHECK-NEXT: store ptr null, ptr [[PST_2_0_N]], align 8
-; CHECK-NEXT: [[PST_2_1_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 9
+; CHECK-NEXT: [[PST_2_1_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 72
; CHECK-NEXT: [[MEMCHR_CMP6:%.*]] = icmp eq i64 [[N]], 0
; CHECK-NEXT: [[CHR_2_1_N:%.*]] = select i1 [[MEMCHR_CMP6]], ptr null, ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 0, i64 1)
; CHECK-NEXT: store ptr [[CHR_2_1_N]], ptr [[PST_2_1_N]], align 8
-; CHECK-NEXT: [[PST_2_2_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 10
+; CHECK-NEXT: [[PST_2_2_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 80
; CHECK-NEXT: [[MEMCHR_CMP7:%.*]] = icmp ult i64 [[N]], 3
; CHECK-NEXT: [[CHR_2_2_N:%.*]] = select i1 [[MEMCHR_CMP7]], ptr null, ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 1, i64 0)
; CHECK-NEXT: store ptr [[CHR_2_2_N]], ptr [[PST_2_2_N]], align 8
-; CHECK-NEXT: [[PST_2_3_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 11
+; CHECK-NEXT: [[PST_2_3_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 88
; CHECK-NEXT: [[MEMCHR_CMP8:%.*]] = icmp ult i64 [[N]], 5
; CHECK-NEXT: [[CHR_2_3_N:%.*]] = select i1 [[MEMCHR_CMP8]], ptr null, ptr getelementptr inbounds ([1 x %struct.A], ptr @a, i64 0, i64 0, i32 1, i64 1)
; CHECK-NEXT: store ptr [[CHR_2_3_N]], ptr [[PST_2_3_N]], align 8
-; CHECK-NEXT: [[PST_2_4_N:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 12
+; CHECK-NEXT: [[PST_2_4_N:%.*]] = getelementptr i8, ptr [[PCHR]], i64 96
; CHECK-NEXT: store ptr null, ptr [[PST_2_4_N]], align 8
; CHECK-NEXT: ret void
;
@@ -232,10 +232,10 @@ define void @call_memchr_A_pIb_xs_cst(ptr %pchr) {
; CHECK-LABEL: @call_memchr_A_pIb_xs_cst(
; CHECK-NEXT: [[CHR_1_0_0_2:%.*]] = call ptr @memchr(ptr noundef nonnull dereferenceable(1) getelementptr inbounds ([1 x %struct.A], ptr @a, i64 1, i64 0), i32 0, i64 2)
; CHECK-NEXT: store ptr [[CHR_1_0_0_2]], ptr [[PCHR:%.*]], align 8
-; CHECK-NEXT: [[PST_1_0_1_2:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 1
+; CHECK-NEXT: [[PST_1_0_1_2:%.*]] = getelementptr i8, ptr [[PCHR]], i64 8
; CHECK-NEXT: [[CHR_1_0_1_2:%.*]] = call ptr @memchr(ptr noundef nonnull dereferenceable(1) getelementptr inbounds ([1 x %struct.A], ptr @a, i64 1, i64 0), i32 0, i64 2)
; CHECK-NEXT: store ptr [[CHR_1_0_1_2]], ptr [[PST_1_0_1_2]], align 8
-; CHECK-NEXT: [[PST_0_0_8_2:%.*]] = getelementptr ptr, ptr [[PCHR]], i64 2
+; CHECK-NEXT: [[PST_0_0_8_2:%.*]] = getelementptr i8, ptr [[PCHR]], i64 16
; CHECK-NEXT: [[CHR_0_0_8_2:%.*]] = call ptr @memchr(ptr noundef nonnull dereferenceable(1) getelementptr inbounds ([1 x %struct.A], ptr @a, i64 1, i64 0, i32 0, i64 0), i32 0, i64 2)
; CHECK-NEXT: store ptr [[CHR_0_0_8_2]], ptr [[PST_0_0_8_2]], align 8
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/InstCombine/memcmp-3.ll b/llvm/test/Transforms/InstCombine/memcmp-3.ll
index fd4d54ebd7c1c6..ae7c2fc3304811 100644
--- a/llvm/test/Transforms/InstCombine/memcmp-3.ll
+++ b/llvm/test/Transforms/InstCombine/memcmp-3.ll
@@ -18,41 +18,41 @@ declare i32 @memcmp(ptr, ptr, i64)
define void @fold_memcmp_ia16a_i8a(ptr %pcmp) {
; BE-LABEL: @fold_memcmp_ia16a_i8a(
; BE-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; BE-NEXT: store i32 0, ptr [[PSTOR1]], align 4
-; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; BE-NEXT: store i32 0, ptr [[PSTOR2]], align 4
-; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; BE-NEXT: store i32 0, ptr [[PSTOR3]], align 4
-; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; BE-NEXT: store i32 0, ptr [[PSTOR4]], align 4
-; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; BE-NEXT: store i32 0, ptr [[PSTOR5]], align 4
-; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; BE-NEXT: store i32 0, ptr [[PSTOR6]], align 4
-; BE-NEXT: [[PSTOR7:%.*]] = getelementptr i32, ptr [[PCMP]], i64 7
+; BE-NEXT: [[PSTOR7:%.*]] = getelementptr i8, ptr [[PCMP]], i64 28
; BE-NEXT: store i32 0, ptr [[PSTOR7]], align 4
-; BE-NEXT: [[PSTOR8:%.*]] = getelementptr i32, ptr [[PCMP]], i64 8
+; BE-NEXT: [[PSTOR8:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; BE-NEXT: store i32 1, ptr [[PSTOR8]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memcmp_ia16a_i8a(
; LE-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; LE-NEXT: store i32 1, ptr [[PSTOR1]], align 4
-; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; LE-NEXT: store i32 1, ptr [[PSTOR2]], align 4
-; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; LE-NEXT: store i32 1, ptr [[PSTOR3]], align 4
-; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; LE-NEXT: store i32 1, ptr [[PSTOR4]], align 4
-; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; LE-NEXT: store i32 1, ptr [[PSTOR5]], align 4
-; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; LE-NEXT: store i32 1, ptr [[PSTOR6]], align 4
-; LE-NEXT: [[PSTOR7:%.*]] = getelementptr i32, ptr [[PCMP]], i64 7
+; LE-NEXT: [[PSTOR7:%.*]] = getelementptr i8, ptr [[PCMP]], i64 28
; LE-NEXT: store i32 1, ptr [[PSTOR7]], align 4
-; LE-NEXT: [[PSTOR8:%.*]] = getelementptr i32, ptr [[PCMP]], i64 8
+; LE-NEXT: [[PSTOR8:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; LE-NEXT: store i32 1, ptr [[PSTOR8]], align 4
; LE-NEXT: ret void
;
@@ -101,33 +101,33 @@ define void @fold_memcmp_ia16a_i8a(ptr %pcmp) {
define void @fold_memcmp_ia16a_p1_i8a_p1(ptr %pcmp) {
; BE-LABEL: @fold_memcmp_ia16a_p1_i8a_p1(
; BE-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; BE-NEXT: store i32 1, ptr [[PSTOR1]], align 4
-; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; BE-NEXT: store i32 1, ptr [[PSTOR2]], align 4
-; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; BE-NEXT: store i32 1, ptr [[PSTOR3]], align 4
-; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; BE-NEXT: store i32 1, ptr [[PSTOR4]], align 4
-; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; BE-NEXT: store i32 1, ptr [[PSTOR5]], align 4
-; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; BE-NEXT: store i32 1, ptr [[PSTOR6]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memcmp_ia16a_p1_i8a_p1(
; LE-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; LE-NEXT: store i32 1, ptr [[PSTOR1]], align 4
-; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; LE-NEXT: store i32 1, ptr [[PSTOR2]], align 4
-; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; LE-NEXT: store i32 1, ptr [[PSTOR3]], align 4
-; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; LE-NEXT: store i32 1, ptr [[PSTOR4]], align 4
-; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; LE-NEXT: store i32 1, ptr [[PSTOR5]], align 4
-; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; LE-NEXT: store i32 1, ptr [[PSTOR6]], align 4
; LE-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memcmp-4.ll b/llvm/test/Transforms/InstCombine/memcmp-4.ll
index c63f305581b034..8c839bbfc47bd6 100644
--- a/llvm/test/Transforms/InstCombine/memcmp-4.ll
+++ b/llvm/test/Transforms/InstCombine/memcmp-4.ll
@@ -20,13 +20,13 @@ declare i32 @memcmp(ptr, ptr, i64)
define void @fold_memcmp_mismatch_too_big(ptr %pcmp) {
; BE-LABEL: @fold_memcmp_mismatch_too_big(
; BE-NEXT: store i32 -1, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR_CB:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR_CB:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; BE-NEXT: store i32 1, ptr [[PSTOR_CB]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memcmp_mismatch_too_big(
; LE-NEXT: store i32 -1, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR_CB:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR_CB:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; LE-NEXT: store i32 1, ptr [[PSTOR_CB]], align 4
; LE-NEXT: ret void
;
@@ -49,13 +49,13 @@ define void @fold_memcmp_mismatch_too_big(ptr %pcmp) {
define void @fold_memcmp_match_too_big(ptr %pcmp) {
; BE-LABEL: @fold_memcmp_match_too_big(
; BE-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR_AB_M1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR_AB_M1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; BE-NEXT: store i32 0, ptr [[PSTOR_AB_M1]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memcmp_match_too_big(
; LE-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR_AB_M1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR_AB_M1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; LE-NEXT: store i32 0, ptr [[PSTOR_AB_M1]], align 4
; LE-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memcmp-5.ll b/llvm/test/Transforms/InstCombine/memcmp-5.ll
index 591131cdb79076..fa97a206b45bac 100644
--- a/llvm/test/Transforms/InstCombine/memcmp-5.ll
+++ b/llvm/test/Transforms/InstCombine/memcmp-5.ll
@@ -20,21 +20,21 @@ define void @fold_memcmp_a_b_n(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ne i64 [[N:%.*]], 0
; CHECK-NEXT: [[C0_1:%.*]] = sext i1 [[TMP1]] to i32
-; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[C0_1]], ptr [[S0_1]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_2:%.*]] = sext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[C0_2]], ptr [[S0_2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_3:%.*]] = sext i1 [[TMP3]] to i32
-; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 [[C0_3]], ptr [[S0_3]], align 4
-; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 0, ptr [[S0_4]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_5:%.*]] = sext i1 [[TMP4]] to i32
-; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 [[C0_5]], ptr [[S0_5]], align 4
; CHECK-NEXT: ret void
;
@@ -107,23 +107,23 @@ define void @fold_memcmp_a_c_n(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 [[C0_0]], ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_1:%.*]] = sext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[C0_1]], ptr [[S0_1]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_2:%.*]] = sext i1 [[TMP3]] to i32
-; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[C0_2]], ptr [[S0_2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_3:%.*]] = sext i1 [[TMP4]] to i32
-; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 [[C0_3]], ptr [[S0_3]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = icmp ugt i64 [[N]], 3
; CHECK-NEXT: [[C0_4:%.*]] = sext i1 [[TMP5]] to i32
-; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 [[C0_4]], ptr [[S0_4]], align 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp ugt i64 [[N]], 3
; CHECK-NEXT: [[C0_5:%.*]] = sext i1 [[TMP6]] to i32
-; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 [[C0_5]], ptr [[S0_5]], align 4
; CHECK-NEXT: ret void
;
@@ -178,11 +178,11 @@ define void @fold_memcmp_a_d_n(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 [[C0_0]], ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_1:%.*]] = sext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[C0_1]], ptr [[S0_1]], align 4
-; CHECK-NEXT: [[S1_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[S1_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 0, ptr [[S1_1]], align 4
-; CHECK-NEXT: [[S6_6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[S6_6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 0, ptr [[S6_6]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memcmp-6.ll b/llvm/test/Transforms/InstCombine/memcmp-6.ll
index e5e304895135c4..be704bec0a8d02 100644
--- a/llvm/test/Transforms/InstCombine/memcmp-6.ll
+++ b/llvm/test/Transforms/InstCombine/memcmp-6.ll
@@ -16,11 +16,11 @@ declare i32 @memcmp(ptr, ptr, i64)
define void @fold_memcmp_cst_cst(ptr %pcmp) {
; CHECK-LABEL: @fold_memcmp_cst_cst(
; CHECK-NEXT: store i32 -1, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[SB5_A5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[SB5_A5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 1, ptr [[SB5_A5]], align 4
-; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 -1, ptr [[SA6_B6]], align 4
-; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 1, ptr [[SB6_A6]], align 4
; CHECK-NEXT: ret void
;
@@ -63,15 +63,15 @@ define void @fold_memcmp_cst_var(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 [[CA0_B0]], ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt i64 [[N]], 6
; CHECK-NEXT: [[CB0_A0:%.*]] = zext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[SB0_A0:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[SB0_A0:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[CB0_A0]], ptr [[SB0_A0]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[CA6_B6:%.*]] = sext i1 [[TMP3]] to i32
-; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[CA6_B6]], ptr [[SA6_B6]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[CB6_A6:%.*]] = zext i1 [[TMP4]] to i32
-; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 [[CB6_A6]], ptr [[SB6_A6]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memcmp-7.ll b/llvm/test/Transforms/InstCombine/memcmp-7.ll
index 72dccfd86f801e..8a6ce565f502d6 100644
--- a/llvm/test/Transforms/InstCombine/memcmp-7.ll
+++ b/llvm/test/Transforms/InstCombine/memcmp-7.ll
@@ -14,9 +14,9 @@ declare i32 @memcmp(ptr, ptr, i64)
define void @fold_memcmp_i32a_i32b_pIb(i32 %I, ptr %pcmp)
; CHECK-LABEL: @fold_memcmp_i32a_i32b_pIb(
; CHECK-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[PST_1_1_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[PST_1_1_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 0, ptr [[PST_1_1_2]], align 4
-; CHECK-NEXT: [[PST_1_1_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[PST_1_1_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 0, ptr [[PST_1_1_3]], align 4
; CHECK-NEXT: ret void
;
@@ -50,23 +50,23 @@ define void @fold_memcmp_i32a_i32b_pIb(i32 %I, ptr %pcmp)
define void @fold_memcmp_A_B_pIb(i32 %I, ptr %pcmp) {
; CHECK-LABEL: @fold_memcmp_A_B_pIb(
; CHECK-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[PST_0_0_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[PST_0_0_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 0, ptr [[PST_0_0_2]], align 4
-; CHECK-NEXT: [[PST_0_0_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[PST_0_0_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 0, ptr [[PST_0_0_3]], align 4
-; CHECK-NEXT: [[PST_0_0_4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[PST_0_0_4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 0, ptr [[PST_0_0_4]], align 4
-; CHECK-NEXT: [[PST_0_1_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[PST_0_1_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 -1, ptr [[PST_0_1_1]], align 4
-; CHECK-NEXT: [[PST_0_1_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[PST_0_1_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 -1, ptr [[PST_0_1_2]], align 4
-; CHECK-NEXT: [[PST_0_1_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; CHECK-NEXT: [[PST_0_1_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; CHECK-NEXT: store i32 -1, ptr [[PST_0_1_3]], align 4
-; CHECK-NEXT: [[PST_1_0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[PST_1_0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 1, ptr [[PST_1_0_1]], align 4
-; CHECK-NEXT: [[PST_1_0_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[PST_1_0_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 1, ptr [[PST_1_0_2]], align 4
-; CHECK-NEXT: [[PST_1_0_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; CHECK-NEXT: [[PST_1_0_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; CHECK-NEXT: store i32 1, ptr [[PST_1_0_3]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memcpy_alloca.ll b/llvm/test/Transforms/InstCombine/memcpy_alloca.ll
index b40f4b8ca5a305..e58e357684e18c 100644
--- a/llvm/test/Transforms/InstCombine/memcpy_alloca.ll
+++ b/llvm/test/Transforms/InstCombine/memcpy_alloca.ll
@@ -59,7 +59,7 @@ define void @test5(ptr %dest) {
define void @test6(ptr %dest) {
; CHECK-LABEL: @test6(
; CHECK-NEXT: [[A:%.*]] = alloca [7 x i8], align 1
-; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds i16, ptr [[A]], i64 1
+; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 2
; CHECK-NEXT: store i16 42, ptr [[P2]], align 2
; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 1 dereferenceable(7) [[DEST:%.*]], ptr noundef nonnull align 1 dereferenceable(7) [[P2]], i64 7, i1 false)
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/InstCombine/memrchr-5.ll b/llvm/test/Transforms/InstCombine/memrchr-5.ll
index 535c01fd6fe797..2646f8d2e94086 100644
--- a/llvm/test/Transforms/InstCombine/memrchr-5.ll
+++ b/llvm/test/Transforms/InstCombine/memrchr-5.ll
@@ -16,37 +16,37 @@ declare ptr @memrchr(ptr, i32, i64)
define void @fold_memrchr_a_16(ptr %pcmp) {
; BE-LABEL: @fold_memrchr_a_16(
; BE-NEXT: store i64 0, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; BE-NEXT: store i64 1, ptr [[PSTOR1]], align 4
-; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; BE-NEXT: store i64 2, ptr [[PSTOR2]], align 4
-; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; BE-NEXT: store i64 3, ptr [[PSTOR3]], align 4
-; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; BE-NEXT: store i64 13, ptr [[PSTOR4]], align 4
-; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i64, ptr [[PCMP]], i64 6
+; BE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 48
; BE-NEXT: store i64 14, ptr [[PSTOR6]], align 4
-; BE-NEXT: [[PSTOR7:%.*]] = getelementptr i64, ptr [[PCMP]], i64 7
+; BE-NEXT: [[PSTOR7:%.*]] = getelementptr i8, ptr [[PCMP]], i64 56
; BE-NEXT: store i64 15, ptr [[PSTOR7]], align 4
-; BE-NEXT: [[PSTOR8:%.*]] = getelementptr i64, ptr [[PCMP]], i64 8
+; BE-NEXT: [[PSTOR8:%.*]] = getelementptr i8, ptr [[PCMP]], i64 64
; BE-NEXT: store i64 0, ptr [[PSTOR8]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memrchr_a_16(
; LE-NEXT: store i64 3, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; LE-NEXT: store i64 2, ptr [[PSTOR1]], align 4
-; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; LE-NEXT: store i64 1, ptr [[PSTOR2]], align 4
-; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; LE-NEXT: store i64 0, ptr [[PSTOR3]], align 4
-; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; LE-NEXT: store i64 14, ptr [[PSTOR4]], align 4
-; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i64, ptr [[PCMP]], i64 6
+; LE-NEXT: [[PSTOR6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 48
; LE-NEXT: store i64 13, ptr [[PSTOR6]], align 4
-; LE-NEXT: [[PSTOR7:%.*]] = getelementptr i64, ptr [[PCMP]], i64 7
+; LE-NEXT: [[PSTOR7:%.*]] = getelementptr i8, ptr [[PCMP]], i64 56
; LE-NEXT: store i64 12, ptr [[PSTOR7]], align 4
-; LE-NEXT: [[PSTOR8:%.*]] = getelementptr i64, ptr [[PCMP]], i64 8
+; LE-NEXT: [[PSTOR8:%.*]] = getelementptr i8, ptr [[PCMP]], i64 64
; LE-NEXT: store i64 0, ptr [[PSTOR8]], align 4
; LE-NEXT: ret void
;
@@ -123,29 +123,29 @@ define void @fold_memrchr_a_16(ptr %pcmp) {
define void @fold_memrchr_a_p1_16(ptr %pcmp) {
; BE-LABEL: @fold_memrchr_a_p1_16(
; BE-NEXT: store i64 0, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; BE-NEXT: store i64 1, ptr [[PSTOR1]], align 4
-; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; BE-NEXT: store i64 2, ptr [[PSTOR2]], align 4
-; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; BE-NEXT: store i64 3, ptr [[PSTOR3]], align 4
-; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; BE-NEXT: store i64 0, ptr [[PSTOR4]], align 4
-; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i64, ptr [[PCMP]], i64 5
+; BE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 40
; BE-NEXT: store i64 0, ptr [[PSTOR5]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memrchr_a_p1_16(
; LE-NEXT: store i64 3, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; LE-NEXT: store i64 2, ptr [[PSTOR1]], align 4
-; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; LE-NEXT: store i64 1, ptr [[PSTOR2]], align 4
-; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; LE-NEXT: store i64 0, ptr [[PSTOR3]], align 4
-; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; LE-NEXT: store i64 0, ptr [[PSTOR4]], align 4
-; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i64, ptr [[PCMP]], i64 5
+; LE-NEXT: [[PSTOR5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 40
; LE-NEXT: store i64 0, ptr [[PSTOR5]], align 4
; LE-NEXT: ret void
;
@@ -206,25 +206,25 @@ define void @fold_memrchr_a_p1_16(ptr %pcmp) {
define void @fold_memrchr_a_20(ptr %pcmp) {
; BE-LABEL: @fold_memrchr_a_20(
; BE-NEXT: store i64 16, ptr [[PCMP:%.*]], align 4
-; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; BE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; BE-NEXT: store i64 17, ptr [[PSTOR1]], align 4
-; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; BE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; BE-NEXT: store i64 18, ptr [[PSTOR2]], align 4
-; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; BE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; BE-NEXT: store i64 19, ptr [[PSTOR3]], align 4
-; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; BE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; BE-NEXT: store i64 4, ptr [[PSTOR4]], align 4
; BE-NEXT: ret void
;
; LE-LABEL: @fold_memrchr_a_20(
; LE-NEXT: store i64 19, ptr [[PCMP:%.*]], align 4
-; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i64, ptr [[PCMP]], i64 1
+; LE-NEXT: [[PSTOR1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; LE-NEXT: store i64 18, ptr [[PSTOR1]], align 4
-; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i64, ptr [[PCMP]], i64 2
+; LE-NEXT: [[PSTOR2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; LE-NEXT: store i64 17, ptr [[PSTOR2]], align 4
-; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i64, ptr [[PCMP]], i64 3
+; LE-NEXT: [[PSTOR3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; LE-NEXT: store i64 16, ptr [[PSTOR3]], align 4
-; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i64, ptr [[PCMP]], i64 4
+; LE-NEXT: [[PSTOR4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; LE-NEXT: store i64 7, ptr [[PSTOR4]], align 4
; LE-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memset2.ll b/llvm/test/Transforms/InstCombine/memset2.ll
index d81abf659335a1..be5faccca3c43b 100644
--- a/llvm/test/Transforms/InstCombine/memset2.ll
+++ b/llvm/test/Transforms/InstCombine/memset2.ll
@@ -9,7 +9,7 @@ define i32 @test(ptr addrspace(1) nocapture %moves) {
; CHECK-LABEL: define i32 @test
; CHECK-SAME: (ptr addrspace(1) nocapture [[MOVES:%.*]]) {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds [[STRUCT_MOVES:%.*]], ptr addrspace(1) [[MOVES]], i64 1, i32 0, i64 9
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[MOVES]], i64 26
; CHECK-NEXT: store i64 0, ptr addrspace(1) [[GEP]], align 1
; CHECK-NEXT: ret i32 0
;
diff --git a/llvm/test/Transforms/InstCombine/multi-size-address-space-pointer.ll b/llvm/test/Transforms/InstCombine/multi-size-address-space-pointer.ll
index 99eb53edab786b..df3152a6a4353d 100644
--- a/llvm/test/Transforms/InstCombine/multi-size-address-space-pointer.ll
+++ b/llvm/test/Transforms/InstCombine/multi-size-address-space-pointer.ll
@@ -5,7 +5,7 @@ target datalayout = "e-p:32:32:32-p1:64:64:64-p2:8:8:8-p3:16:16:16-p4:16:16:16-i
define i32 @test_as0(ptr addrspace(0) %a) {
; CHECK-LABEL: @test_as0(
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i32, ptr [[A:%.*]], i32 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i8, ptr [[A:%.*]], i32 4
; CHECK-NEXT: [[Y:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: ret i32 [[Y]]
;
@@ -16,7 +16,7 @@ define i32 @test_as0(ptr addrspace(0) %a) {
define i32 @test_as1(ptr addrspace(1) %a) {
; CHECK-LABEL: @test_as1(
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i32, ptr addrspace(1) [[A:%.*]], i64 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i8, ptr addrspace(1) [[A:%.*]], i64 4
; CHECK-NEXT: [[Y:%.*]] = load i32, ptr addrspace(1) [[ARRAYIDX]], align 4
; CHECK-NEXT: ret i32 [[Y]]
;
@@ -27,7 +27,7 @@ define i32 @test_as1(ptr addrspace(1) %a) {
define i32 @test_as2(ptr addrspace(2) %a) {
; CHECK-LABEL: @test_as2(
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i32, ptr addrspace(2) [[A:%.*]], i8 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i8, ptr addrspace(2) [[A:%.*]], i8 4
; CHECK-NEXT: [[Y:%.*]] = load i32, ptr addrspace(2) [[ARRAYIDX]], align 4
; CHECK-NEXT: ret i32 [[Y]]
;
@@ -38,7 +38,7 @@ define i32 @test_as2(ptr addrspace(2) %a) {
define i32 @test_as3(ptr addrspace(3) %a) {
; CHECK-LABEL: @test_as3(
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i32, ptr addrspace(3) [[A:%.*]], i16 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr i8, ptr addrspace(3) [[A:%.*]], i16 4
; CHECK-NEXT: [[Y:%.*]] = load i32, ptr addrspace(3) [[ARRAYIDX]], align 4
; CHECK-NEXT: ret i32 [[Y]]
;
@@ -92,7 +92,7 @@ define <2 x i8> @test_combine_vector_inttoptr(<2 x i8> %a) {
; Check that the GEP index is changed to the address space integer type (i64 -> i8)
define ptr addrspace(2) @shrink_gep_constant_index_64_as2(ptr addrspace(2) %p) {
; CHECK-LABEL: @shrink_gep_constant_index_64_as2(
-; CHECK-NEXT: [[RET:%.*]] = getelementptr i32, ptr addrspace(2) [[P:%.*]], i8 1
+; CHECK-NEXT: [[RET:%.*]] = getelementptr i8, ptr addrspace(2) [[P:%.*]], i8 4
; CHECK-NEXT: ret ptr addrspace(2) [[RET]]
;
%ret = getelementptr i32, ptr addrspace(2) %p, i64 1
@@ -101,7 +101,7 @@ define ptr addrspace(2) @shrink_gep_constant_index_64_as2(ptr addrspace(2) %p) {
define ptr addrspace(2) @shrink_gep_constant_index_32_as2(ptr addrspace(2) %p) {
; CHECK-LABEL: @shrink_gep_constant_index_32_as2(
-; CHECK-NEXT: [[RET:%.*]] = getelementptr i32, ptr addrspace(2) [[P:%.*]], i8 1
+; CHECK-NEXT: [[RET:%.*]] = getelementptr i8, ptr addrspace(2) [[P:%.*]], i8 4
; CHECK-NEXT: ret ptr addrspace(2) [[RET]]
;
%ret = getelementptr i32, ptr addrspace(2) %p, i32 1
@@ -110,7 +110,7 @@ define ptr addrspace(2) @shrink_gep_constant_index_32_as2(ptr addrspace(2) %p) {
define ptr addrspace(3) @shrink_gep_constant_index_64_as3(ptr addrspace(3) %p) {
; CHECK-LABEL: @shrink_gep_constant_index_64_as3(
-; CHECK-NEXT: [[RET:%.*]] = getelementptr i32, ptr addrspace(3) [[P:%.*]], i16 1
+; CHECK-NEXT: [[RET:%.*]] = getelementptr i8, ptr addrspace(3) [[P:%.*]], i16 4
; CHECK-NEXT: ret ptr addrspace(3) [[RET]]
;
%ret = getelementptr i32, ptr addrspace(3) %p, i64 1
diff --git a/llvm/test/Transforms/InstCombine/non-integral-pointers.ll b/llvm/test/Transforms/InstCombine/non-integral-pointers.ll
index 7c82b7ff36f5d2..11e2108f600786 100644
--- a/llvm/test/Transforms/InstCombine/non-integral-pointers.ll
+++ b/llvm/test/Transforms/InstCombine/non-integral-pointers.ll
@@ -56,7 +56,7 @@ define i64 @g(ptr %gp) {
; CHECK-NEXT: [[DOTPRE:%.*]] = load ptr addrspace(4), ptr [[GP:%.*]], align 8
; CHECK-NEXT: [[V74:%.*]] = call ptr addrspace(4) @alloc()
; CHECK-NEXT: [[V75:%.*]] = addrspacecast ptr addrspace(4) [[V74]] to ptr
-; CHECK-NEXT: [[V77:%.*]] = getelementptr ptr addrspace(4), ptr [[V75]], i64 -1
+; CHECK-NEXT: [[V77:%.*]] = getelementptr i8, ptr [[V75]], i64 -8
; CHECK-NEXT: store ptr addrspace(4) [[DOTPRE]], ptr [[V77]], align 8
; CHECK-NEXT: [[V81:%.*]] = load i64, ptr [[V77]], align 8
; CHECK-NEXT: ret i64 [[V81]]
@@ -74,7 +74,7 @@ define i64 @g2(ptr addrspace(4) %gp) {
; CHECK-LABEL: @g2(
; CHECK-NEXT: [[DOTPRE:%.*]] = load ptr, ptr addrspace(4) [[GP:%.*]], align 8
; CHECK-NEXT: [[V74:%.*]] = call ptr addrspace(4) @alloc()
-; CHECK-NEXT: [[V77:%.*]] = getelementptr ptr, ptr addrspace(4) [[V74]], i64 -1
+; CHECK-NEXT: [[V77:%.*]] = getelementptr i8, ptr addrspace(4) [[V74]], i64 -8
; CHECK-NEXT: store ptr [[DOTPRE]], ptr addrspace(4) [[V77]], align 8
; CHECK-NEXT: [[V81_CAST:%.*]] = ptrtoint ptr [[DOTPRE]] to i64
; CHECK-NEXT: ret i64 [[V81_CAST]]
diff --git a/llvm/test/Transforms/InstCombine/opaque-ptr.ll b/llvm/test/Transforms/InstCombine/opaque-ptr.ll
index 4448a49ad92bf5..4d38e2cd37c959 100644
--- a/llvm/test/Transforms/InstCombine/opaque-ptr.ll
+++ b/llvm/test/Transforms/InstCombine/opaque-ptr.ll
@@ -160,7 +160,7 @@ define void @varargs_cast_opaque_to_typed(ptr %a) {
define ptr @geps_combinable(ptr %a) {
; CHECK-LABEL: @geps_combinable(
-; CHECK-NEXT: [[A3:%.*]] = getelementptr { i32, { i32, i32 } }, ptr [[A:%.*]], i64 0, i32 1, i32 1
+; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 8
; CHECK-NEXT: ret ptr [[A3]]
;
%a2 = getelementptr { i32, { i32, i32 } }, ptr %a, i32 0, i32 1
@@ -170,7 +170,7 @@ define ptr @geps_combinable(ptr %a) {
define ptr @geps_combinable_different_elem_type1(ptr %a) {
; CHECK-LABEL: @geps_combinable_different_elem_type1(
-; CHECK-NEXT: [[A3:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 1
+; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 8
; CHECK-NEXT: ret ptr [[A3]]
;
%a2 = getelementptr { i32, i32 }, ptr %a, i32 0, i32 1
@@ -180,7 +180,7 @@ define ptr @geps_combinable_different_elem_type1(ptr %a) {
define ptr @geps_combinable_different_elem_type2(ptr %a) {
; CHECK-LABEL: @geps_combinable_different_elem_type2(
-; CHECK-NEXT: [[A3:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 1
+; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 8
; CHECK-NEXT: ret ptr [[A3]]
;
%a2 = getelementptr { i32, i32 }, ptr %a, i32 0, i32 1
@@ -190,7 +190,7 @@ define ptr @geps_combinable_different_elem_type2(ptr %a) {
define ptr @geps_combinable_different_elem_type3(ptr %a) {
; CHECK-LABEL: @geps_combinable_different_elem_type3(
-; CHECK-NEXT: [[A3:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 1, i32 1
+; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 12
; CHECK-NEXT: ret ptr [[A3]]
;
%a2 = getelementptr { i32, i32 }, ptr %a, i32 0, i32 1
@@ -263,9 +263,9 @@ declare void @use(ptr)
define ptr @geps_combinable_different_elem_type_extra_use1(ptr %a) {
; CHECK-LABEL: @geps_combinable_different_elem_type_extra_use1(
-; CHECK-NEXT: [[A2:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[A2:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 4
; CHECK-NEXT: call void @use(ptr [[A2]])
-; CHECK-NEXT: [[A3:%.*]] = getelementptr { i32, i32 }, ptr [[A]], i64 1
+; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: ret ptr [[A3]]
;
%a2 = getelementptr { i32, i32 }, ptr %a, i32 0, i32 1
@@ -440,7 +440,7 @@ bb2:
define ptr addrspace(1) @gep_of_addrspace_cast(ptr %ptr) {
; CHECK-LABEL: @gep_of_addrspace_cast(
; CHECK-NEXT: [[CAST1:%.*]] = addrspacecast ptr [[PTR:%.*]] to ptr addrspace(1)
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[CAST1]], i64 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[CAST1]], i64 4
; CHECK-NEXT: ret ptr addrspace(1) [[GEP]]
;
%cast1 = addrspacecast ptr %ptr to ptr addrspace(1)
@@ -519,7 +519,7 @@ define ptr @phi_of_gep(i1 %c, ptr %p) {
; CHECK: else:
; CHECK-NEXT: br label [[JOIN]]
; CHECK: join:
-; CHECK-NEXT: [[PHI:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[PHI:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: ret ptr [[PHI]]
;
br i1 %c, label %if, label %else
@@ -541,10 +541,10 @@ define ptr @phi_of_gep_different_type(i1 %c, ptr %p) {
; CHECK-LABEL: @phi_of_gep_different_type(
; CHECK-NEXT: br i1 [[C:%.*]], label [[IF:%.*]], label [[ELSE:%.*]]
; CHECK: if:
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 1
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 4
; CHECK-NEXT: br label [[JOIN:%.*]]
; CHECK: else:
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i64, ptr [[P]], i64 1
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[P]], i64 8
; CHECK-NEXT: br label [[JOIN]]
; CHECK: join:
; CHECK-NEXT: [[PHI:%.*]] = phi ptr [ [[GEP1]], [[IF]] ], [ [[GEP2]], [[ELSE]] ]
@@ -573,9 +573,9 @@ define ptr @gep_of_phi_of_gep(i1 %c, ptr %p) {
; CHECK: else:
; CHECK-NEXT: br label [[JOIN]]
; CHECK: join:
-; CHECK-NEXT: [[TMP1:%.*]] = phi i64 [ 1, [[IF]] ], [ 2, [[ELSE]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[TMP1]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[TMP2]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = phi i64 [ 4, [[IF]] ], [ 8, [[ELSE]] ]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 4
; CHECK-NEXT: ret ptr [[GEP]]
;
br i1 %c, label %if, label %else
@@ -598,14 +598,13 @@ define ptr @gep_of_phi_of_gep_different_type(i1 %c, ptr %p) {
; CHECK-LABEL: @gep_of_phi_of_gep_different_type(
; CHECK-NEXT: br i1 [[C:%.*]], label [[IF:%.*]], label [[ELSE:%.*]]
; CHECK: if:
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 1
; CHECK-NEXT: br label [[JOIN:%.*]]
; CHECK: else:
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i64, ptr [[P]], i64 2
; CHECK-NEXT: br label [[JOIN]]
; CHECK: join:
-; CHECK-NEXT: [[PHI:%.*]] = phi ptr [ [[GEP1]], [[IF]] ], [ [[GEP2]], [[ELSE]] ]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[PHI]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = phi i64 [ 4, [[IF]] ], [ 16, [[ELSE]] ]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 4
; CHECK-NEXT: ret ptr [[GEP]]
;
br i1 %c, label %if, label %else
@@ -626,8 +625,8 @@ join:
define ptr @select_of_gep(i1 %c, ptr %p) {
; CHECK-LABEL: @select_of_gep(
-; CHECK-NEXT: [[S_V:%.*]] = select i1 [[C:%.*]], i64 1, i64 2
-; CHECK-NEXT: [[S:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[S_V]]
+; CHECK-NEXT: [[S_V:%.*]] = select i1 [[C:%.*]], i64 4, i64 8
+; CHECK-NEXT: [[S:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[S_V]]
; CHECK-NEXT: ret ptr [[S]]
;
%gep1 = getelementptr i32, ptr %p, i64 1
@@ -638,9 +637,8 @@ define ptr @select_of_gep(i1 %c, ptr %p) {
define ptr @select_of_gep_different_type(i1 %c, ptr %p) {
; CHECK-LABEL: @select_of_gep_different_type(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 1
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i64, ptr [[P]], i64 2
-; CHECK-NEXT: [[S:%.*]] = select i1 [[C:%.*]], ptr [[GEP1]], ptr [[GEP2]]
+; CHECK-NEXT: [[S_V:%.*]] = select i1 [[C:%.*]], i64 4, i64 16
+; CHECK-NEXT: [[S:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[S_V]]
; CHECK-NEXT: ret ptr [[S]]
;
%gep1 = getelementptr i32, ptr %p, i64 1
diff --git a/llvm/test/Transforms/InstCombine/phi-equal-incoming-pointers.ll b/llvm/test/Transforms/InstCombine/phi-equal-incoming-pointers.ll
index ea75de4e99f88c..998169ec2fe26f 100644
--- a/llvm/test/Transforms/InstCombine/phi-equal-incoming-pointers.ll
+++ b/llvm/test/Transforms/InstCombine/phi-equal-incoming-pointers.ll
@@ -157,7 +157,7 @@ define i32 @test_gep_i32ptr(i1 %cond, i1 %cond2) {
; ALL: bb2:
; ALL-NEXT: br label [[EXIT]]
; ALL: exit:
-; ALL-NEXT: [[PTR_TYPED:%.*]] = getelementptr inbounds i32, ptr [[OBJ]], i64 16
+; ALL-NEXT: [[PTR_TYPED:%.*]] = getelementptr inbounds i8, ptr [[OBJ]], i64 64
; ALL-NEXT: [[RES_PHI:%.*]] = load i32, ptr [[PTR_TYPED]], align 4
; ALL-NEXT: store i32 1, ptr [[PTR_TYPED]], align 4
; ALL-NEXT: [[RES:%.*]] = select i1 [[COND2:%.*]], i32 [[RES_PHI]], i32 1
diff --git a/llvm/test/Transforms/InstCombine/phi-timeout.ll b/llvm/test/Transforms/InstCombine/phi-timeout.ll
index cb28502fdbb602..f30750ef0455c3 100644
--- a/llvm/test/Transforms/InstCombine/phi-timeout.ll
+++ b/llvm/test/Transforms/InstCombine/phi-timeout.ll
@@ -10,7 +10,7 @@ define void @timeout(ptr nocapture readonly %cinfo, ptr %ptr) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds i16, ptr [[CINFO:%.*]], i32 2
+; CHECK-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds i8, ptr [[CINFO:%.*]], i32 4
; CHECK-NEXT: [[L:%.*]] = load i16, ptr [[ARRAYIDX15]], align 2
; CHECK-NEXT: [[CMP17:%.*]] = icmp eq i16 [[L]], 0
; CHECK-NEXT: [[EXTRACT_T1:%.*]] = trunc i16 [[L]] to i8
diff --git a/llvm/test/Transforms/InstCombine/phi.ll b/llvm/test/Transforms/InstCombine/phi.ll
index 90818771675b2e..a95fcef51cf9a5 100644
--- a/llvm/test/Transforms/InstCombine/phi.ll
+++ b/llvm/test/Transforms/InstCombine/phi.ll
@@ -232,7 +232,7 @@ define ptr @test8(ptr %A, i1 %b) {
; CHECK: BB1:
; CHECK-NEXT: br label [[BB2]]
; CHECK: BB2:
-; CHECK-NEXT: [[B:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 4
; CHECK-NEXT: ret ptr [[B]]
;
BB0:
@@ -1739,7 +1739,7 @@ define i1 @pr57488_icmp_of_phi(ptr %ptr.base, i64 %len) {
; CHECK: loop:
; CHECK-NEXT: [[ACCUM:%.*]] = phi i1 [ [[AND:%.*]], [[LOOP]] ], [ true, [[START:%.*]] ]
; CHECK-NEXT: [[PTR:%.*]] = phi ptr [ [[PTR_NEXT:%.*]], [[LOOP]] ], [ [[PTR_BASE]], [[START]] ]
-; CHECK-NEXT: [[PTR_NEXT]] = getelementptr inbounds i64, ptr [[PTR]], i64 1
+; CHECK-NEXT: [[PTR_NEXT]] = getelementptr inbounds i8, ptr [[PTR]], i64 8
; CHECK-NEXT: [[VAL:%.*]] = load i64, ptr [[PTR]], align 8
; CHECK-NEXT: [[VAL_ZERO:%.*]] = icmp eq i64 [[VAL]], 0
; CHECK-NEXT: [[AND]] = and i1 [[ACCUM]], [[VAL_ZERO]]
diff --git a/llvm/test/Transforms/InstCombine/pr39908.ll b/llvm/test/Transforms/InstCombine/pr39908.ll
index 0f0eae68439d7a..ca143f417fb27d 100644
--- a/llvm/test/Transforms/InstCombine/pr39908.ll
+++ b/llvm/test/Transforms/InstCombine/pr39908.ll
@@ -8,7 +8,7 @@ target datalayout = "p:32:32"
define i1 @test(ptr %p, i32 %n) {
; CHECK-LABEL: @test(
; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds [0 x %S], ptr [[P:%.*]], i32 0, i32 [[N:%.*]], i32 0, i32 0
-; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds [[S:%.*]], ptr [[END]], i32 -1
+; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds i8, ptr [[END]], i32 -8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[LAST]], [[P]]
; CHECK-NEXT: ret i1 [[CMP]]
;
@@ -23,7 +23,7 @@ define i1 @test64(ptr %p, i64 %n) {
; CHECK-LABEL: @test64(
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[N:%.*]] to i32
; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds [0 x %S], ptr [[P:%.*]], i32 0, i32 [[TMP1]], i32 0, i32 0
-; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds [[S:%.*]], ptr [[END]], i32 -1
+; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds i8, ptr [[END]], i32 -8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[LAST]], [[P]]
; CHECK-NEXT: ret i1 [[CMP]]
;
@@ -38,7 +38,7 @@ define i1 @test64_overflow(ptr %p, i64 %n) {
; CHECK-LABEL: @test64_overflow(
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[N:%.*]] to i32
; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds [0 x %S], ptr [[P:%.*]], i32 0, i32 [[TMP1]], i32 0, i32 0
-; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds [[S:%.*]], ptr [[END]], i32 -1
+; CHECK-NEXT: [[LAST:%.*]] = getelementptr inbounds i8, ptr [[END]], i32 -8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[LAST]], [[P]]
; CHECK-NEXT: ret i1 [[CMP]]
;
diff --git a/llvm/test/Transforms/InstCombine/pr44242.ll b/llvm/test/Transforms/InstCombine/pr44242.ll
index e7b0f7da03476b..e86c17057fe27c 100644
--- a/llvm/test/Transforms/InstCombine/pr44242.ll
+++ b/llvm/test/Transforms/InstCombine/pr44242.ll
@@ -118,7 +118,7 @@ define void @store_address(i32 %x) {
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[X:%.*]], 0
; CHECK-NEXT: br i1 [[CMP]], label [[END:%.*]], label [[LOOP]]
; CHECK: loop:
-; CHECK-NEXT: [[VAL_INCR]] = getelementptr float, ptr [[VAL]], i64 1
+; CHECK-NEXT: [[VAL_INCR]] = getelementptr i8, ptr [[VAL]], i64 4
; CHECK-NEXT: br label [[LOOP_HEADER]]
; CHECK: end:
; CHECK-NEXT: store i32 0, ptr [[VAL]], align 4
diff --git a/llvm/test/Transforms/InstCombine/pr58901.ll b/llvm/test/Transforms/InstCombine/pr58901.ll
index f94c3f131b2be3..1eea4b5305545d 100644
--- a/llvm/test/Transforms/InstCombine/pr58901.ll
+++ b/llvm/test/Transforms/InstCombine/pr58901.ll
@@ -3,8 +3,9 @@
define ptr @f1(ptr %arg, i64 %arg1) {
; CHECK-LABEL: @f1(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [6 x i32], ptr [[ARG:%.*]], i64 3, i64 [[ARG1:%.*]]
-; CHECK-NEXT: ret ptr [[TMP1]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[ARG:%.*]], i64 72
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr [6 x i32], ptr [[TMP1]], i64 0, i64 [[ARG1:%.*]]
+; CHECK-NEXT: ret ptr [[TMP2]]
;
%1 = getelementptr [6 x i32], ptr %arg, i64 3
%2 = getelementptr [6 x i32], ptr %1, i64 0, i64 %arg1
@@ -13,7 +14,7 @@ define ptr @f1(ptr %arg, i64 %arg1) {
define ptr @f2(ptr %arg, i64 %arg1) {
; CHECK-LABEL: @f2(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [6 x i32], ptr [[ARG:%.*]], i64 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[ARG:%.*]], i64 72
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr [6 x i32], ptr [[TMP1]], i64 [[ARG1:%.*]], i64 [[ARG1]]
; CHECK-NEXT: ret ptr [[TMP2]]
;
diff --git a/llvm/test/Transforms/InstCombine/ptr-replace-alloca.ll b/llvm/test/Transforms/InstCombine/ptr-replace-alloca.ll
index 6a0e7098ab744e..a7aa3a8ef1bebf 100644
--- a/llvm/test/Transforms/InstCombine/ptr-replace-alloca.ll
+++ b/llvm/test/Transforms/InstCombine/ptr-replace-alloca.ll
@@ -45,10 +45,10 @@ define i8 @volatile_load_keep_alloca(i1 %cond) {
; CHECK-NEXT: call void @llvm.memcpy.p1.p0.i64(ptr addrspace(1) noundef align 4 dereferenceable(32) [[ALLOCA]], ptr noundef nonnull align 16 dereferenceable(32) @g1, i64 32, i1 false)
; CHECK-NEXT: br i1 [[COND:%.*]], label [[IF:%.*]], label [[ELSE:%.*]]
; CHECK: if:
-; CHECK-NEXT: [[VAL_IF:%.*]] = getelementptr inbounds [32 x i8], ptr addrspace(1) [[ALLOCA]], i64 0, i64 1
+; CHECK-NEXT: [[VAL_IF:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[ALLOCA]], i64 1
; CHECK-NEXT: br label [[SINK:%.*]]
; CHECK: else:
-; CHECK-NEXT: [[VAL_ELSE:%.*]] = getelementptr inbounds [32 x i8], ptr addrspace(1) [[ALLOCA]], i64 0, i64 2
+; CHECK-NEXT: [[VAL_ELSE:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[ALLOCA]], i64 2
; CHECK-NEXT: br label [[SINK]]
; CHECK: sink:
; CHECK-NEXT: [[PTR:%.*]] = phi ptr addrspace(1) [ [[VAL_IF]], [[IF]] ], [ [[VAL_ELSE]], [[ELSE]] ]
@@ -81,10 +81,10 @@ define i8 @no_memcpy_keep_alloca(i1 %cond) {
; CHECK-NEXT: [[ALLOCA:%.*]] = alloca [32 x i8], align 4, addrspace(1)
; CHECK-NEXT: br i1 [[COND:%.*]], label [[IF:%.*]], label [[ELSE:%.*]]
; CHECK: if:
-; CHECK-NEXT: [[VAL_IF:%.*]] = getelementptr inbounds [32 x i8], ptr addrspace(1) [[ALLOCA]], i64 0, i64 1
+; CHECK-NEXT: [[VAL_IF:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[ALLOCA]], i64 1
; CHECK-NEXT: br label [[SINK:%.*]]
; CHECK: else:
-; CHECK-NEXT: [[VAL_ELSE:%.*]] = getelementptr inbounds [32 x i8], ptr addrspace(1) [[ALLOCA]], i64 0, i64 2
+; CHECK-NEXT: [[VAL_ELSE:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[ALLOCA]], i64 2
; CHECK-NEXT: br label [[SINK]]
; CHECK: sink:
; CHECK-NEXT: [[PTR:%.*]] = phi ptr addrspace(1) [ [[VAL_IF]], [[IF]] ], [ [[VAL_ELSE]], [[ELSE]] ]
@@ -445,7 +445,7 @@ define i8 @call_readonly_keep_alloca2() {
; CHECK-LABEL: @call_readonly_keep_alloca2(
; CHECK-NEXT: [[ALLOCA:%.*]] = alloca [32 x i8], align 1, addrspace(1)
; CHECK-NEXT: call void @llvm.memcpy.p1.p0.i64(ptr addrspace(1) noundef align 1 dereferenceable(16) [[ALLOCA]], ptr noundef nonnull align 16 dereferenceable(16) @g1, i64 16, i1 false)
-; CHECK-NEXT: [[A1:%.*]] = getelementptr inbounds [32 x i8], ptr addrspace(1) [[ALLOCA]], i64 0, i64 16
+; CHECK-NEXT: [[A1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[ALLOCA]], i64 16
; CHECK-NEXT: call void @llvm.memcpy.p1.p1.i64(ptr addrspace(1) noundef align 1 dereferenceable(16) [[A1]], ptr addrspace(1) noundef align 16 dereferenceable(16) @g2, i64 16, i1 false)
; CHECK-NEXT: [[P:%.*]] = addrspacecast ptr addrspace(1) [[ALLOCA]] to ptr
; CHECK-NEXT: [[V:%.*]] = call i8 @readonly_callee(ptr [[P]])
diff --git a/llvm/test/Transforms/InstCombine/select-cmp-br.ll b/llvm/test/Transforms/InstCombine/select-cmp-br.ll
index 92dc562281b81a..f1c62a5593f0f2 100644
--- a/llvm/test/Transforms/InstCombine/select-cmp-br.ll
+++ b/llvm/test/Transforms/InstCombine/select-cmp-br.ll
@@ -12,7 +12,7 @@ define void @test1(ptr %arg) {
; CHECK-LABEL: @test1(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[M:%.*]] = load ptr, ptr [[ARG:%.*]], align 8
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[C:%.*]], ptr [[ARG]], i64 1, i32 0, i32 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[ARG]], i64 16
; CHECK-NEXT: [[N:%.*]] = load ptr, ptr [[TMP1]], align 8
; CHECK-NEXT: [[TMP5_NOT:%.*]] = icmp eq ptr [[M]], [[N]]
; CHECK-NEXT: br i1 [[TMP5_NOT]], label [[BB8:%.*]], label [[BB10:%.*]]
@@ -22,7 +22,7 @@ define void @test1(ptr %arg) {
; CHECK-NEXT: tail call void @bar(ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb10:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[M]], i64 9
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[M]], i64 72
; CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP2]], align 8
; CHECK-NEXT: [[TMP11:%.*]] = tail call i64 [[TMP4]](ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB]]
@@ -54,7 +54,7 @@ define void @test2(ptr %arg) {
; CHECK-LABEL: @test2(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[M:%.*]] = load ptr, ptr [[ARG:%.*]], align 8
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[C:%.*]], ptr [[ARG]], i64 1, i32 0, i32 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[ARG]], i64 16
; CHECK-NEXT: [[N:%.*]] = load ptr, ptr [[TMP1]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq ptr [[M]], [[N]]
; CHECK-NEXT: br i1 [[TMP5]], label [[BB10:%.*]], label [[BB8:%.*]]
@@ -64,7 +64,7 @@ define void @test2(ptr %arg) {
; CHECK-NEXT: tail call void @bar(ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb10:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[M]], i64 9
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[M]], i64 72
; CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP2]], align 8
; CHECK-NEXT: [[TMP11:%.*]] = tail call i64 [[TMP4]](ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB]]
@@ -96,7 +96,7 @@ define void @test3(ptr %arg) {
; CHECK-LABEL: @test3(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[M:%.*]] = load ptr, ptr [[ARG:%.*]], align 8
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[C:%.*]], ptr [[ARG]], i64 1, i32 0, i32 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[ARG]], i64 16
; CHECK-NEXT: [[N:%.*]] = load ptr, ptr [[TMP1]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq ptr [[M]], [[N]]
; CHECK-NEXT: br i1 [[TMP5]], label [[BB8:%.*]], label [[BB10:%.*]]
@@ -106,7 +106,7 @@ define void @test3(ptr %arg) {
; CHECK-NEXT: tail call void @bar(ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb10:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[M]], i64 9
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[M]], i64 72
; CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP2]], align 8
; CHECK-NEXT: [[TMP11:%.*]] = tail call i64 [[TMP4]](ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB]]
@@ -138,7 +138,7 @@ define void @test4(ptr %arg) {
; CHECK-LABEL: @test4(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[M:%.*]] = load ptr, ptr [[ARG:%.*]], align 8
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[C:%.*]], ptr [[ARG]], i64 1, i32 0, i32 0
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[ARG]], i64 16
; CHECK-NEXT: [[N:%.*]] = load ptr, ptr [[TMP1]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq ptr [[M]], [[N]]
; CHECK-NEXT: br i1 [[TMP5]], label [[BB10:%.*]], label [[BB8:%.*]]
@@ -148,7 +148,7 @@ define void @test4(ptr %arg) {
; CHECK-NEXT: tail call void @bar(ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb10:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[M]], i64 9
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[M]], i64 72
; CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP2]], align 8
; CHECK-NEXT: [[TMP11:%.*]] = tail call i64 [[TMP4]](ptr nonnull [[ARG]])
; CHECK-NEXT: br label [[BB]]
diff --git a/llvm/test/Transforms/InstCombine/select-gep.ll b/llvm/test/Transforms/InstCombine/select-gep.ll
index ad2b029c517644..53bbcda85bb229 100644
--- a/llvm/test/Transforms/InstCombine/select-gep.ll
+++ b/llvm/test/Transforms/InstCombine/select-gep.ll
@@ -5,7 +5,7 @@ define ptr @test1a(ptr %p, ptr %q) {
; CHECK-LABEL: @test1a(
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt ptr [[P:%.*]], [[Q:%.*]]
; CHECK-NEXT: [[SELECT_V:%.*]] = select i1 [[CMP]], ptr [[P]], ptr [[Q]]
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr i32, ptr [[SELECT_V]], i64 4
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr i8, ptr [[SELECT_V]], i64 16
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep1 = getelementptr i32, ptr %p, i64 4
@@ -19,7 +19,7 @@ define ptr @test1b(ptr %p, ptr %q) {
; CHECK-LABEL: @test1b(
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt ptr [[P:%.*]], [[Q:%.*]]
; CHECK-NEXT: [[SELECT_V:%.*]] = select i1 [[CMP]], ptr [[P]], ptr [[Q]]
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr i32, ptr [[SELECT_V]], i64 4
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr i8, ptr [[SELECT_V]], i64 16
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep1 = getelementptr inbounds i32, ptr %p, i64 4
@@ -33,7 +33,7 @@ define ptr @test1c(ptr %p, ptr %q) {
; CHECK-LABEL: @test1c(
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt ptr [[P:%.*]], [[Q:%.*]]
; CHECK-NEXT: [[SELECT_V:%.*]] = select i1 [[CMP]], ptr [[P]], ptr [[Q]]
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr i32, ptr [[SELECT_V]], i64 4
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr i8, ptr [[SELECT_V]], i64 16
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep1 = getelementptr i32, ptr %p, i64 4
@@ -47,7 +47,7 @@ define ptr @test1d(ptr %p, ptr %q) {
; CHECK-LABEL: @test1d(
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt ptr [[P:%.*]], [[Q:%.*]]
; CHECK-NEXT: [[SELECT_V:%.*]] = select i1 [[CMP]], ptr [[P]], ptr [[Q]]
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i32, ptr [[SELECT_V]], i64 4
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i8, ptr [[SELECT_V]], i64 16
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep1 = getelementptr inbounds i32, ptr %p, i64 4
@@ -103,8 +103,8 @@ define ptr @test2c(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2c(
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[X:%.*]]
; CHECK-NEXT: [[ICMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
-; CHECK-NEXT: [[SEL_IDX:%.*]] = select i1 [[ICMP]], i64 0, i64 6
-; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds i32, ptr [[GEP1]], i64 [[SEL_IDX]]
+; CHECK-NEXT: [[SEL_IDX:%.*]] = select i1 [[ICMP]], i64 0, i64 24
+; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds i8, ptr [[GEP1]], i64 [[SEL_IDX]]
; CHECK-NEXT: ret ptr [[SEL]]
;
%gep1 = getelementptr inbounds i32, ptr %p, i64 %x
@@ -119,8 +119,8 @@ define ptr @test2d(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2d(
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[X:%.*]]
; CHECK-NEXT: [[ICMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
-; CHECK-NEXT: [[SEL_IDX:%.*]] = select i1 [[ICMP]], i64 6, i64 0
-; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds i32, ptr [[GEP1]], i64 [[SEL_IDX]]
+; CHECK-NEXT: [[SEL_IDX:%.*]] = select i1 [[ICMP]], i64 24, i64 0
+; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds i8, ptr [[GEP1]], i64 [[SEL_IDX]]
; CHECK-NEXT: ret ptr [[SEL]]
;
%gep1 = getelementptr inbounds i32, ptr %p, i64 %x
diff --git a/llvm/test/Transforms/InstCombine/shift.ll b/llvm/test/Transforms/InstCombine/shift.ll
index 6f9ce1544cc0a1..d783adbe938632 100644
--- a/llvm/test/Transforms/InstCombine/shift.ll
+++ b/llvm/test/Transforms/InstCombine/shift.ll
@@ -119,8 +119,8 @@ define i8 @test11a(i8 %A) {
;; (A >> 8) << 8 === A & -256
define i32 @test12(i32 %A) {
; CHECK-LABEL: @test12(
-; CHECK-NEXT: [[C:%.*]] = and i32 [[A:%.*]], -256
-; CHECK-NEXT: ret i32 [[C]]
+; CHECK-NEXT: [[B:%.*]] = and i32 [[A:%.*]], -256
+; CHECK-NEXT: ret i32 [[B]]
;
%B = ashr i32 %A, 8
%C = shl i32 %B, 8
@@ -1754,7 +1754,7 @@ define void @ashr_out_of_range_1(ptr %A) {
; CHECK-NEXT: [[B:%.*]] = select i1 [[TMP1]], i177 0, i177 [[L_FROZEN]]
; CHECK-NEXT: [[TMP2:%.*]] = trunc i177 [[B]] to i64
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i177, ptr [[A]], i64 [[TMP2]]
-; CHECK-NEXT: [[G11:%.*]] = getelementptr i177, ptr [[TMP3]], i64 -1
+; CHECK-NEXT: [[G11:%.*]] = getelementptr i8, ptr [[TMP3]], i64 -24
; CHECK-NEXT: [[C17:%.*]] = icmp sgt i177 [[B]], [[L_FROZEN]]
; CHECK-NEXT: [[TMP4:%.*]] = sext i1 [[C17]] to i64
; CHECK-NEXT: [[G62:%.*]] = getelementptr i177, ptr [[G11]], i64 [[TMP4]]
@@ -1783,7 +1783,7 @@ define void @ashr_out_of_range_1(ptr %A) {
define void @ossfuzz_38078(i32 %arg, i32 %arg1, ptr %ptr, ptr %ptr2, ptr %ptr3, ptr %ptr4, ptr %ptr5, ptr %ptr6, ptr %ptr7) {
; CHECK-LABEL: @ossfuzz_38078(
; CHECK-NEXT: bb:
-; CHECK-NEXT: [[G1:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 -1
+; CHECK-NEXT: [[G1:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 -4
; CHECK-NEXT: [[I2:%.*]] = sub i32 0, [[ARG1:%.*]]
; CHECK-NEXT: [[I5:%.*]] = icmp eq i32 [[I2]], [[ARG:%.*]]
; CHECK-NEXT: call void @llvm.assume(i1 [[I5]])
diff --git a/llvm/test/Transforms/InstCombine/sink_sideeffecting_instruction.ll b/llvm/test/Transforms/InstCombine/sink_sideeffecting_instruction.ll
index f6262fd3c353ab..1da6903ae46664 100644
--- a/llvm/test/Transforms/InstCombine/sink_sideeffecting_instruction.ll
+++ b/llvm/test/Transforms/InstCombine/sink_sideeffecting_instruction.ll
@@ -187,7 +187,7 @@ define i32 @sink_gep1(i1 %c) {
; CHECK: early_return:
; CHECK-NEXT: ret i32 0
; CHECK: use_block:
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[VAR]], i64 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[VAR]], i64 4
; CHECK-NEXT: [[VAR3:%.*]] = call i32 @unknown(ptr nonnull [[GEP]]) #[[ATTR1]]
; CHECK-NEXT: ret i32 [[VAR3]]
;
diff --git a/llvm/test/Transforms/InstCombine/sprintf-2.ll b/llvm/test/Transforms/InstCombine/sprintf-2.ll
index 086be81bca9ec3..6caca2cc17825d 100644
--- a/llvm/test/Transforms/InstCombine/sprintf-2.ll
+++ b/llvm/test/Transforms/InstCombine/sprintf-2.ll
@@ -18,21 +18,21 @@ declare i32 @snprintf(ptr, i64, ptr, ...)
define void @fold_snprintf_member_pC(ptr %pi) {
; CHECK-LABEL: @fold_snprintf_member_pC(
; CHECK-NEXT: store i32 1, ptr [[PI:%.*]], align 4
-; CHECK-NEXT: [[PIA0AP1:%.*]] = getelementptr i32, ptr [[PI]], i64 1
+; CHECK-NEXT: [[PIA0AP1:%.*]] = getelementptr i8, ptr [[PI]], i64 4
; CHECK-NEXT: store i32 0, ptr [[PIA0AP1]], align 4
-; CHECK-NEXT: [[PIA0B:%.*]] = getelementptr i32, ptr [[PI]], i64 2
+; CHECK-NEXT: [[PIA0B:%.*]] = getelementptr i8, ptr [[PI]], i64 8
; CHECK-NEXT: store i32 2, ptr [[PIA0B]], align 4
-; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i32, ptr [[PI]], i64 3
+; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i8, ptr [[PI]], i64 12
; CHECK-NEXT: store i32 1, ptr [[PIA0BP1]], align 4
-; CHECK-NEXT: [[PIA0BP2:%.*]] = getelementptr i32, ptr [[PI]], i64 4
+; CHECK-NEXT: [[PIA0BP2:%.*]] = getelementptr i8, ptr [[PI]], i64 16
; CHECK-NEXT: store i32 0, ptr [[PIA0BP2]], align 4
-; CHECK-NEXT: [[PIA0C:%.*]] = getelementptr i32, ptr [[PI]], i64 5
+; CHECK-NEXT: [[PIA0C:%.*]] = getelementptr i8, ptr [[PI]], i64 20
; CHECK-NEXT: store i32 3, ptr [[PIA0C]], align 4
-; CHECK-NEXT: [[PIA1A:%.*]] = getelementptr i32, ptr [[PI]], i64 6
+; CHECK-NEXT: [[PIA1A:%.*]] = getelementptr i8, ptr [[PI]], i64 24
; CHECK-NEXT: store i32 4, ptr [[PIA1A]], align 4
-; CHECK-NEXT: [[PIA1B:%.*]] = getelementptr i32, ptr [[PI]], i64 7
+; CHECK-NEXT: [[PIA1B:%.*]] = getelementptr i8, ptr [[PI]], i64 28
; CHECK-NEXT: store i32 5, ptr [[PIA1B]], align 4
-; CHECK-NEXT: [[PIA1C:%.*]] = getelementptr i32, ptr [[PI]], i64 8
+; CHECK-NEXT: [[PIA1C:%.*]] = getelementptr i8, ptr [[PI]], i64 32
; CHECK-NEXT: store i32 6, ptr [[PIA1C]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/statepoint-cleanup.ll b/llvm/test/Transforms/InstCombine/statepoint-cleanup.ll
index 61c2b85553fd9a..95906f0f456d64 100644
--- a/llvm/test/Transforms/InstCombine/statepoint-cleanup.ll
+++ b/llvm/test/Transforms/InstCombine/statepoint-cleanup.ll
@@ -10,7 +10,7 @@ declare void @func()
define void @test(ptr addrspace(1) %b) gc "statepoint-example" {
; CHECK-LABEL: @test(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[D:%.*]] = getelementptr i32, ptr addrspace(1) [[B:%.*]], i64 16
+; CHECK-NEXT: [[D:%.*]] = getelementptr i8, ptr addrspace(1) [[B:%.*]], i64 64
; CHECK-NEXT: [[SAFEPOINT_TOKEN:%.*]] = tail call token (i64, i32, ptr, i32, i32, ...) @llvm.experimental.gc.statepoint.p0(i64 0, i32 0, ptr nonnull elementtype(void ()) @func, i32 0, i32 0, i32 0, i32 0) [ "gc-live"(ptr addrspace(1) [[B]], ptr addrspace(1) [[D]]) ]
; CHECK-NEXT: [[B_NEW_1:%.*]] = call ptr addrspace(1) @llvm.experimental.gc.relocate.p1(token [[SAFEPOINT_TOKEN]], i32 0, i32 0)
; CHECK-NEXT: [[B_NEW_2:%.*]] = call ptr addrspace(1) @llvm.experimental.gc.relocate.p1(token [[SAFEPOINT_TOKEN]], i32 0, i32 0)
@@ -68,7 +68,7 @@ entry:
define void @test_no_base_use(ptr addrspace(1) %b) gc "statepoint-example" {
; CHECK-LABEL: @test_no_base_use(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[D:%.*]] = getelementptr i32, ptr addrspace(1) [[B:%.*]], i64 16
+; CHECK-NEXT: [[D:%.*]] = getelementptr i8, ptr addrspace(1) [[B:%.*]], i64 64
; CHECK-NEXT: [[SAFEPOINT_TOKEN:%.*]] = tail call token (i64, i32, ptr, i32, i32, ...) @llvm.experimental.gc.statepoint.p0(i64 0, i32 0, ptr nonnull elementtype(void ()) @func, i32 0, i32 0, i32 0, i32 0) [ "gc-live"(ptr addrspace(1) [[B]], ptr addrspace(1) [[D]]) ]
; CHECK-NEXT: [[D_NEW_1:%.*]] = call ptr addrspace(1) @llvm.experimental.gc.relocate.p1(token [[SAFEPOINT_TOKEN]], i32 0, i32 1)
; CHECK-NEXT: store i32 1, ptr addrspace(1) [[D_NEW_1]], align 4
@@ -90,7 +90,7 @@ entry:
define void @test_invoke(ptr addrspace(1) %b) gc "statepoint-example" personality ptr @fake_personality_function {
; CHECK-LABEL: @test_invoke(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[D:%.*]] = getelementptr i32, ptr addrspace(1) [[B:%.*]], i64 16
+; CHECK-NEXT: [[D:%.*]] = getelementptr i8, ptr addrspace(1) [[B:%.*]], i64 64
; CHECK-NEXT: [[SAFEPOINT_TOKEN:%.*]] = invoke token (i64, i32, ptr, i32, i32, ...) @llvm.experimental.gc.statepoint.p0(i64 0, i32 0, ptr nonnull elementtype(void ()) @func, i32 0, i32 0, i32 0, i32 0) [ "gc-live"(ptr addrspace(1) [[B]], ptr addrspace(1) [[D]]) ]
; CHECK-NEXT: to label [[NORMAL_DEST:%.*]] unwind label [[UNWIND_DEST:%.*]]
; CHECK: normal_dest:
@@ -209,7 +209,7 @@ unwind_dest:
define void @test_no_base_use_invoke(ptr addrspace(1) %b) gc "statepoint-example" personality ptr @fake_personality_function {
; CHECK-LABEL: @test_no_base_use_invoke(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[D:%.*]] = getelementptr i32, ptr addrspace(1) [[B:%.*]], i64 16
+; CHECK-NEXT: [[D:%.*]] = getelementptr i8, ptr addrspace(1) [[B:%.*]], i64 64
; CHECK-NEXT: [[SAFEPOINT_TOKEN:%.*]] = invoke token (i64, i32, ptr, i32, i32, ...) @llvm.experimental.gc.statepoint.p0(i64 0, i32 0, ptr nonnull elementtype(void ()) @func, i32 0, i32 0, i32 0, i32 0) [ "gc-live"(ptr addrspace(1) [[B]], ptr addrspace(1) [[D]]) ]
; CHECK-NEXT: to label [[NORMAL_DEST:%.*]] unwind label [[UNWIND_DEST:%.*]]
; CHECK: normal_dest:
diff --git a/llvm/test/Transforms/InstCombine/str-int-3.ll b/llvm/test/Transforms/InstCombine/str-int-3.ll
index 7d46297b5d143b..f319a16d211f9a 100644
--- a/llvm/test/Transforms/InstCombine/str-int-3.ll
+++ b/llvm/test/Transforms/InstCombine/str-int-3.ll
@@ -21,11 +21,11 @@ declare i64 @strtoll(ptr, ptr, i32)
define void @fold_atoi_member(ptr %pi) {
; CHECK-LABEL: @fold_atoi_member(
; CHECK-NEXT: store i32 1, ptr [[PI:%.*]], align 4
-; CHECK-NEXT: [[PIA0B:%.*]] = getelementptr i32, ptr [[PI]], i64 1
+; CHECK-NEXT: [[PIA0B:%.*]] = getelementptr i8, ptr [[PI]], i64 4
; CHECK-NEXT: store i32 12, ptr [[PIA0B]], align 4
-; CHECK-NEXT: [[PIA1A:%.*]] = getelementptr i32, ptr [[PI]], i64 2
+; CHECK-NEXT: [[PIA1A:%.*]] = getelementptr i8, ptr [[PI]], i64 8
; CHECK-NEXT: store i32 123, ptr [[PIA1A]], align 4
-; CHECK-NEXT: [[PIA1B:%.*]] = getelementptr i32, ptr [[PI]], i64 3
+; CHECK-NEXT: [[PIA1B:%.*]] = getelementptr i8, ptr [[PI]], i64 12
; CHECK-NEXT: store i32 1234, ptr [[PIA1B]], align 4
; CHECK-NEXT: ret void
;
@@ -92,15 +92,15 @@ define void @fold_atoi_offset_out_of_bounds(ptr %pi) {
define void @fold_atol_member(ptr %pi) {
; CHECK-LABEL: @fold_atol_member(
; CHECK-NEXT: store i64 1, ptr [[PI:%.*]], align 4
-; CHECK-NEXT: [[PIA0B:%.*]] = getelementptr i64, ptr [[PI]], i64 1
+; CHECK-NEXT: [[PIA0B:%.*]] = getelementptr i8, ptr [[PI]], i64 8
; CHECK-NEXT: store i64 12, ptr [[PIA0B]], align 4
-; CHECK-NEXT: [[PIA0C:%.*]] = getelementptr i64, ptr [[PI]], i64 2
+; CHECK-NEXT: [[PIA0C:%.*]] = getelementptr i8, ptr [[PI]], i64 16
; CHECK-NEXT: store i64 56789, ptr [[PIA0C]], align 4
-; CHECK-NEXT: [[PIA1A:%.*]] = getelementptr i64, ptr [[PI]], i64 3
+; CHECK-NEXT: [[PIA1A:%.*]] = getelementptr i8, ptr [[PI]], i64 24
; CHECK-NEXT: store i64 123, ptr [[PIA1A]], align 4
-; CHECK-NEXT: [[PIA1B:%.*]] = getelementptr i64, ptr [[PI]], i64 4
+; CHECK-NEXT: [[PIA1B:%.*]] = getelementptr i8, ptr [[PI]], i64 32
; CHECK-NEXT: store i64 1234, ptr [[PIA1B]], align 4
-; CHECK-NEXT: [[PIA1C:%.*]] = getelementptr i64, ptr [[PI]], i64 5
+; CHECK-NEXT: [[PIA1C:%.*]] = getelementptr i8, ptr [[PI]], i64 40
; CHECK-NEXT: store i64 67890, ptr [[PIA1C]], align 4
; CHECK-NEXT: ret void
;
@@ -148,15 +148,15 @@ define void @fold_atol_member(ptr %pi) {
define void @fold_atoll_member_pC(ptr %pi) {
; CHECK-LABEL: @fold_atoll_member_pC(
; CHECK-NEXT: store i64 1, ptr [[PI:%.*]], align 4
-; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i64, ptr [[PI]], i64 1
+; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i8, ptr [[PI]], i64 8
; CHECK-NEXT: store i64 2, ptr [[PIA0BP1]], align 4
-; CHECK-NEXT: [[PIA0CP3:%.*]] = getelementptr i64, ptr [[PI]], i64 2
+; CHECK-NEXT: [[PIA0CP3:%.*]] = getelementptr i8, ptr [[PI]], i64 16
; CHECK-NEXT: store i64 89, ptr [[PIA0CP3]], align 4
-; CHECK-NEXT: [[PIA1AP2:%.*]] = getelementptr i64, ptr [[PI]], i64 3
+; CHECK-NEXT: [[PIA1AP2:%.*]] = getelementptr i8, ptr [[PI]], i64 24
; CHECK-NEXT: store i64 3, ptr [[PIA1AP2]], align 4
-; CHECK-NEXT: [[PIA1BP3:%.*]] = getelementptr i64, ptr [[PI]], i64 4
+; CHECK-NEXT: [[PIA1BP3:%.*]] = getelementptr i8, ptr [[PI]], i64 32
; CHECK-NEXT: store i64 4, ptr [[PIA1BP3]], align 4
-; CHECK-NEXT: [[PIA1CP4:%.*]] = getelementptr i64, ptr [[PI]], i64 5
+; CHECK-NEXT: [[PIA1CP4:%.*]] = getelementptr i8, ptr [[PI]], i64 40
; CHECK-NEXT: store i64 0, ptr [[PIA1CP4]], align 4
; CHECK-NEXT: ret void
;
@@ -204,15 +204,15 @@ define void @fold_atoll_member_pC(ptr %pi) {
define void @fold_strtol_member_pC(ptr %pi) {
; CHECK-LABEL: @fold_strtol_member_pC(
; CHECK-NEXT: store i64 1, ptr [[PI:%.*]], align 4
-; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i64, ptr [[PI]], i64 1
+; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i8, ptr [[PI]], i64 8
; CHECK-NEXT: store i64 2, ptr [[PIA0BP1]], align 4
-; CHECK-NEXT: [[PIA0CP3:%.*]] = getelementptr i64, ptr [[PI]], i64 2
+; CHECK-NEXT: [[PIA0CP3:%.*]] = getelementptr i8, ptr [[PI]], i64 16
; CHECK-NEXT: store i64 89, ptr [[PIA0CP3]], align 4
-; CHECK-NEXT: [[PIA1AP2:%.*]] = getelementptr i64, ptr [[PI]], i64 3
+; CHECK-NEXT: [[PIA1AP2:%.*]] = getelementptr i8, ptr [[PI]], i64 24
; CHECK-NEXT: store i64 3, ptr [[PIA1AP2]], align 4
-; CHECK-NEXT: [[PIA1BP3:%.*]] = getelementptr i64, ptr [[PI]], i64 4
+; CHECK-NEXT: [[PIA1BP3:%.*]] = getelementptr i8, ptr [[PI]], i64 32
; CHECK-NEXT: store i64 4, ptr [[PIA1BP3]], align 4
-; CHECK-NEXT: [[PIA1CP4:%.*]] = getelementptr i64, ptr [[PI]], i64 5
+; CHECK-NEXT: [[PIA1CP4:%.*]] = getelementptr i8, ptr [[PI]], i64 40
; CHECK-NEXT: store i64 0, ptr [[PIA1CP4]], align 4
; CHECK-NEXT: ret void
;
@@ -260,15 +260,15 @@ define void @fold_strtol_member_pC(ptr %pi) {
define void @fold_strtoll_member_pC(ptr %pi) {
; CHECK-LABEL: @fold_strtoll_member_pC(
; CHECK-NEXT: store i64 1, ptr [[PI:%.*]], align 4
-; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i64, ptr [[PI]], i64 1
+; CHECK-NEXT: [[PIA0BP1:%.*]] = getelementptr i8, ptr [[PI]], i64 8
; CHECK-NEXT: store i64 2, ptr [[PIA0BP1]], align 4
-; CHECK-NEXT: [[PIA0CP3:%.*]] = getelementptr i64, ptr [[PI]], i64 2
+; CHECK-NEXT: [[PIA0CP3:%.*]] = getelementptr i8, ptr [[PI]], i64 16
; CHECK-NEXT: store i64 89, ptr [[PIA0CP3]], align 4
-; CHECK-NEXT: [[PIA1AP2:%.*]] = getelementptr i64, ptr [[PI]], i64 3
+; CHECK-NEXT: [[PIA1AP2:%.*]] = getelementptr i8, ptr [[PI]], i64 24
; CHECK-NEXT: store i64 3, ptr [[PIA1AP2]], align 4
-; CHECK-NEXT: [[PIA1BP3:%.*]] = getelementptr i64, ptr [[PI]], i64 4
+; CHECK-NEXT: [[PIA1BP3:%.*]] = getelementptr i8, ptr [[PI]], i64 32
; CHECK-NEXT: store i64 4, ptr [[PIA1BP3]], align 4
-; CHECK-NEXT: [[PIA1CP4:%.*]] = getelementptr i64, ptr [[PI]], i64 5
+; CHECK-NEXT: [[PIA1CP4:%.*]] = getelementptr i8, ptr [[PI]], i64 40
; CHECK-NEXT: store i64 0, ptr [[PIA1CP4]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/str-int-4.ll b/llvm/test/Transforms/InstCombine/str-int-4.ll
index 9d3302ca16085b..6efc5fb4ed1fbb 100644
--- a/llvm/test/Transforms/InstCombine/str-int-4.ll
+++ b/llvm/test/Transforms/InstCombine/str-int-4.ll
@@ -45,37 +45,37 @@ define void @fold_strtol(ptr %ps) {
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_im123, i64 0, i64 10), ptr @endptr, align 8
; CHECK-NEXT: store i32 -123, ptr [[PS:%.*]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_ip234, i64 0, i64 10), ptr @endptr, align 8
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i32, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 4
; CHECK-NEXT: store i32 234, ptr [[PS1]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([3 x i8], ptr @i0, i64 0, i64 2), ptr @endptr, align 8
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i32, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i32 0, ptr [[PS2]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([3 x i8], ptr @i9, i64 0, i64 2), ptr @endptr, align 8
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i32, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 12
; CHECK-NEXT: store i32 9, ptr [[PS3]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([3 x i8], ptr @ia, i64 0, i64 2), ptr @endptr, align 8
-; CHECK-NEXT: [[PS4:%.*]] = getelementptr i32, ptr [[PS]], i64 4
+; CHECK-NEXT: [[PS4:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i32 10, ptr [[PS4]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([7 x i8], ptr @i19azAZ, i64 0, i64 6), ptr @endptr, align 8
-; CHECK-NEXT: [[PS5:%.*]] = getelementptr i32, ptr [[PS]], i64 5
+; CHECK-NEXT: [[PS5:%.*]] = getelementptr i8, ptr [[PS]], i64 20
; CHECK-NEXT: store i32 76095035, ptr [[PS5]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @i32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS6:%.*]] = getelementptr i32, ptr [[PS]], i64 6
+; CHECK-NEXT: [[PS6:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i32 -2147483648, ptr [[PS6]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([15 x i8], ptr @mo32min, i64 0, i64 14), ptr @endptr, align 8
-; CHECK-NEXT: [[PS7:%.*]] = getelementptr i32, ptr [[PS]], i64 7
+; CHECK-NEXT: [[PS7:%.*]] = getelementptr i8, ptr [[PS]], i64 28
; CHECK-NEXT: store i32 -2147483648, ptr [[PS7]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @mx32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS8:%.*]] = getelementptr i32, ptr [[PS]], i64 8
+; CHECK-NEXT: [[PS8:%.*]] = getelementptr i8, ptr [[PS]], i64 32
; CHECK-NEXT: store i32 -2147483648, ptr [[PS8]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @mx32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS9:%.*]] = getelementptr i32, ptr [[PS]], i64 9
+; CHECK-NEXT: [[PS9:%.*]] = getelementptr i8, ptr [[PS]], i64 36
; CHECK-NEXT: store i32 -2147483648, ptr [[PS9]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([12 x i8], ptr @i32max, i64 0, i64 11), ptr @endptr, align 8
-; CHECK-NEXT: [[PS10:%.*]] = getelementptr i32, ptr [[PS]], i64 10
+; CHECK-NEXT: [[PS10:%.*]] = getelementptr i8, ptr [[PS]], i64 40
; CHECK-NEXT: store i32 2147483647, ptr [[PS10]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([12 x i8], ptr @x32max, i64 0, i64 11), ptr @endptr, align 8
-; CHECK-NEXT: [[PS11:%.*]] = getelementptr i32, ptr [[PS]], i64 11
+; CHECK-NEXT: [[PS11:%.*]] = getelementptr i8, ptr [[PS]], i64 44
; CHECK-NEXT: store i32 2147483647, ptr [[PS11]], align 4
; CHECK-NEXT: ret void
;
@@ -149,46 +149,46 @@ define void @call_strtol(ptr %ps) {
; CHECK-NEXT: [[MINM1:%.*]] = call i32 @strtol(ptr nonnull @i32min_m1, ptr nonnull @endptr, i32 10)
; CHECK-NEXT: store i32 [[MINM1]], ptr [[PS:%.*]], align 4
; CHECK-NEXT: [[MAXP1:%.*]] = call i32 @strtol(ptr nonnull @i32max_p1, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i32, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 4
; CHECK-NEXT: store i32 [[MAXP1]], ptr [[PS1]], align 4
; CHECK-NEXT: [[IPLUS:%.*]] = call i32 @strtol(ptr nonnull @wsplus, ptr nonnull @endptr, i32 0)
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i32, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i32 [[IPLUS]], ptr [[PS2]], align 4
; CHECK-NEXT: [[IA:%.*]] = call i32 @strtol(ptr nonnull @ia, ptr nonnull @endptr, i32 0)
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i32, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 12
; CHECK-NEXT: store i32 [[IA]], ptr [[PS3]], align 4
; CHECK-NEXT: [[I8:%.*]] = call i32 @strtol(ptr nonnull @i8, ptr nonnull @endptr, i32 8)
-; CHECK-NEXT: [[PS4:%.*]] = getelementptr i32, ptr [[PS]], i64 4
+; CHECK-NEXT: [[PS4:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i32 [[I8]], ptr [[PS4]], align 4
; CHECK-NEXT: [[I0X:%.*]] = call i32 @strtol(ptr nonnull @x0x, ptr nonnull @endptr, i32 0)
-; CHECK-NEXT: [[PS5:%.*]] = getelementptr i32, ptr [[PS]], i64 5
+; CHECK-NEXT: [[PS5:%.*]] = getelementptr i8, ptr [[PS]], i64 20
; CHECK-NEXT: store i32 [[I0X]], ptr [[PS5]], align 4
; CHECK-NEXT: [[IWSPWS0:%.*]] = call i32 @strtol(ptr nonnull @wsplusws0, ptr nonnull @endptr, i32 0)
-; CHECK-NEXT: [[PS6:%.*]] = getelementptr i32, ptr [[PS]], i64 6
+; CHECK-NEXT: [[PS6:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i32 [[IWSPWS0]], ptr [[PS6]], align 4
; CHECK-NEXT: [[I19AZAZ:%.*]] = call i32 @strtol(ptr nonnull @i19azAZ, ptr nonnull @endptr, i32 35)
-; CHECK-NEXT: [[PS7:%.*]] = getelementptr i32, ptr [[PS]], i64 7
+; CHECK-NEXT: [[PS7:%.*]] = getelementptr i8, ptr [[PS]], i64 28
; CHECK-NEXT: store i32 [[I19AZAZ]], ptr [[PS7]], align 4
; CHECK-NEXT: [[O32MIN:%.*]] = call i32 @strtol(ptr nonnull @o32min, ptr nonnull @endptr, i32 0)
-; CHECK-NEXT: [[PS8:%.*]] = getelementptr i32, ptr [[PS]], i64 8
+; CHECK-NEXT: [[PS8:%.*]] = getelementptr i8, ptr [[PS]], i64 32
; CHECK-NEXT: store i32 [[O32MIN]], ptr [[PS8]], align 4
; CHECK-NEXT: [[X32MIN:%.*]] = call i32 @strtol(ptr nonnull @x32min, ptr nonnull @endptr, i32 0)
-; CHECK-NEXT: [[PS9:%.*]] = getelementptr i32, ptr [[PS]], i64 9
+; CHECK-NEXT: [[PS9:%.*]] = getelementptr i8, ptr [[PS]], i64 36
; CHECK-NEXT: store i32 [[X32MIN]], ptr [[PS9]], align 4
; CHECK-NEXT: [[X32MIN_10:%.*]] = call i32 @strtol(ptr nonnull @x32min, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS10:%.*]] = getelementptr i32, ptr [[PS]], i64 10
+; CHECK-NEXT: [[PS10:%.*]] = getelementptr i8, ptr [[PS]], i64 40
; CHECK-NEXT: store i32 [[X32MIN_10]], ptr [[PS10]], align 4
; CHECK-NEXT: [[NWS:%.*]] = call i32 @strtol(ptr nonnull @ws, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS11:%.*]] = getelementptr i32, ptr [[PS]], i64 11
+; CHECK-NEXT: [[PS11:%.*]] = getelementptr i8, ptr [[PS]], i64 44
; CHECK-NEXT: store i32 [[NWS]], ptr [[PS11]], align 4
; CHECK-NEXT: [[NWSP6:%.*]] = call i32 @strtol(ptr nonnull getelementptr inbounds ([7 x i8], ptr @ws, i64 0, i64 6), ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS12:%.*]] = getelementptr i32, ptr [[PS]], i64 12
+; CHECK-NEXT: [[PS12:%.*]] = getelementptr i8, ptr [[PS]], i64 48
; CHECK-NEXT: store i32 [[NWSP6]], ptr [[PS12]], align 4
; CHECK-NEXT: [[I0B1:%.*]] = call i32 @strtol(ptr nonnull @i0, ptr nonnull @endptr, i32 1)
-; CHECK-NEXT: [[PS13:%.*]] = getelementptr i32, ptr [[PS]], i64 13
+; CHECK-NEXT: [[PS13:%.*]] = getelementptr i8, ptr [[PS]], i64 52
; CHECK-NEXT: store i32 [[I0B1]], ptr [[PS13]], align 4
; CHECK-NEXT: [[I0B256:%.*]] = call i32 @strtol(ptr nonnull @i0, ptr nonnull @endptr, i32 256)
-; CHECK-NEXT: [[PS14:%.*]] = getelementptr i32, ptr [[PS]], i64 14
+; CHECK-NEXT: [[PS14:%.*]] = getelementptr i8, ptr [[PS]], i64 56
; CHECK-NEXT: store i32 [[I0B256]], ptr [[PS14]], align 4
; CHECK-NEXT: ret void
;
@@ -290,13 +290,13 @@ define void @fold_strtoll(ptr %ps) {
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_im123, i64 0, i64 10), ptr @endptr, align 8
; CHECK-NEXT: store i64 -123, ptr [[PS:%.*]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_ip234, i64 0, i64 10), ptr @endptr, align 8
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i64, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i64 234, ptr [[PS1]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([22 x i8], ptr @i64min, i64 0, i64 21), ptr @endptr, align 8
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i64, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i64 -9223372036854775808, ptr [[PS2]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([21 x i8], ptr @i64max, i64 0, i64 20), ptr @endptr, align 8
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i64, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i64 9223372036854775807, ptr [[PS3]], align 4
; CHECK-NEXT: ret void
;
@@ -330,13 +330,13 @@ define void @call_strtoll(ptr %ps) {
; CHECK-NEXT: [[MINM1:%.*]] = call i64 @strtoll(ptr nonnull @i64min_m1, ptr nonnull @endptr, i32 10)
; CHECK-NEXT: store i64 [[MINM1]], ptr [[PS:%.*]], align 4
; CHECK-NEXT: [[MAXP1:%.*]] = call i64 @strtoll(ptr nonnull @i64max_p1, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i64, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i64 [[MAXP1]], ptr [[PS1]], align 4
; CHECK-NEXT: [[NWS:%.*]] = call i64 @strtoll(ptr nonnull @ws, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i64, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i64 [[NWS]], ptr [[PS2]], align 4
; CHECK-NEXT: [[NWSP6:%.*]] = call i64 @strtoll(ptr nonnull getelementptr inbounds ([7 x i8], ptr @ws, i64 0, i64 6), ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i64, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i64 [[NWSP6]], ptr [[PS3]], align 4
; CHECK-NEXT: ret void
;
@@ -373,13 +373,13 @@ define void @call_strtoll(ptr %ps) {
define void @call_strtol_trailing_space(ptr %ps) {
; CHECK-LABEL: @call_strtol_trailing_space(
; CHECK-NEXT: [[N1:%.*]] = call i32 @strtol(ptr nonnull @i_1_2_3_, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i32, ptr [[PS:%.*]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS:%.*]], i64 4
; CHECK-NEXT: store i32 [[N1]], ptr [[PS1]], align 4
; CHECK-NEXT: [[N2:%.*]] = call i32 @strtol(ptr nonnull getelementptr inbounds ([9 x i8], ptr @i_1_2_3_, i64 0, i64 2), ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i32, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i32 [[N2]], ptr [[PS2]], align 4
; CHECK-NEXT: [[N3:%.*]] = call i32 @strtol(ptr nonnull getelementptr inbounds ([9 x i8], ptr @i_1_2_3_, i64 0, i64 4), ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i32, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 12
; CHECK-NEXT: store i32 [[N3]], ptr [[PS3]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/str-int-5.ll b/llvm/test/Transforms/InstCombine/str-int-5.ll
index f8976bcb827576..ff4f2bffd9776a 100644
--- a/llvm/test/Transforms/InstCombine/str-int-5.ll
+++ b/llvm/test/Transforms/InstCombine/str-int-5.ll
@@ -49,37 +49,37 @@ define void @fold_strtoul(ptr %ps) {
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_im123, i64 0, i64 10), ptr @endptr, align 8
; CHECK-NEXT: store i32 -123, ptr [[PS:%.*]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_ip234, i64 0, i64 10), ptr @endptr, align 8
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i32, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 4
; CHECK-NEXT: store i32 234, ptr [[PS1]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @i32min_m1, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i32, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i32 2147483647, ptr [[PS2]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @i32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i32, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 12
; CHECK-NEXT: store i32 -2147483648, ptr [[PS3]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([15 x i8], ptr @o32min, i64 0, i64 14), ptr @endptr, align 8
-; CHECK-NEXT: [[PS4:%.*]] = getelementptr i32, ptr [[PS]], i64 4
+; CHECK-NEXT: [[PS4:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i32 -2147483648, ptr [[PS4]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([15 x i8], ptr @mo32min, i64 0, i64 14), ptr @endptr, align 8
-; CHECK-NEXT: [[PS5:%.*]] = getelementptr i32, ptr [[PS]], i64 5
+; CHECK-NEXT: [[PS5:%.*]] = getelementptr i8, ptr [[PS]], i64 20
; CHECK-NEXT: store i32 -2147483648, ptr [[PS5]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @x32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS6:%.*]] = getelementptr i32, ptr [[PS]], i64 6
+; CHECK-NEXT: [[PS6:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i32 -2147483648, ptr [[PS6]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @mx32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS7:%.*]] = getelementptr i32, ptr [[PS]], i64 7
+; CHECK-NEXT: [[PS7:%.*]] = getelementptr i8, ptr [[PS]], i64 28
; CHECK-NEXT: store i32 -2147483648, ptr [[PS7]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([12 x i8], ptr @i32max, i64 0, i64 11), ptr @endptr, align 8
-; CHECK-NEXT: [[PS8:%.*]] = getelementptr i32, ptr [[PS]], i64 8
+; CHECK-NEXT: [[PS8:%.*]] = getelementptr i8, ptr [[PS]], i64 32
; CHECK-NEXT: store i32 2147483647, ptr [[PS8]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([6 x i8], ptr @mX01, i64 0, i64 5), ptr @endptr, align 8
-; CHECK-NEXT: [[PS9:%.*]] = getelementptr i32, ptr [[PS]], i64 9
+; CHECK-NEXT: [[PS9:%.*]] = getelementptr i8, ptr [[PS]], i64 36
; CHECK-NEXT: store i32 -1, ptr [[PS9]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([12 x i8], ptr @i32max_p1, i64 0, i64 11), ptr @endptr, align 8
-; CHECK-NEXT: [[PS10:%.*]] = getelementptr i32, ptr [[PS]], i64 10
+; CHECK-NEXT: [[PS10:%.*]] = getelementptr i8, ptr [[PS]], i64 40
; CHECK-NEXT: store i32 -2147483648, ptr [[PS10]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([12 x i8], ptr @ui32max, i64 0, i64 11), ptr @endptr, align 8
-; CHECK-NEXT: [[PS11:%.*]] = getelementptr i32, ptr [[PS]], i64 11
+; CHECK-NEXT: [[PS11:%.*]] = getelementptr i8, ptr [[PS]], i64 44
; CHECK-NEXT: store i32 -1, ptr [[PS11]], align 4
; CHECK-NEXT: ret void
;
@@ -154,13 +154,13 @@ define void @call_strtoul(ptr %ps) {
; CHECK-NEXT: [[MINM1:%.*]] = call i32 @strtoul(ptr nonnull @i64min_m1, ptr nonnull @endptr, i32 10)
; CHECK-NEXT: store i32 [[MINM1]], ptr [[PS:%.*]], align 4
; CHECK-NEXT: [[MAXP1:%.*]] = call i32 @strtoul(ptr nonnull @ui32max_p1, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i32, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 4
; CHECK-NEXT: store i32 [[MAXP1]], ptr [[PS1]], align 4
; CHECK-NEXT: [[NWS:%.*]] = call i32 @strtoul(ptr nonnull @ws, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i32, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i32 [[NWS]], ptr [[PS2]], align 4
; CHECK-NEXT: [[NWSP6:%.*]] = call i32 @strtoul(ptr nonnull getelementptr inbounds ([7 x i8], ptr @ws, i64 0, i64 6), ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i32, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 12
; CHECK-NEXT: store i32 [[NWSP6]], ptr [[PS3]], align 4
; CHECK-NEXT: ret void
;
@@ -198,34 +198,34 @@ define void @fold_strtoull(ptr %ps) {
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_im123, i64 0, i64 10), ptr @endptr, align 8
; CHECK-NEXT: store i64 -123, ptr [[PS:%.*]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([11 x i8], ptr @ws_ip234, i64 0, i64 10), ptr @endptr, align 8
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i64, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i64 234, ptr [[PS1]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([22 x i8], ptr @i64min_m1, i64 0, i64 21), ptr @endptr, align 8
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i64, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i64 9223372036854775807, ptr [[PS2]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @i32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i64, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i64 -2147483648, ptr [[PS3]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([15 x i8], ptr @o32min, i64 0, i64 14), ptr @endptr, align 8
-; CHECK-NEXT: [[PS4:%.*]] = getelementptr i64, ptr [[PS]], i64 4
+; CHECK-NEXT: [[PS4:%.*]] = getelementptr i8, ptr [[PS]], i64 32
; CHECK-NEXT: store i64 2147483648, ptr [[PS4]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([13 x i8], ptr @x32min, i64 0, i64 12), ptr @endptr, align 8
-; CHECK-NEXT: [[PS5:%.*]] = getelementptr i64, ptr [[PS]], i64 5
+; CHECK-NEXT: [[PS5:%.*]] = getelementptr i8, ptr [[PS]], i64 40
; CHECK-NEXT: store i64 2147483648, ptr [[PS5]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([22 x i8], ptr @i64min, i64 0, i64 21), ptr @endptr, align 8
-; CHECK-NEXT: [[PS6:%.*]] = getelementptr i64, ptr [[PS]], i64 6
+; CHECK-NEXT: [[PS6:%.*]] = getelementptr i8, ptr [[PS]], i64 48
; CHECK-NEXT: store i64 -9223372036854775808, ptr [[PS6]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([21 x i8], ptr @i64max, i64 0, i64 20), ptr @endptr, align 8
-; CHECK-NEXT: [[PS7:%.*]] = getelementptr i64, ptr [[PS]], i64 7
+; CHECK-NEXT: [[PS7:%.*]] = getelementptr i8, ptr [[PS]], i64 56
; CHECK-NEXT: store i64 9223372036854775807, ptr [[PS7]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([21 x i8], ptr @i64max_p1, i64 0, i64 20), ptr @endptr, align 8
-; CHECK-NEXT: [[PS8:%.*]] = getelementptr i64, ptr [[PS]], i64 8
+; CHECK-NEXT: [[PS8:%.*]] = getelementptr i8, ptr [[PS]], i64 64
; CHECK-NEXT: store i64 -9223372036854775808, ptr [[PS8]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([22 x i8], ptr @ui64max, i64 0, i64 21), ptr @endptr, align 8
-; CHECK-NEXT: [[PS9:%.*]] = getelementptr i64, ptr [[PS]], i64 9
+; CHECK-NEXT: [[PS9:%.*]] = getelementptr i8, ptr [[PS]], i64 72
; CHECK-NEXT: store i64 -1, ptr [[PS9]], align 4
; CHECK-NEXT: store ptr getelementptr inbounds ([20 x i8], ptr @x64max, i64 0, i64 19), ptr @endptr, align 8
-; CHECK-NEXT: [[PS10:%.*]] = getelementptr i64, ptr [[PS]], i64 10
+; CHECK-NEXT: [[PS10:%.*]] = getelementptr i8, ptr [[PS]], i64 80
; CHECK-NEXT: store i64 -1, ptr [[PS10]], align 4
; CHECK-NEXT: ret void
;
@@ -293,13 +293,13 @@ define void @fold_strtoull(ptr %ps) {
define void @call_strtoull(ptr %ps) {
; CHECK-LABEL: @call_strtoull(
; CHECK-NEXT: [[MAXP1:%.*]] = call i64 @strtoull(ptr nonnull @ui64max_p1, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i64, ptr [[PS:%.*]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS:%.*]], i64 8
; CHECK-NEXT: store i64 [[MAXP1]], ptr [[PS1]], align 4
; CHECK-NEXT: [[NWS:%.*]] = call i64 @strtoull(ptr nonnull @ws, ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr i64, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i64 [[NWS]], ptr [[PS2]], align 4
; CHECK-NEXT: [[NWSP6:%.*]] = call i64 @strtoull(ptr nonnull getelementptr inbounds ([7 x i8], ptr @ws, i64 0, i64 6), ptr nonnull @endptr, i32 10)
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i64, ptr [[PS]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i64 [[NWSP6]], ptr [[PS3]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/str-int.ll b/llvm/test/Transforms/InstCombine/str-int.ll
index 8b4126b5c95762..718bfe4133332a 100644
--- a/llvm/test/Transforms/InstCombine/str-int.ll
+++ b/llvm/test/Transforms/InstCombine/str-int.ll
@@ -45,7 +45,7 @@ define i32 @strtol_hex() #0 {
define i32 @strtol_endptr_not_null(ptr %pend) {
; CHECK-LABEL: @strtol_endptr_not_null(
-; CHECK-NEXT: [[ENDP1:%.*]] = getelementptr inbounds ptr, ptr [[PEND:%.*]], i64 1
+; CHECK-NEXT: [[ENDP1:%.*]] = getelementptr inbounds i8, ptr [[PEND:%.*]], i64 8
; CHECK-NEXT: store ptr getelementptr inbounds ([3 x i8], ptr @.str, i64 0, i64 2), ptr [[ENDP1]], align 8
; CHECK-NEXT: ret i32 12
;
diff --git a/llvm/test/Transforms/InstCombine/strcall-bad-sig.ll b/llvm/test/Transforms/InstCombine/strcall-bad-sig.ll
index 98f5e6ea1b27c2..5e59db5ef88aa5 100644
--- a/llvm/test/Transforms/InstCombine/strcall-bad-sig.ll
+++ b/llvm/test/Transforms/InstCombine/strcall-bad-sig.ll
@@ -15,10 +15,10 @@ define void @call_bad_ato(ptr %ps) {
; CHECK-NEXT: [[IR:%.*]] = call ptr @atoi(ptr nonnull @a)
; CHECK-NEXT: store ptr [[IR]], ptr [[PS:%.*]], align 8
; CHECK-NEXT: [[LR:%.*]] = call ptr @atol(ptr nonnull @a)
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr ptr, ptr [[PS]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store ptr [[LR]], ptr [[PS1]], align 8
; CHECK-NEXT: [[LLR:%.*]] = call ptr @atol(ptr nonnull @a)
-; CHECK-NEXT: [[PS2:%.*]] = getelementptr ptr, ptr [[PS]], i64 2
+; CHECK-NEXT: [[PS2:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store ptr [[LLR]], ptr [[PS2]], align 8
; CHECK-NEXT: ret void
;
@@ -114,12 +114,12 @@ define void @call_bad_strto(ptr %psi32, ptr %psi64) {
; CHECK-NEXT: [[LR:%.*]] = call i32 @strtol(ptr nonnull @a, ptr null)
; CHECK-NEXT: store i32 [[LR]], ptr [[PSI32:%.*]], align 4
; CHECK-NEXT: [[ULR:%.*]] = call i32 @strtoul(ptr nonnull @a, ptr null)
-; CHECK-NEXT: [[PS1:%.*]] = getelementptr i32, ptr [[PSI32]], i64 1
+; CHECK-NEXT: [[PS1:%.*]] = getelementptr i8, ptr [[PSI32]], i64 4
; CHECK-NEXT: store i32 [[ULR]], ptr [[PS1]], align 4
; CHECK-NEXT: [[LLR:%.*]] = call i64 @strtoll(ptr nonnull @a, ptr null)
; CHECK-NEXT: store i64 [[LLR]], ptr [[PSI64:%.*]], align 4
; CHECK-NEXT: [[ULLR:%.*]] = call i64 @strtoull(ptr nonnull @a, ptr null)
-; CHECK-NEXT: [[PS3:%.*]] = getelementptr i64, ptr [[PSI64]], i64 3
+; CHECK-NEXT: [[PS3:%.*]] = getelementptr i8, ptr [[PSI64]], i64 24
; CHECK-NEXT: store i64 [[ULLR]], ptr [[PS3]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/strcall-no-nul.ll b/llvm/test/Transforms/InstCombine/strcall-no-nul.ll
index fef06a03639c7c..30221ad5b0962a 100644
--- a/llvm/test/Transforms/InstCombine/strcall-no-nul.ll
+++ b/llvm/test/Transforms/InstCombine/strcall-no-nul.ll
@@ -62,7 +62,7 @@ define ptr @fold_strchr_past_end() {
define void @fold_strcmp_past_end(ptr %pcmp) {
; CHECK-LABEL: @fold_strcmp_past_end(
; CHECK-NEXT: store i32 1, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[PC50:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[PC50:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 -1, ptr [[PC50]], align 4
; CHECK-NEXT: ret void
;
@@ -84,7 +84,7 @@ define void @fold_strcmp_past_end(ptr %pcmp) {
define void @fold_strncmp_past_end(ptr %pcmp) {
; CHECK-LABEL: @fold_strncmp_past_end(
; CHECK-NEXT: store i32 1, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[PC50:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[PC50:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 -1, ptr [[PC50]], align 4
; CHECK-NEXT: ret void
;
@@ -118,7 +118,7 @@ define ptr @fold_strrchr_past_end(i32 %c) {
define void @fold_strstr_past_end(ptr %psub) {
; CHECK-LABEL: @fold_strstr_past_end(
; CHECK-NEXT: store ptr @a5, ptr [[PSUB:%.*]], align 8
-; CHECK-NEXT: [[PS50:%.*]] = getelementptr ptr, ptr [[PSUB]], i64 1
+; CHECK-NEXT: [[PS50:%.*]] = getelementptr i8, ptr [[PSUB]], i64 8
; CHECK-NEXT: store ptr null, ptr [[PS50]], align 8
; CHECK-NEXT: ret void
;
@@ -202,7 +202,7 @@ define ptr @fold_strncpy_past_end(ptr %dst) {
define void @fold_strpbrk_past_end(ptr %psub) {
; CHECK-LABEL: @fold_strpbrk_past_end(
; CHECK-NEXT: store ptr null, ptr [[PSUB:%.*]], align 8
-; CHECK-NEXT: [[PS50:%.*]] = getelementptr ptr, ptr [[PSUB]], i64 1
+; CHECK-NEXT: [[PS50:%.*]] = getelementptr i8, ptr [[PSUB]], i64 8
; CHECK-NEXT: store ptr null, ptr [[PS50]], align 8
; CHECK-NEXT: ret void
;
@@ -224,7 +224,7 @@ define void @fold_strpbrk_past_end(ptr %psub) {
define void @fold_strspn_past_end(ptr %poff) {
; CHECK-LABEL: @fold_strspn_past_end(
; CHECK-NEXT: store i64 0, ptr [[POFF:%.*]], align 4
-; CHECK-NEXT: [[PO50:%.*]] = getelementptr i64, ptr [[POFF]], i64 1
+; CHECK-NEXT: [[PO50:%.*]] = getelementptr i8, ptr [[POFF]], i64 8
; CHECK-NEXT: store i64 0, ptr [[PO50]], align 4
; CHECK-NEXT: ret void
;
@@ -246,7 +246,7 @@ define void @fold_strspn_past_end(ptr %poff) {
define void @fold_strcspn_past_end(ptr %poff) {
; CHECK-LABEL: @fold_strcspn_past_end(
; CHECK-NEXT: store i64 2, ptr [[POFF:%.*]], align 4
-; CHECK-NEXT: [[PO50:%.*]] = getelementptr i64, ptr [[POFF]], i64 1
+; CHECK-NEXT: [[PO50:%.*]] = getelementptr i8, ptr [[POFF]], i64 8
; CHECK-NEXT: store i64 0, ptr [[PO50]], align 4
; CHECK-NEXT: ret void
;
@@ -285,19 +285,19 @@ define void @fold_atol_strtol_past_end(ptr %ps) {
; CHECK-NEXT: [[I0:%.*]] = call i64 @atol(ptr nocapture nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0))
; CHECK-NEXT: store i64 [[I0]], ptr [[PS:%.*]], align 4
; CHECK-NEXT: [[I1:%.*]] = call i64 @atoll(ptr nocapture nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0))
-; CHECK-NEXT: [[P1:%.*]] = getelementptr i64, ptr [[PS]], i64 1
+; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[PS]], i64 8
; CHECK-NEXT: store i64 [[I1]], ptr [[P1]], align 4
; CHECK-NEXT: [[I2:%.*]] = call i64 @strtol(ptr nocapture nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0), ptr null, i32 0)
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i64, ptr [[PS]], i64 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[PS]], i64 16
; CHECK-NEXT: store i64 [[I2]], ptr [[P2]], align 4
; CHECK-NEXT: [[I3:%.*]] = call i64 @strtoul(ptr nocapture nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0), ptr null, i32 8)
-; CHECK-NEXT: [[P3:%.*]] = getelementptr i64, ptr [[PS]], i64 3
+; CHECK-NEXT: [[P3:%.*]] = getelementptr i8, ptr [[PS]], i64 24
; CHECK-NEXT: store i64 [[I3]], ptr [[P3]], align 4
; CHECK-NEXT: [[I4:%.*]] = call i64 @strtoll(ptr nocapture nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0), ptr null, i32 10)
-; CHECK-NEXT: [[P4:%.*]] = getelementptr i64, ptr [[PS]], i64 4
+; CHECK-NEXT: [[P4:%.*]] = getelementptr i8, ptr [[PS]], i64 32
; CHECK-NEXT: store i64 [[I4]], ptr [[P4]], align 4
; CHECK-NEXT: [[I5:%.*]] = call i64 @strtoul(ptr nocapture nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0), ptr null, i32 16)
-; CHECK-NEXT: [[P5:%.*]] = getelementptr i64, ptr [[PS]], i64 5
+; CHECK-NEXT: [[P5:%.*]] = getelementptr i8, ptr [[PS]], i64 40
; CHECK-NEXT: store i64 [[I5]], ptr [[P5]], align 4
; CHECK-NEXT: ret void
;
@@ -336,7 +336,7 @@ define void @fold_atol_strtol_past_end(ptr %ps) {
define void @fold_sprintf_past_end(ptr %pcnt, ptr %dst) {
; CHECK-LABEL: @fold_sprintf_past_end(
; CHECK-NEXT: store i32 0, ptr [[PCNT:%.*]], align 4
-; CHECK-NEXT: [[PN05:%.*]] = getelementptr i32, ptr [[PCNT]], i64 1
+; CHECK-NEXT: [[PN05:%.*]] = getelementptr i8, ptr [[PCNT]], i64 4
; CHECK-NEXT: store i32 0, ptr [[PN05]], align 4
; CHECK-NEXT: ret void
;
@@ -361,7 +361,7 @@ define void @fold_snprintf_past_end(ptr %pcnt, ptr %dst, i64 %n) {
; CHECK-NEXT: [[N5_:%.*]] = call i32 (ptr, i64, ptr, ...) @snprintf(ptr [[DST:%.*]], i64 [[N:%.*]], ptr nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0))
; CHECK-NEXT: store i32 [[N5_]], ptr [[PCNT:%.*]], align 4
; CHECK-NEXT: [[N05:%.*]] = call i32 (ptr, i64, ptr, ...) @snprintf(ptr [[DST]], i64 [[N]], ptr nonnull @a5, ptr nonnull getelementptr inbounds ([5 x i8], ptr @a5, i64 1, i64 0))
-; CHECK-NEXT: [[PN05:%.*]] = getelementptr i32, ptr [[PCNT]], i64 1
+; CHECK-NEXT: [[PN05:%.*]] = getelementptr i8, ptr [[PCNT]], i64 4
; CHECK-NEXT: store i32 [[N05]], ptr [[PN05]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/strlen-7.ll b/llvm/test/Transforms/InstCombine/strlen-7.ll
index 1a12a791329b52..6eb7885abd0416 100644
--- a/llvm/test/Transforms/InstCombine/strlen-7.ll
+++ b/llvm/test/Transforms/InstCombine/strlen-7.ll
@@ -16,39 +16,39 @@ declare i64 @strlen(ptr)
define void @fold_strlen_A(ptr %plen) {
; CHECK-LABEL: @fold_strlen_A(
; CHECK-NEXT: store i64 1, ptr [[PLEN:%.*]], align 4
-; CHECK-NEXT: [[PLEN1:%.*]] = getelementptr i64, ptr [[PLEN]], i64 1
+; CHECK-NEXT: [[PLEN1:%.*]] = getelementptr i8, ptr [[PLEN]], i64 8
; CHECK-NEXT: store i64 0, ptr [[PLEN1]], align 4
-; CHECK-NEXT: [[PLEN2:%.*]] = getelementptr i64, ptr [[PLEN]], i64 2
+; CHECK-NEXT: [[PLEN2:%.*]] = getelementptr i8, ptr [[PLEN]], i64 16
; CHECK-NEXT: store i64 0, ptr [[PLEN2]], align 4
-; CHECK-NEXT: [[PLEN3:%.*]] = getelementptr i64, ptr [[PLEN]], i64 3
+; CHECK-NEXT: [[PLEN3:%.*]] = getelementptr i8, ptr [[PLEN]], i64 24
; CHECK-NEXT: store i64 0, ptr [[PLEN3]], align 4
-; CHECK-NEXT: [[PLEN4:%.*]] = getelementptr i64, ptr [[PLEN]], i64 4
+; CHECK-NEXT: [[PLEN4:%.*]] = getelementptr i8, ptr [[PLEN]], i64 32
; CHECK-NEXT: store i64 2, ptr [[PLEN4]], align 4
-; CHECK-NEXT: [[PLEN5:%.*]] = getelementptr i64, ptr [[PLEN]], i64 5
+; CHECK-NEXT: [[PLEN5:%.*]] = getelementptr i8, ptr [[PLEN]], i64 40
; CHECK-NEXT: store i64 1, ptr [[PLEN5]], align 4
-; CHECK-NEXT: [[PLEN6:%.*]] = getelementptr i64, ptr [[PLEN]], i64 6
+; CHECK-NEXT: [[PLEN6:%.*]] = getelementptr i8, ptr [[PLEN]], i64 48
; CHECK-NEXT: store i64 0, ptr [[PLEN6]], align 4
-; CHECK-NEXT: [[PLEN7:%.*]] = getelementptr i64, ptr [[PLEN]], i64 7
+; CHECK-NEXT: [[PLEN7:%.*]] = getelementptr i8, ptr [[PLEN]], i64 56
; CHECK-NEXT: store i64 0, ptr [[PLEN7]], align 4
-; CHECK-NEXT: [[PLEN8:%.*]] = getelementptr i64, ptr [[PLEN]], i64 8
+; CHECK-NEXT: [[PLEN8:%.*]] = getelementptr i8, ptr [[PLEN]], i64 64
; CHECK-NEXT: store i64 0, ptr [[PLEN8]], align 4
-; CHECK-NEXT: [[PLEN9:%.*]] = getelementptr i64, ptr [[PLEN]], i64 9
+; CHECK-NEXT: [[PLEN9:%.*]] = getelementptr i8, ptr [[PLEN]], i64 72
; CHECK-NEXT: store i64 3, ptr [[PLEN9]], align 4
-; CHECK-NEXT: [[PLEN10:%.*]] = getelementptr i64, ptr [[PLEN]], i64 10
+; CHECK-NEXT: [[PLEN10:%.*]] = getelementptr i8, ptr [[PLEN]], i64 80
; CHECK-NEXT: store i64 2, ptr [[PLEN10]], align 4
-; CHECK-NEXT: [[PLEN11:%.*]] = getelementptr i64, ptr [[PLEN]], i64 11
+; CHECK-NEXT: [[PLEN11:%.*]] = getelementptr i8, ptr [[PLEN]], i64 88
; CHECK-NEXT: store i64 1, ptr [[PLEN11]], align 4
-; CHECK-NEXT: [[PLEN12:%.*]] = getelementptr i64, ptr [[PLEN]], i64 12
+; CHECK-NEXT: [[PLEN12:%.*]] = getelementptr i8, ptr [[PLEN]], i64 96
; CHECK-NEXT: store i64 0, ptr [[PLEN12]], align 4
-; CHECK-NEXT: [[PLEN14:%.*]] = getelementptr i64, ptr [[PLEN]], i64 14
+; CHECK-NEXT: [[PLEN14:%.*]] = getelementptr i8, ptr [[PLEN]], i64 112
; CHECK-NEXT: store i64 4, ptr [[PLEN14]], align 4
-; CHECK-NEXT: [[PLEN15:%.*]] = getelementptr i64, ptr [[PLEN]], i64 15
+; CHECK-NEXT: [[PLEN15:%.*]] = getelementptr i8, ptr [[PLEN]], i64 120
; CHECK-NEXT: store i64 3, ptr [[PLEN15]], align 4
-; CHECK-NEXT: [[PLEN16:%.*]] = getelementptr i64, ptr [[PLEN]], i64 16
+; CHECK-NEXT: [[PLEN16:%.*]] = getelementptr i8, ptr [[PLEN]], i64 128
; CHECK-NEXT: store i64 2, ptr [[PLEN16]], align 4
-; CHECK-NEXT: [[PLEN17:%.*]] = getelementptr i64, ptr [[PLEN]], i64 17
+; CHECK-NEXT: [[PLEN17:%.*]] = getelementptr i8, ptr [[PLEN]], i64 136
; CHECK-NEXT: store i64 1, ptr [[PLEN17]], align 4
-; CHECK-NEXT: [[PLEN18:%.*]] = getelementptr i64, ptr [[PLEN]], i64 18
+; CHECK-NEXT: [[PLEN18:%.*]] = getelementptr i8, ptr [[PLEN]], i64 144
; CHECK-NEXT: store i64 0, ptr [[PLEN18]], align 4
; CHECK-NEXT: ret void
;
@@ -172,15 +172,15 @@ define void @fold_strlen_A_pI(ptr %plen, i64 %I) {
; CHECK-NEXT: store i64 [[LENA0A]], ptr [[PLEN:%.*]], align 4
; CHECK-NEXT: [[PA0B:%.*]] = getelementptr [2 x %struct.A], ptr @a, i64 0, i64 0, i32 1, i64 [[I]]
; CHECK-NEXT: [[LENA0B:%.*]] = call i64 @strlen(ptr noundef nonnull dereferenceable(1) [[PA0B]])
-; CHECK-NEXT: [[PLEN1:%.*]] = getelementptr i64, ptr [[PLEN]], i64 1
+; CHECK-NEXT: [[PLEN1:%.*]] = getelementptr i8, ptr [[PLEN]], i64 8
; CHECK-NEXT: store i64 [[LENA0B]], ptr [[PLEN1]], align 4
; CHECK-NEXT: [[PA1A:%.*]] = getelementptr [2 x %struct.A], ptr @a, i64 0, i64 1, i32 0, i64 [[I]]
; CHECK-NEXT: [[LENA1A:%.*]] = call i64 @strlen(ptr noundef nonnull dereferenceable(1) [[PA1A]])
-; CHECK-NEXT: [[PLEN2:%.*]] = getelementptr i64, ptr [[PLEN]], i64 2
+; CHECK-NEXT: [[PLEN2:%.*]] = getelementptr i8, ptr [[PLEN]], i64 16
; CHECK-NEXT: store i64 [[LENA1A]], ptr [[PLEN2]], align 4
; CHECK-NEXT: [[PA1B:%.*]] = getelementptr [2 x %struct.A], ptr @a, i64 0, i64 1, i32 1, i64 [[I]]
; CHECK-NEXT: [[LENA1B:%.*]] = call i64 @strlen(ptr noundef nonnull dereferenceable(1) [[PA1B]])
-; CHECK-NEXT: [[PLEN3:%.*]] = getelementptr i64, ptr [[PLEN]], i64 3
+; CHECK-NEXT: [[PLEN3:%.*]] = getelementptr i8, ptr [[PLEN]], i64 24
; CHECK-NEXT: store i64 [[LENA1B]], ptr [[PLEN3]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/strlen-9.ll b/llvm/test/Transforms/InstCombine/strlen-9.ll
index 4e0d728fb184e5..e29e426c6a6a39 100644
--- a/llvm/test/Transforms/InstCombine/strlen-9.ll
+++ b/llvm/test/Transforms/InstCombine/strlen-9.ll
@@ -20,23 +20,23 @@ declare i64 @strlen(ptr)
define void @fold_strlen_no_nul(ptr %plen, i32 %i) {
; CHECK-LABEL: @fold_strlen_no_nul(
; CHECK-NEXT: store i64 5, ptr [[PLEN:%.*]], align 4
-; CHECK-NEXT: [[PNA5_P5:%.*]] = getelementptr i64, ptr [[PLEN]], i64 1
+; CHECK-NEXT: [[PNA5_P5:%.*]] = getelementptr i8, ptr [[PLEN]], i64 8
; CHECK-NEXT: store i64 0, ptr [[PNA5_P5]], align 4
-; CHECK-NEXT: [[PNS5_P6:%.*]] = getelementptr i64, ptr [[PLEN]], i64 2
+; CHECK-NEXT: [[PNS5_P6:%.*]] = getelementptr i8, ptr [[PLEN]], i64 16
; CHECK-NEXT: store i64 0, ptr [[PNS5_P6]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[I:%.*]] to i64
; CHECK-NEXT: [[PA5_PI:%.*]] = getelementptr [5 x i8], ptr @a5, i64 0, i64 [[TMP1]]
; CHECK-NEXT: [[NA5_PI:%.*]] = call i64 @strlen(ptr noundef nonnull dereferenceable(1) [[PA5_PI]])
-; CHECK-NEXT: [[PNA5_PI:%.*]] = getelementptr i64, ptr [[PLEN]], i64 3
+; CHECK-NEXT: [[PNA5_PI:%.*]] = getelementptr i8, ptr [[PLEN]], i64 24
; CHECK-NEXT: store i64 [[NA5_PI]], ptr [[PNA5_PI]], align 4
-; CHECK-NEXT: [[PNZ0_P0:%.*]] = getelementptr i64, ptr [[PLEN]], i64 4
+; CHECK-NEXT: [[PNZ0_P0:%.*]] = getelementptr i8, ptr [[PLEN]], i64 32
; CHECK-NEXT: store i64 0, ptr [[PNZ0_P0]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = sext i32 [[I]] to i64
; CHECK-NEXT: [[PZ0_PI:%.*]] = getelementptr [0 x i8], ptr @z0, i64 0, i64 [[TMP2]]
; CHECK-NEXT: [[NZ0_PI:%.*]] = call i64 @strlen(ptr noundef nonnull dereferenceable(1) [[PZ0_PI]])
-; CHECK-NEXT: [[PNZ0_PI:%.*]] = getelementptr i64, ptr [[PLEN]], i64 5
+; CHECK-NEXT: [[PNZ0_PI:%.*]] = getelementptr i8, ptr [[PLEN]], i64 40
; CHECK-NEXT: store i64 [[NZ0_PI]], ptr [[PNZ0_PI]], align 4
-; CHECK-NEXT: [[PNZ5_P5:%.*]] = getelementptr i64, ptr [[PLEN]], i64 6
+; CHECK-NEXT: [[PNZ5_P5:%.*]] = getelementptr i8, ptr [[PLEN]], i64 48
; CHECK-NEXT: store i64 0, ptr [[PNZ5_P5]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/strncmp-4.ll b/llvm/test/Transforms/InstCombine/strncmp-4.ll
index 0e019dad3a02ac..b22650176b0d08 100644
--- a/llvm/test/Transforms/InstCombine/strncmp-4.ll
+++ b/llvm/test/Transforms/InstCombine/strncmp-4.ll
@@ -13,23 +13,23 @@ declare i32 @strncmp(ptr, ptr, i64)
define void @fold_strncmp_Aa_b(ptr %pcmp) {
; CHECK-LABEL: @fold_strncmp_Aa_b(
; CHECK-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[PCMP1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[PCMP1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 0, ptr [[PCMP1]], align 4
-; CHECK-NEXT: [[PCMP2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[PCMP2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 0, ptr [[PCMP2]], align 4
-; CHECK-NEXT: [[PCMP3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[PCMP3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 0, ptr [[PCMP3]], align 4
-; CHECK-NEXT: [[PCMP4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[PCMP4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 0, ptr [[PCMP4]], align 4
-; CHECK-NEXT: [[PCMP5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[PCMP5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 0, ptr [[PCMP5]], align 4
-; CHECK-NEXT: [[PCMP6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; CHECK-NEXT: [[PCMP6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; CHECK-NEXT: store i32 0, ptr [[PCMP6]], align 4
-; CHECK-NEXT: [[PCMP7:%.*]] = getelementptr i32, ptr [[PCMP]], i64 7
+; CHECK-NEXT: [[PCMP7:%.*]] = getelementptr i8, ptr [[PCMP]], i64 28
; CHECK-NEXT: store i32 -1, ptr [[PCMP7]], align 4
-; CHECK-NEXT: [[PCMP8:%.*]] = getelementptr i32, ptr [[PCMP]], i64 8
+; CHECK-NEXT: [[PCMP8:%.*]] = getelementptr i8, ptr [[PCMP]], i64 32
; CHECK-NEXT: store i32 -1, ptr [[PCMP8]], align 4
-; CHECK-NEXT: [[PCMP9:%.*]] = getelementptr i32, ptr [[PCMP]], i64 9
+; CHECK-NEXT: [[PCMP9:%.*]] = getelementptr i8, ptr [[PCMP]], i64 36
; CHECK-NEXT: store i32 -1, ptr [[PCMP9]], align 4
; CHECK-NEXT: ret void
;
@@ -98,15 +98,15 @@ define void @fold_strncmp_Aa_b(ptr %pcmp) {
define void @fold_strncmp_Ab_a(ptr %pcmp) {
; CHECK-LABEL: @fold_strncmp_Ab_a(
; CHECK-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[PCMP1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[PCMP1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 0, ptr [[PCMP1]], align 4
-; CHECK-NEXT: [[PCMP2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[PCMP2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 0, ptr [[PCMP2]], align 4
-; CHECK-NEXT: [[PCMP3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[PCMP3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 0, ptr [[PCMP3]], align 4
-; CHECK-NEXT: [[PCMP4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[PCMP4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 1, ptr [[PCMP4]], align 4
-; CHECK-NEXT: [[PCMP5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[PCMP5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 1, ptr [[PCMP5]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/strncmp-5.ll b/llvm/test/Transforms/InstCombine/strncmp-5.ll
index 1065091fc4c715..4c6585ef2a983e 100644
--- a/llvm/test/Transforms/InstCombine/strncmp-5.ll
+++ b/llvm/test/Transforms/InstCombine/strncmp-5.ll
@@ -21,25 +21,25 @@ define void @fold_strncmp_a_b_n(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ne i64 [[N:%.*]], 0
; CHECK-NEXT: [[C0_1:%.*]] = sext i1 [[TMP1]] to i32
-; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[C0_1]], ptr [[S0_1]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_2:%.*]] = sext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[C0_2]], ptr [[S0_2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_3:%.*]] = sext i1 [[TMP3]] to i32
-; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 [[C0_3]], ptr [[S0_3]], align 4
-; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 0, ptr [[S0_4]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_5:%.*]] = sext i1 [[TMP4]] to i32
-; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 [[C0_5]], ptr [[S0_5]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C5_0:%.*]] = zext i1 [[TMP5]] to i32
-; CHECK-NEXT: [[S5_0:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; CHECK-NEXT: [[S5_0:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; CHECK-NEXT: store i32 [[C5_0]], ptr [[S5_0]], align 4
; CHECK-NEXT: ret void
;
@@ -117,23 +117,23 @@ define void @fold_strncmp_a_c_n(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 [[C0_0]], ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_1:%.*]] = sext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[C0_1]], ptr [[S0_1]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_2:%.*]] = sext i1 [[TMP3]] to i32
-; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[S0_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[C0_2]], ptr [[S0_2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_3:%.*]] = sext i1 [[TMP4]] to i32
-; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[S0_3:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 [[C0_3]], ptr [[S0_3]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = icmp ugt i64 [[N]], 3
; CHECK-NEXT: [[C0_4:%.*]] = sext i1 [[TMP5]] to i32
-; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[S0_4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 [[C0_4]], ptr [[S0_4]], align 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp ugt i64 [[N]], 3
; CHECK-NEXT: [[C0_5:%.*]] = sext i1 [[TMP6]] to i32
-; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[S0_5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 [[C0_5]], ptr [[S0_5]], align 4
; CHECK-NEXT: ret void
;
@@ -188,27 +188,27 @@ define void @fold_strncmp_a_d_n(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 [[C0_0]], ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C0_1:%.*]] = sext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[C0_1]], ptr [[S0_1]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ugt i64 [[N]], 3
; CHECK-NEXT: [[C1_1:%.*]] = zext i1 [[TMP3]] to i32
-; CHECK-NEXT: [[S1_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[S1_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[C1_1]], ptr [[S1_1]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp ugt i64 [[N]], 2
; CHECK-NEXT: [[C2_2:%.*]] = zext i1 [[TMP4]] to i32
-; CHECK-NEXT: [[S2_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[S2_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 [[C2_2]], ptr [[S2_2]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C4_4:%.*]] = zext i1 [[TMP5]] to i32
-; CHECK-NEXT: [[S4_4:%.*]] = getelementptr i32, ptr [[PCMP]], i64 4
+; CHECK-NEXT: [[S4_4:%.*]] = getelementptr i8, ptr [[PCMP]], i64 16
; CHECK-NEXT: store i32 [[C4_4]], ptr [[S4_4]], align 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C4_4_2:%.*]] = sext i1 [[TMP6]] to i32
-; CHECK-NEXT: [[S4_4_2:%.*]] = getelementptr i32, ptr [[PCMP]], i64 5
+; CHECK-NEXT: [[S4_4_2:%.*]] = getelementptr i8, ptr [[PCMP]], i64 20
; CHECK-NEXT: store i32 [[C4_4_2]], ptr [[S4_4_2]], align 4
-; CHECK-NEXT: [[S5_5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 6
+; CHECK-NEXT: [[S5_5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 24
; CHECK-NEXT: store i32 0, ptr [[S5_5]], align 4
-; CHECK-NEXT: [[S6_6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 7
+; CHECK-NEXT: [[S6_6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 28
; CHECK-NEXT: store i32 0, ptr [[S6_6]], align 4
; CHECK-NEXT: ret void
;
@@ -301,13 +301,13 @@ define void @fold_strncmp_d_e_n(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 0, ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ne i64 [[N:%.*]], 0
; CHECK-NEXT: [[C0_1:%.*]] = zext i1 [[TMP1]] to i32
-; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[S0_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[C0_1]], ptr [[S0_1]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[C1_0:%.*]] = sext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[S1_0:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[S1_0:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[C1_0]], ptr [[S1_0]], align 4
-; CHECK-NEXT: [[S1_1:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[S1_1:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 0, ptr [[S1_1]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/strncmp-6.ll b/llvm/test/Transforms/InstCombine/strncmp-6.ll
index e4ab9e78583f43..1fb462be2221fd 100644
--- a/llvm/test/Transforms/InstCombine/strncmp-6.ll
+++ b/llvm/test/Transforms/InstCombine/strncmp-6.ll
@@ -16,11 +16,11 @@ declare i32 @strncmp(ptr, ptr, i64)
define void @fold_strncmp_cst_cst(ptr %pcmp) {
; CHECK-LABEL: @fold_strncmp_cst_cst(
; CHECK-NEXT: store i32 -1, ptr [[PCMP:%.*]], align 4
-; CHECK-NEXT: [[SB5_A5:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[SB5_A5:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 1, ptr [[SB5_A5]], align 4
-; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 -1, ptr [[SA6_B6]], align 4
-; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 1, ptr [[SB6_A6]], align 4
; CHECK-NEXT: ret void
;
@@ -63,15 +63,15 @@ define void @fold_strncmp_cst_var(ptr %pcmp, i64 %n) {
; CHECK-NEXT: store i32 [[CA0_B0]], ptr [[PCMP:%.*]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt i64 [[N]], 6
; CHECK-NEXT: [[CB0_A0:%.*]] = zext i1 [[TMP2]] to i32
-; CHECK-NEXT: [[SB0_A0:%.*]] = getelementptr i32, ptr [[PCMP]], i64 1
+; CHECK-NEXT: [[SB0_A0:%.*]] = getelementptr i8, ptr [[PCMP]], i64 4
; CHECK-NEXT: store i32 [[CB0_A0]], ptr [[SB0_A0]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[CA6_B6:%.*]] = sext i1 [[TMP3]] to i32
-; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 2
+; CHECK-NEXT: [[SA6_B6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 8
; CHECK-NEXT: store i32 [[CA6_B6]], ptr [[SA6_B6]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp ne i64 [[N]], 0
; CHECK-NEXT: [[CB6_A6:%.*]] = zext i1 [[TMP4]] to i32
-; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i32, ptr [[PCMP]], i64 3
+; CHECK-NEXT: [[SB6_A6:%.*]] = getelementptr i8, ptr [[PCMP]], i64 12
; CHECK-NEXT: store i32 [[CB6_A6]], ptr [[SB6_A6]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/sub.ll b/llvm/test/Transforms/InstCombine/sub.ll
index 1fceca321da124..494cdf62c7975e 100644
--- a/llvm/test/Transforms/InstCombine/sub.ll
+++ b/llvm/test/Transforms/InstCombine/sub.ll
@@ -1124,7 +1124,7 @@ define i64 @test58(ptr %foo, i64 %i, i64 %j) {
define i64 @test59(ptr %foo, i64 %i) {
; CHECK-LABEL: @test59(
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds [100 x [100 x i8]], ptr [[FOO:%.*]], i64 0, i64 42, i64 [[I:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds [100 x [100 x i8]], ptr [[FOO]], i64 0, i64 42, i64 0
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[FOO]], i64 4200
; CHECK-NEXT: store ptr [[GEP1]], ptr @dummy_global1, align 8
; CHECK-NEXT: store ptr [[GEP2]], ptr @dummy_global2, align 8
; CHECK-NEXT: ret i64 [[I]]
@@ -1143,7 +1143,7 @@ define i64 @test59(ptr %foo, i64 %i) {
define i64 @test60(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test60(
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds [100 x [100 x i8]], ptr [[FOO:%.*]], i64 0, i64 [[J:%.*]], i64 [[I:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds [100 x [100 x i8]], ptr [[FOO]], i64 0, i64 42, i64 0
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[FOO]], i64 4200
; CHECK-NEXT: [[CAST1:%.*]] = ptrtoint ptr [[GEP1]] to i64
; CHECK-NEXT: [[CAST2:%.*]] = ptrtoint ptr [[GEP2]] to i64
; CHECK-NEXT: [[SUB:%.*]] = sub i64 [[CAST1]], [[CAST2]]
@@ -1162,7 +1162,7 @@ define i64 @test60(ptr %foo, i64 %i, i64 %j) {
define i64 @test61(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test61(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds [100 x [100 x i8]], ptr [[FOO:%.*]], i64 0, i64 42, i64 0
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[FOO:%.*]], i64 4200
; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds [100 x [100 x i8]], ptr [[FOO]], i64 0, i64 [[J:%.*]], i64 [[I:%.*]]
; CHECK-NEXT: [[CAST1:%.*]] = ptrtoint ptr [[GEP1]] to i64
; CHECK-NEXT: [[CAST2:%.*]] = ptrtoint ptr [[GEP2]] to i64
diff --git a/llvm/test/Transforms/InstCombine/unpack-fca.ll b/llvm/test/Transforms/InstCombine/unpack-fca.ll
index 4ac78b9e32ce9b..b0cd3823dcca99 100644
--- a/llvm/test/Transforms/InstCombine/unpack-fca.ll
+++ b/llvm/test/Transforms/InstCombine/unpack-fca.ll
@@ -24,7 +24,7 @@ define void @storeA(ptr %a.ptr) {
define void @storeB(ptr %b.ptr) {
; CHECK-LABEL: @storeB(
; CHECK-NEXT: store ptr null, ptr [[B_PTR:%.*]], align 8
-; CHECK-NEXT: [[B_PTR_REPACK1:%.*]] = getelementptr inbounds [[B:%.*]], ptr [[B_PTR]], i64 0, i32 1
+; CHECK-NEXT: [[B_PTR_REPACK1:%.*]] = getelementptr inbounds i8, ptr [[B_PTR]], i64 8
; CHECK-NEXT: store i64 42, ptr [[B_PTR_REPACK1]], align 8
; CHECK-NEXT: ret void
;
@@ -76,14 +76,14 @@ define void @storeArrayOfB(ptr %ab.ptr, [2 x %B] %ab) {
; CHECK-NEXT: [[AB_ELT:%.*]] = extractvalue [2 x %B] [[AB:%.*]], 0
; CHECK-NEXT: [[AB_ELT_ELT:%.*]] = extractvalue [[B:%.*]] [[AB_ELT]], 0
; CHECK-NEXT: store ptr [[AB_ELT_ELT]], ptr [[AB_PTR:%.*]], align 8
-; CHECK-NEXT: [[AB_PTR_REPACK3:%.*]] = getelementptr inbounds [[B]], ptr [[AB_PTR]], i64 0, i32 1
+; CHECK-NEXT: [[AB_PTR_REPACK3:%.*]] = getelementptr inbounds i8, ptr [[AB_PTR]], i64 8
; CHECK-NEXT: [[AB_ELT_ELT4:%.*]] = extractvalue [[B]] [[AB_ELT]], 1
; CHECK-NEXT: store i64 [[AB_ELT_ELT4]], ptr [[AB_PTR_REPACK3]], align 8
-; CHECK-NEXT: [[AB_PTR_REPACK1:%.*]] = getelementptr inbounds [2 x %B], ptr [[AB_PTR]], i64 0, i64 1
+; CHECK-NEXT: [[AB_PTR_REPACK1:%.*]] = getelementptr inbounds i8, ptr [[AB_PTR]], i64 16
; CHECK-NEXT: [[AB_ELT2:%.*]] = extractvalue [2 x %B] [[AB]], 1
; CHECK-NEXT: [[AB_ELT2_ELT:%.*]] = extractvalue [[B]] [[AB_ELT2]], 0
; CHECK-NEXT: store ptr [[AB_ELT2_ELT]], ptr [[AB_PTR_REPACK1]], align 8
-; CHECK-NEXT: [[AB_PTR_REPACK1_REPACK5:%.*]] = getelementptr inbounds [2 x %B], ptr [[AB_PTR]], i64 0, i64 1, i32 1
+; CHECK-NEXT: [[AB_PTR_REPACK1_REPACK5:%.*]] = getelementptr inbounds i8, ptr [[AB_PTR]], i64 24
; CHECK-NEXT: [[AB_ELT2_ELT6:%.*]] = extractvalue [[B]] [[AB_ELT2]], 1
; CHECK-NEXT: store i64 [[AB_ELT2_ELT6]], ptr [[AB_PTR_REPACK1_REPACK5]], align 8
; CHECK-NEXT: ret void
@@ -106,7 +106,7 @@ define %B @loadB(ptr %b.ptr) {
; CHECK-LABEL: @loadB(
; CHECK-NEXT: [[DOTUNPACK:%.*]] = load ptr, ptr [[B_PTR:%.*]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = insertvalue [[B:%.*]] poison, ptr [[DOTUNPACK]], 0
-; CHECK-NEXT: [[DOTELT1:%.*]] = getelementptr inbounds [[B]], ptr [[B_PTR]], i64 0, i32 1
+; CHECK-NEXT: [[DOTELT1:%.*]] = getelementptr inbounds i8, ptr [[B_PTR]], i64 8
; CHECK-NEXT: [[DOTUNPACK2:%.*]] = load i64, ptr [[DOTELT1]], align 8
; CHECK-NEXT: [[TMP2:%.*]] = insertvalue [[B]] [[TMP1]], i64 [[DOTUNPACK2]], 1
; CHECK-NEXT: ret [[B]] [[TMP2]]
@@ -162,9 +162,9 @@ define { %A } @structOfA(ptr %sa.ptr) {
define %B @structB(ptr %b.ptr) {
; CHECK-LABEL: @structB(
; CHECK-NEXT: store ptr null, ptr [[B_PTR:%.*]], align 8
-; CHECK-NEXT: [[B_PTR_REPACK1:%.*]] = getelementptr inbounds [[B:%.*]], ptr [[B_PTR]], i64 0, i32 1
+; CHECK-NEXT: [[B_PTR_REPACK1:%.*]] = getelementptr inbounds i8, ptr [[B_PTR]], i64 8
; CHECK-NEXT: store i64 42, ptr [[B_PTR_REPACK1]], align 8
-; CHECK-NEXT: ret [[B]] { ptr null, i64 42 }
+; CHECK-NEXT: ret [[B:%.*]] { ptr null, i64 42 }
;
store %B { ptr null, i64 42 }, ptr %b.ptr, align 8
%1 = load %B, ptr %b.ptr, align 8
@@ -175,14 +175,14 @@ define [2 x %B] @loadArrayOfB(ptr %ab.ptr) {
; CHECK-LABEL: @loadArrayOfB(
; CHECK-NEXT: [[DOTUNPACK_UNPACK:%.*]] = load ptr, ptr [[AB_PTR:%.*]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = insertvalue [[B:%.*]] poison, ptr [[DOTUNPACK_UNPACK]], 0
-; CHECK-NEXT: [[DOTUNPACK_ELT3:%.*]] = getelementptr inbounds [[B]], ptr [[AB_PTR]], i64 0, i32 1
+; CHECK-NEXT: [[DOTUNPACK_ELT3:%.*]] = getelementptr inbounds i8, ptr [[AB_PTR]], i64 8
; CHECK-NEXT: [[DOTUNPACK_UNPACK4:%.*]] = load i64, ptr [[DOTUNPACK_ELT3]], align 8
; CHECK-NEXT: [[DOTUNPACK5:%.*]] = insertvalue [[B]] [[TMP1]], i64 [[DOTUNPACK_UNPACK4]], 1
; CHECK-NEXT: [[TMP2:%.*]] = insertvalue [2 x %B] poison, [[B]] [[DOTUNPACK5]], 0
-; CHECK-NEXT: [[DOTELT1:%.*]] = getelementptr inbounds [2 x %B], ptr [[AB_PTR]], i64 0, i64 1
+; CHECK-NEXT: [[DOTELT1:%.*]] = getelementptr inbounds i8, ptr [[AB_PTR]], i64 16
; CHECK-NEXT: [[DOTUNPACK2_UNPACK:%.*]] = load ptr, ptr [[DOTELT1]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = insertvalue [[B]] poison, ptr [[DOTUNPACK2_UNPACK]], 0
-; CHECK-NEXT: [[DOTUNPACK2_ELT6:%.*]] = getelementptr inbounds [2 x %B], ptr [[AB_PTR]], i64 0, i64 1, i32 1
+; CHECK-NEXT: [[DOTUNPACK2_ELT6:%.*]] = getelementptr inbounds i8, ptr [[AB_PTR]], i64 24
; CHECK-NEXT: [[DOTUNPACK2_UNPACK7:%.*]] = load i64, ptr [[DOTUNPACK2_ELT6]], align 8
; CHECK-NEXT: [[DOTUNPACK28:%.*]] = insertvalue [[B]] [[TMP3]], i64 [[DOTUNPACK2_UNPACK7]], 1
; CHECK-NEXT: [[TMP4:%.*]] = insertvalue [2 x %B] [[TMP2]], [[B]] [[DOTUNPACK28]], 1
@@ -207,7 +207,7 @@ define [2000 x %B] @loadLargeArrayOfB(ptr %ab.ptr) {
; Make sure that we do not increase alignment of packed struct element
define i32 @packed_alignment(ptr dereferenceable(9) %s) {
; CHECK-LABEL: @packed_alignment(
-; CHECK-NEXT: [[TV_ELT1:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], ptr [[S:%.*]], i64 0, i32 1, i32 1
+; CHECK-NEXT: [[TV_ELT1:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 5
; CHECK-NEXT: [[TV_UNPACK2:%.*]] = load i32, ptr [[TV_ELT1]], align 1
; CHECK-NEXT: ret i32 [[TV_UNPACK2]]
;
@@ -222,38 +222,38 @@ define i32 @packed_alignment(ptr dereferenceable(9) %s) {
define void @check_alignment(ptr %u, ptr %v) {
; CHECK-LABEL: @check_alignment(
; CHECK-NEXT: [[DOTUNPACK:%.*]] = load i8, ptr [[U:%.*]], align 8
-; CHECK-NEXT: [[DOTELT1:%.*]] = getelementptr inbounds [[STRUCT_U:%.*]], ptr [[U]], i64 0, i32 1
+; CHECK-NEXT: [[DOTELT1:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 1
; CHECK-NEXT: [[DOTUNPACK2:%.*]] = load i8, ptr [[DOTELT1]], align 1
-; CHECK-NEXT: [[DOTELT3:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[U]], i64 0, i32 2
+; CHECK-NEXT: [[DOTELT3:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 2
; CHECK-NEXT: [[DOTUNPACK4:%.*]] = load i8, ptr [[DOTELT3]], align 2
-; CHECK-NEXT: [[DOTELT5:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[U]], i64 0, i32 3
+; CHECK-NEXT: [[DOTELT5:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 3
; CHECK-NEXT: [[DOTUNPACK6:%.*]] = load i8, ptr [[DOTELT5]], align 1
-; CHECK-NEXT: [[DOTELT7:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[U]], i64 0, i32 4
+; CHECK-NEXT: [[DOTELT7:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 4
; CHECK-NEXT: [[DOTUNPACK8:%.*]] = load i8, ptr [[DOTELT7]], align 4
-; CHECK-NEXT: [[DOTELT9:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[U]], i64 0, i32 5
+; CHECK-NEXT: [[DOTELT9:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 5
; CHECK-NEXT: [[DOTUNPACK10:%.*]] = load i8, ptr [[DOTELT9]], align 1
-; CHECK-NEXT: [[DOTELT11:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[U]], i64 0, i32 6
+; CHECK-NEXT: [[DOTELT11:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 6
; CHECK-NEXT: [[DOTUNPACK12:%.*]] = load i8, ptr [[DOTELT11]], align 2
-; CHECK-NEXT: [[DOTELT13:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[U]], i64 0, i32 7
+; CHECK-NEXT: [[DOTELT13:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 7
; CHECK-NEXT: [[DOTUNPACK14:%.*]] = load i8, ptr [[DOTELT13]], align 1
-; CHECK-NEXT: [[DOTELT15:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[U]], i64 0, i32 8
+; CHECK-NEXT: [[DOTELT15:%.*]] = getelementptr inbounds i8, ptr [[U]], i64 8
; CHECK-NEXT: [[DOTUNPACK16:%.*]] = load i64, ptr [[DOTELT15]], align 8
; CHECK-NEXT: store i8 [[DOTUNPACK]], ptr [[V:%.*]], align 8
-; CHECK-NEXT: [[V_REPACK17:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 1
+; CHECK-NEXT: [[V_REPACK17:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 1
; CHECK-NEXT: store i8 [[DOTUNPACK2]], ptr [[V_REPACK17]], align 1
-; CHECK-NEXT: [[V_REPACK19:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 2
+; CHECK-NEXT: [[V_REPACK19:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 2
; CHECK-NEXT: store i8 [[DOTUNPACK4]], ptr [[V_REPACK19]], align 2
-; CHECK-NEXT: [[V_REPACK21:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 3
+; CHECK-NEXT: [[V_REPACK21:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 3
; CHECK-NEXT: store i8 [[DOTUNPACK6]], ptr [[V_REPACK21]], align 1
-; CHECK-NEXT: [[V_REPACK23:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 4
+; CHECK-NEXT: [[V_REPACK23:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 4
; CHECK-NEXT: store i8 [[DOTUNPACK8]], ptr [[V_REPACK23]], align 4
-; CHECK-NEXT: [[V_REPACK25:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 5
+; CHECK-NEXT: [[V_REPACK25:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 5
; CHECK-NEXT: store i8 [[DOTUNPACK10]], ptr [[V_REPACK25]], align 1
-; CHECK-NEXT: [[V_REPACK27:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 6
+; CHECK-NEXT: [[V_REPACK27:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 6
; CHECK-NEXT: store i8 [[DOTUNPACK12]], ptr [[V_REPACK27]], align 2
-; CHECK-NEXT: [[V_REPACK29:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 7
+; CHECK-NEXT: [[V_REPACK29:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 7
; CHECK-NEXT: store i8 [[DOTUNPACK14]], ptr [[V_REPACK29]], align 1
-; CHECK-NEXT: [[V_REPACK31:%.*]] = getelementptr inbounds [[STRUCT_U]], ptr [[V]], i64 0, i32 8
+; CHECK-NEXT: [[V_REPACK31:%.*]] = getelementptr inbounds i8, ptr [[V]], i64 8
; CHECK-NEXT: store i64 [[DOTUNPACK16]], ptr [[V_REPACK31]], align 8
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll b/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll
index 25f5c39c2c9484..738ef1bc1ad2f6 100644
--- a/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll
@@ -512,7 +512,7 @@ define ptr @gep_vbase_w_s_idx(<2 x ptr> %base, i64 %index) {
define ptr @gep_splat_base_w_s_idx(ptr %base) {
; CHECK-LABEL: @gep_splat_base_w_s_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 1
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 4
; CHECK-NEXT: ret ptr [[EE]]
;
%basevec1 = insertelement <2 x ptr> poison, ptr %base, i32 0
@@ -576,7 +576,7 @@ define ptr @gep_cvbase_w_cv_idx(<2 x ptr> %base, i64 %raw_addr) {
define ptr @gep_sbase_w_cv_idx(ptr %base) {
; CHECK-LABEL: @gep_sbase_w_cv_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 1
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 4
; CHECK-NEXT: ret ptr [[EE]]
;
%gep = getelementptr i32, ptr %base, <2 x i64> <i64 0, i64 1>
diff --git a/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll b/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
index c84d39c5aa9dd2..4e5fdbb1eab991 100644
--- a/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
+++ b/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
@@ -515,7 +515,7 @@ define ptr @gep_vbase_w_s_idx(<2 x ptr> %base, i64 %index) {
define ptr @gep_splat_base_w_s_idx(ptr %base) {
; CHECK-LABEL: @gep_splat_base_w_s_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 1
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 4
; CHECK-NEXT: ret ptr [[EE]]
;
%basevec1 = insertelement <2 x ptr> undef, ptr %base, i32 0
@@ -579,7 +579,7 @@ define ptr @gep_cvbase_w_cv_idx(<2 x ptr> %base, i64 %raw_addr) {
define ptr @gep_sbase_w_cv_idx(ptr %base) {
; CHECK-LABEL: @gep_sbase_w_cv_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 1
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 4
; CHECK-NEXT: ret ptr [[EE]]
;
%gep = getelementptr i32, ptr %base, <2 x i64> <i64 0, i64 1>
diff --git a/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg-inseltpoison.ll b/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg-inseltpoison.ll
index 40b76d140b5a19..4ac1d084cda9a5 100644
--- a/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg-inseltpoison.ll
@@ -3,7 +3,7 @@
define <4 x ptr> @PR41270(ptr %x) {
; CHECK-LABEL: @PR41270(
-; CHECK-NEXT: [[T3:%.*]] = getelementptr inbounds [4 x i16], ptr [[X:%.*]], i64 0, i64 3
+; CHECK-NEXT: [[T3:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 6
; CHECK-NEXT: [[INS2:%.*]] = insertelement <4 x ptr> poison, ptr [[T3]], i64 0
; CHECK-NEXT: ret <4 x ptr> [[INS2]]
;
diff --git a/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg.ll b/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg.ll
index 4e4fa7defa8cfc..8b442133335423 100644
--- a/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg.ll
+++ b/llvm/test/Transforms/InstCombine/vec_gep_scalar_arg.ll
@@ -3,7 +3,7 @@
define <4 x ptr> @PR41270(ptr %x) {
; CHECK-LABEL: @PR41270(
-; CHECK-NEXT: [[T3:%.*]] = getelementptr inbounds [4 x i16], ptr [[X:%.*]], i64 0, i64 3
+; CHECK-NEXT: [[T3:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 6
; CHECK-NEXT: [[INS2:%.*]] = insertelement <4 x ptr> <ptr poison, ptr undef, ptr undef, ptr undef>, ptr [[T3]], i64 0
; CHECK-NEXT: ret <4 x ptr> [[INS2]]
;
diff --git a/llvm/test/Transforms/InstCombine/vscale_gep.ll b/llvm/test/Transforms/InstCombine/vscale_gep.ll
index 2a1865f69fe30d..534888dc1a1c49 100644
--- a/llvm/test/Transforms/InstCombine/vscale_gep.ll
+++ b/llvm/test/Transforms/InstCombine/vscale_gep.ll
@@ -40,7 +40,7 @@ define void @gep_bitcast(ptr %p) {
define i32 @gep_alloca_inbounds_vscale_zero() {
; CHECK-LABEL: @gep_alloca_inbounds_vscale_zero(
; CHECK-NEXT: [[A:%.*]] = alloca <vscale x 4 x i32>, align 16
-; CHECK-NEXT: [[TMP:%.*]] = getelementptr inbounds <vscale x 4 x i32>, ptr [[A]], i64 0, i64 2
+; CHECK-NEXT: [[TMP:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 8
; CHECK-NEXT: [[LOAD:%.*]] = load i32, ptr [[TMP]], align 4
; CHECK-NEXT: ret i32 [[LOAD]]
;
diff --git a/llvm/test/Transforms/InstCombine/wcslen-5.ll b/llvm/test/Transforms/InstCombine/wcslen-5.ll
index 29874e7de5928b..3e0cdf34a6422a 100644
--- a/llvm/test/Transforms/InstCombine/wcslen-5.ll
+++ b/llvm/test/Transforms/InstCombine/wcslen-5.ll
@@ -42,7 +42,7 @@ define dso_local i64 @fold_wcslen_s3_pi_p1_s5(i1 zeroext %0, i64 %1) {
; XFAIL-CHECK-NEXT: ret i64 [[SEL]]
; CHECK-LABEL: @fold_wcslen_s3_pi_p1_s5(
; CHECK-NEXT: [[PS3_PI:%.*]] = getelementptr inbounds [4 x i32], ptr @ws3, i64 0, i64 [[TMP1:%.*]]
-; CHECK-NEXT: [[PS3_PI_P1:%.*]] = getelementptr inbounds i32, ptr [[PS3_PI]], i64 1
+; CHECK-NEXT: [[PS3_PI_P1:%.*]] = getelementptr inbounds i8, ptr [[PS3_PI]], i64 4
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[TMP0:%.*]], ptr [[PS3_PI_P1]], ptr @ws5
; CHECK-NEXT: [[LEN:%.*]] = tail call i64 @wcslen(ptr nonnull [[SEL]])
; CHECK-NEXT: ret i64 [[LEN]]
diff --git a/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll b/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll
index b364116e78e4f9..141af3e66949a3 100644
--- a/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll
+++ b/llvm/test/Transforms/LoopUnroll/ARM/upperbound.ll
@@ -17,7 +17,7 @@ define void @test(ptr %x, i32 %n) {
; CHECK-NEXT: store i32 0, ptr [[X]], align 4
; CHECK-NEXT: br label [[IF_END]]
; CHECK: if.end:
-; CHECK-NEXT: [[INCDEC_PTR:%.*]] = getelementptr inbounds i32, ptr [[X]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 4
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[REM]], 1
; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY_1:%.*]]
; CHECK: while.body.1:
@@ -28,7 +28,7 @@ define void @test(ptr %x, i32 %n) {
; CHECK-NEXT: store i32 0, ptr [[INCDEC_PTR]], align 4
; CHECK-NEXT: br label [[IF_END_1]]
; CHECK: if.end.1:
-; CHECK-NEXT: [[INCDEC_PTR_1:%.*]] = getelementptr inbounds i32, ptr [[X]], i32 2
+; CHECK-NEXT: [[INCDEC_PTR_1:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 8
; CHECK-NEXT: [[CMP_1:%.*]] = icmp sgt i32 [[REM]], 2
; CHECK-NEXT: br i1 [[CMP_1]], label [[WHILE_BODY_2:%.*]], label [[WHILE_END]]
; CHECK: while.body.2:
diff --git a/llvm/test/Transforms/LoopUnroll/peel-loop.ll b/llvm/test/Transforms/LoopUnroll/peel-loop.ll
index 88ffe60075f9b7..a76facae3cb9d9 100644
--- a/llvm/test/Transforms/LoopUnroll/peel-loop.ll
+++ b/llvm/test/Transforms/LoopUnroll/peel-loop.ll
@@ -16,20 +16,20 @@ define void @basic(ptr %p, i32 %k) #0 {
; CHECK-NEXT: [[CMP_PEEL_NOT:%.*]] = icmp eq i32 [[K]], 1
; CHECK-NEXT: br i1 [[CMP_PEEL_NOT]], label [[FOR_END]], label [[FOR_BODY_PEEL2:%.*]]
; CHECK: for.body.peel2:
-; CHECK-NEXT: [[INCDEC_PTR_PEEL:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR_PEEL:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 4
; CHECK-NEXT: store i32 1, ptr [[INCDEC_PTR_PEEL]], align 4
; CHECK-NEXT: [[CMP_PEEL5:%.*]] = icmp sgt i32 [[K]], 2
; CHECK-NEXT: br i1 [[CMP_PEEL5]], label [[FOR_BODY_PEEL7:%.*]], label [[FOR_END]]
; CHECK: for.body.peel7:
-; CHECK-NEXT: [[INCDEC_PTR_PEEL3:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 2
-; CHECK-NEXT: [[INCDEC_PTR_PEEL8:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 3
+; CHECK-NEXT: [[INCDEC_PTR_PEEL3:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 8
+; CHECK-NEXT: [[INCDEC_PTR_PEEL8:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 12
; CHECK-NEXT: store i32 2, ptr [[INCDEC_PTR_PEEL3]], align 4
; CHECK-NEXT: [[CMP_PEEL10_NOT:%.*]] = icmp eq i32 [[K]], 3
; CHECK-NEXT: br i1 [[CMP_PEEL10_NOT]], label [[FOR_END]], label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I_05:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 3, [[FOR_BODY_PEEL7]] ]
; CHECK-NEXT: [[P_ADDR_04:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[INCDEC_PTR_PEEL8]], [[FOR_BODY_PEEL7]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[P_ADDR_04]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[P_ADDR_04]], i64 4
; CHECK-NEXT: store i32 [[I_05]], ptr [[P_ADDR_04]], align 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_05]], 1
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[INC]], [[K]]
@@ -76,20 +76,20 @@ define i32 @output(ptr %p, i32 %k) #0 {
; CHECK-NEXT: [[CMP_PEEL_NOT:%.*]] = icmp eq i32 [[K]], 1
; CHECK-NEXT: br i1 [[CMP_PEEL_NOT]], label [[FOR_END]], label [[FOR_BODY_PEEL2:%.*]]
; CHECK: for.body.peel2:
-; CHECK-NEXT: [[INCDEC_PTR_PEEL:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR_PEEL:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 4
; CHECK-NEXT: store i32 1, ptr [[INCDEC_PTR_PEEL]], align 4
; CHECK-NEXT: [[CMP_PEEL5:%.*]] = icmp sgt i32 [[K]], 2
; CHECK-NEXT: br i1 [[CMP_PEEL5]], label [[FOR_BODY_PEEL7:%.*]], label [[FOR_END]]
; CHECK: for.body.peel7:
-; CHECK-NEXT: [[INCDEC_PTR_PEEL3:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 2
-; CHECK-NEXT: [[INCDEC_PTR_PEEL8:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 3
+; CHECK-NEXT: [[INCDEC_PTR_PEEL3:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 8
+; CHECK-NEXT: [[INCDEC_PTR_PEEL8:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 12
; CHECK-NEXT: store i32 2, ptr [[INCDEC_PTR_PEEL3]], align 4
; CHECK-NEXT: [[CMP_PEEL10_NOT:%.*]] = icmp eq i32 [[K]], 3
; CHECK-NEXT: br i1 [[CMP_PEEL10_NOT]], label [[FOR_END]], label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I_05:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 3, [[FOR_BODY_PEEL7]] ]
; CHECK-NEXT: [[P_ADDR_04:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[INCDEC_PTR_PEEL8]], [[FOR_BODY_PEEL7]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[P_ADDR_04]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[P_ADDR_04]], i64 4
; CHECK-NEXT: store i32 [[I_05]], ptr [[P_ADDR_04]], align 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_05]], 1
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[INC]], [[K]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/deterministic-type-shrinkage.ll b/llvm/test/Transforms/LoopVectorize/AArch64/deterministic-type-shrinkage.ll
index bcf5c1888786a6..03f055413583f3 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/deterministic-type-shrinkage.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/deterministic-type-shrinkage.ll
@@ -459,7 +459,7 @@ define void @old_and_new_size_equalko(ptr noalias %src, ptr noalias %dst) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; CHECK-NEXT: store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP1]], align 4
; CHECK-NEXT: store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/intrinsiccost.ll b/llvm/test/Transforms/LoopVectorize/AArch64/intrinsiccost.ll
index 95dc010811e6cc..b772a3814a64ac 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/intrinsiccost.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/intrinsiccost.ll
@@ -38,12 +38,12 @@ define void @saddsat(ptr nocapture readonly %pSrc, i16 signext %offset, ptr noca
; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[PSRC]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[INDEX]], 1
; CHECK-NEXT: [[NEXT_GEP6:%.*]] = getelementptr i8, ptr [[PDST]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i16, ptr [[NEXT_GEP]], i64 8
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[NEXT_GEP]], align 2
; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <8 x i16>, ptr [[TMP5]], align 2
; CHECK-NEXT: [[TMP6:%.*]] = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> [[WIDE_LOAD]], <8 x i16> [[BROADCAST_SPLAT]])
; CHECK-NEXT: [[TMP7:%.*]] = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> [[WIDE_LOAD8]], <8 x i16> [[BROADCAST_SPLAT]])
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i16, ptr [[NEXT_GEP6]], i64 8
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[NEXT_GEP6]], i64 16
; CHECK-NEXT: store <8 x i16> [[TMP6]], ptr [[NEXT_GEP6]], align 2
; CHECK-NEXT: store <8 x i16> [[TMP7]], ptr [[TMP8]], align 2
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
@@ -61,10 +61,10 @@ define void @saddsat(ptr nocapture readonly %pSrc, i16 signext %offset, ptr noca
; CHECK-NEXT: [[BLKCNT_09:%.*]] = phi i32 [ [[DEC:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[PSRC_ADDR_08:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[PDST_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR3:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i16, ptr [[PSRC_ADDR_08]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRC_ADDR_08]], i64 2
; CHECK-NEXT: [[TMP10:%.*]] = load i16, ptr [[PSRC_ADDR_08]], align 2
; CHECK-NEXT: [[TMP11:%.*]] = tail call i16 @llvm.sadd.sat.i16(i16 [[TMP10]], i16 [[OFFSET]])
-; CHECK-NEXT: [[INCDEC_PTR3]] = getelementptr inbounds i16, ptr [[PDST_ADDR_07]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR3]] = getelementptr inbounds i8, ptr [[PDST_ADDR_07]], i64 2
; CHECK-NEXT: store i16 [[TMP11]], ptr [[PDST_ADDR_07]], align 2
; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_09]], -1
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-cond-inv-loads.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-cond-inv-loads.ll
index 7848d0836d546f..6f62f2f2096f17 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-cond-inv-loads.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-cond-inv-loads.ll
@@ -173,7 +173,7 @@ define void @invariant_load_cond(ptr noalias nocapture %a, ptr nocapture readonl
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 42
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 168
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x ptr> poison, ptr [[TMP5]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x ptr> [[BROADCAST_SPLATINSERT]], <vscale x 4 x ptr> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[COND:%.*]], i64 [[INDEX]]
@@ -201,7 +201,7 @@ define void @invariant_load_cond(ptr noalias nocapture %a, ptr nocapture readonl
; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP12]], 0
; CHECK-NEXT: br i1 [[TOBOOL_NOT]], label [[FOR_INC]], label [[IF_THEN:%.*]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 42
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 168
; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX3]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll
index f600a0d5877d33..c6aec2264e3792 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll
@@ -26,8 +26,8 @@ define void @test_array_load2_store2(i32 %C, i32 %D) #1 {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[TMP8]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[C:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[D:%.*]], i64 0
@@ -36,19 +36,19 @@ define void @test_array_load2_store2(i32 %C, i32 %D) #1 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1024 x i32], ptr @AB, i64 0, i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP0]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1024 x i32], ptr @AB, i64 0, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
-; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 1
-; CHECK-NEXT: [[TMP4:%.*]] = add nsw <vscale x 4 x i32> [[TMP1]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: [[TMP5:%.*]] = mul nsw <vscale x 4 x i32> [[TMP2]], [[BROADCAST_SPLAT2]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1024 x i32], ptr @CD, i64 0, i64 [[TMP3]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i64 -1
-; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]])
-; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP7]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
+; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
+; CHECK-NEXT: [[TMP5:%.*]] = or disjoint i64 [[OFFSET_IDX]], 1
+; CHECK-NEXT: [[TMP6:%.*]] = add nsw <vscale x 4 x i32> [[TMP3]], [[BROADCAST_SPLAT]]
+; CHECK-NEXT: [[TMP7:%.*]] = mul nsw <vscale x 4 x i32> [[TMP4]], [[BROADCAST_SPLAT2]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1024 x i32], ptr @CD, i64 0, i64 [[TMP5]]
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP8]], i64 -4
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]])
+; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP9]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 512
; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
@@ -106,13 +106,13 @@ define void @test_array_load2_i16_store2(i32 %C, i32 %D) #1 {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[TMP14]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP1:%.*]] = shl <vscale x 4 x i64> [[TMP0]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[TMP2]], 3
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP3]], i64 0
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP3:%.*]] = shl <vscale x 4 x i64> [[TMP2]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i64 [[TMP4]], 3
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP5]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[C:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
@@ -121,22 +121,22 @@ define void @test_array_load2_i16_store2(i32 %C, i32 %D) #1 {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP1]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1024 x i16], ptr @AB_i16, i64 0, <vscale x 4 x i64> [[VEC_IND]]
-; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP4]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison)
-; CHECK-NEXT: [[TMP5:%.*]] = or disjoint <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1024 x i16], ptr @AB_i16, i64 0, <vscale x 4 x i64> [[TMP5]]
-; CHECK-NEXT: [[WIDE_MASKED_GATHER1:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP6]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison)
-; CHECK-NEXT: [[TMP7:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER]] to <vscale x 4 x i32>
-; CHECK-NEXT: [[TMP8:%.*]] = add nsw <vscale x 4 x i32> [[BROADCAST_SPLAT]], [[TMP7]]
-; CHECK-NEXT: [[TMP9:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER1]] to <vscale x 4 x i32>
-; CHECK-NEXT: [[TMP10:%.*]] = mul nsw <vscale x 4 x i32> [[BROADCAST_SPLAT3]], [[TMP9]]
-; CHECK-NEXT: [[TMP11:%.*]] = extractelement <vscale x 4 x i64> [[TMP5]], i64 0
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1024 x i32], ptr @CD, i64 0, i64 [[TMP11]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i64 -1
-; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[TMP8]], <vscale x 4 x i32> [[TMP10]])
-; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP13]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP15]]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP3]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1024 x i16], ptr @AB_i16, i64 0, <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP6]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison)
+; CHECK-NEXT: [[TMP7:%.*]] = or disjoint <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1024 x i16], ptr @AB_i16, i64 0, <vscale x 4 x i64> [[TMP7]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER1:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP8]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison)
+; CHECK-NEXT: [[TMP9:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER]] to <vscale x 4 x i32>
+; CHECK-NEXT: [[TMP10:%.*]] = add nsw <vscale x 4 x i32> [[BROADCAST_SPLAT]], [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER1]] to <vscale x 4 x i32>
+; CHECK-NEXT: [[TMP12:%.*]] = mul nsw <vscale x 4 x i32> [[BROADCAST_SPLAT3]], [[TMP11]]
+; CHECK-NEXT: [[TMP13:%.*]] = extractelement <vscale x 4 x i64> [[TMP7]], i64 0
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1024 x i32], ptr @CD, i64 0, i64 [[TMP13]]
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP14]], i64 -4
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[TMP10]], <vscale x 4 x i32> [[TMP12]])
+; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP15]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 512
; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
@@ -197,13 +197,13 @@ define void @test_array_load2_store2_i16(i32 noundef %C, i32 noundef %D) #1 {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[TMP14]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP1:%.*]] = shl <vscale x 4 x i64> [[TMP0]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[TMP2]], 3
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP3]], i64 0
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP3:%.*]] = shl <vscale x 4 x i64> [[TMP2]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i64 [[TMP4]], 3
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP5]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[C:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
@@ -212,23 +212,23 @@ define void @test_array_load2_store2_i16(i32 noundef %C, i32 noundef %D) #1 {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP1]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP3]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1024 x i32], ptr @AB, i64 0, i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP4]], align 4
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1024 x i32], ptr @AB, i64 0, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
-; CHECK-NEXT: [[TMP7:%.*]] = or disjoint <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP8:%.*]] = add nsw <vscale x 4 x i32> [[TMP5]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: [[TMP9:%.*]] = trunc <vscale x 4 x i32> [[TMP8]] to <vscale x 4 x i16>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds [1024 x i16], ptr @CD_i16, i64 0, <vscale x 4 x i64> [[VEC_IND]]
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i16.nxv4p0(<vscale x 4 x i16> [[TMP9]], <vscale x 4 x ptr> [[TMP10]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[TMP11:%.*]] = mul nsw <vscale x 4 x i32> [[TMP6]], [[BROADCAST_SPLAT2]]
-; CHECK-NEXT: [[TMP12:%.*]] = trunc <vscale x 4 x i32> [[TMP11]] to <vscale x 4 x i16>
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1024 x i16], ptr @CD_i16, i64 0, <vscale x 4 x i64> [[TMP7]]
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i16.nxv4p0(<vscale x 4 x i16> [[TMP12]], <vscale x 4 x ptr> [[TMP13]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP15]]
+; CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP8:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
+; CHECK-NEXT: [[TMP9:%.*]] = or disjoint <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP10:%.*]] = add nsw <vscale x 4 x i32> [[TMP7]], [[BROADCAST_SPLAT]]
+; CHECK-NEXT: [[TMP11:%.*]] = trunc <vscale x 4 x i32> [[TMP10]] to <vscale x 4 x i16>
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1024 x i16], ptr @CD_i16, i64 0, <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i16.nxv4p0(<vscale x 4 x i16> [[TMP11]], <vscale x 4 x ptr> [[TMP12]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[TMP13:%.*]] = mul nsw <vscale x 4 x i32> [[TMP8]], [[BROADCAST_SPLAT2]]
+; CHECK-NEXT: [[TMP14:%.*]] = trunc <vscale x 4 x i32> [[TMP13]] to <vscale x 4 x i16>
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1024 x i16], ptr @CD_i16, i64 0, <vscale x 4 x i64> [[TMP9]]
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i16.nxv4p0(<vscale x 4 x i16> [[TMP14]], <vscale x 4 x ptr> [[TMP15]], i32 2, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 512
; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
@@ -277,42 +277,42 @@ define i32 @test_struct_load6(%struct.ST6* %S) #1 {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP15:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP16:%.*]] = shl nuw nsw i64 [[TMP15]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 2
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP2]], i64 0
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[TMP3]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP4]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP0]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ST6:%.*]], ptr [[S:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
-; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP3]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 1
-; CHECK-NEXT: [[WIDE_MASKED_GATHER1:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP4]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 2
-; CHECK-NEXT: [[WIDE_MASKED_GATHER2:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP5]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 3
-; CHECK-NEXT: [[WIDE_MASKED_GATHER3:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP6]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 4
-; CHECK-NEXT: [[WIDE_MASKED_GATHER4:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP7]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 5
-; CHECK-NEXT: [[WIDE_MASKED_GATHER5:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP8]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP9:%.*]] = add <vscale x 4 x i32> [[WIDE_MASKED_GATHER]], [[VEC_PHI]]
-; CHECK-NEXT: [[TMP10:%.*]] = add <vscale x 4 x i32> [[TMP9]], [[WIDE_MASKED_GATHER2]]
-; CHECK-NEXT: [[TMP11:%.*]] = add <vscale x 4 x i32> [[WIDE_MASKED_GATHER1]], [[WIDE_MASKED_GATHER3]]
-; CHECK-NEXT: [[TMP12:%.*]] = add <vscale x 4 x i32> [[TMP11]], [[WIDE_MASKED_GATHER4]]
-; CHECK-NEXT: [[TMP13:%.*]] = add <vscale x 4 x i32> [[TMP12]], [[WIDE_MASKED_GATHER5]]
-; CHECK-NEXT: [[TMP14]] = sub <vscale x 4 x i32> [[TMP10]], [[TMP13]]
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP16]]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP2]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP16:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_ST6:%.*]], ptr [[S:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP5]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 1
+; CHECK-NEXT: [[WIDE_MASKED_GATHER1:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP6]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 2
+; CHECK-NEXT: [[WIDE_MASKED_GATHER2:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP7]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 3
+; CHECK-NEXT: [[WIDE_MASKED_GATHER3:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP8]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 4
+; CHECK-NEXT: [[WIDE_MASKED_GATHER4:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP9]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]], i32 5
+; CHECK-NEXT: [[WIDE_MASKED_GATHER5:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> [[TMP10]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i32> poison)
+; CHECK-NEXT: [[TMP11:%.*]] = add <vscale x 4 x i32> [[WIDE_MASKED_GATHER]], [[VEC_PHI]]
+; CHECK-NEXT: [[TMP12:%.*]] = add <vscale x 4 x i32> [[TMP11]], [[WIDE_MASKED_GATHER2]]
+; CHECK-NEXT: [[TMP13:%.*]] = add <vscale x 4 x i32> [[WIDE_MASKED_GATHER1]], [[WIDE_MASKED_GATHER3]]
+; CHECK-NEXT: [[TMP14:%.*]] = add <vscale x 4 x i32> [[TMP13]], [[WIDE_MASKED_GATHER4]]
+; CHECK-NEXT: [[TMP15:%.*]] = add <vscale x 4 x i32> [[TMP14]], [[WIDE_MASKED_GATHER5]]
+; CHECK-NEXT: [[TMP16]] = sub <vscale x 4 x i32> [[TMP12]], [[TMP15]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; CHECK-NEXT: br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: middle.block:
-; CHECK-NEXT: [[TMP18:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP14]])
+; CHECK-NEXT: [[TMP18:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP16]])
; CHECK-NEXT: br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
@@ -381,12 +381,12 @@ define void @test_reversed_load2_store2(%struct.ST2* noalias nocapture readonly
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP18:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP19:%.*]] = shl nuw nsw i64 [[TMP18]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.experimental.stepvector.nxv4i32()
-; CHECK-NEXT: [[INDUCTION:%.*]] = sub <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1023, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), [[TMP0]]
-; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.vscale.i32()
-; CHECK-NEXT: [[DOTNEG:%.*]] = mul nsw i32 [[TMP1]], -4
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.experimental.stepvector.nxv4i32()
+; CHECK-NEXT: [[INDUCTION:%.*]] = sub <vscale x 4 x i32> shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1023, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer), [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = call i32 @llvm.vscale.i32()
+; CHECK-NEXT: [[DOTNEG:%.*]] = mul nsw i32 [[TMP3]], -4
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[DOTNEG]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
@@ -394,31 +394,31 @@ define void @test_reversed_load2_store2(%struct.ST2* noalias nocapture readonly
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_ST2:%.*]], ptr [[A:%.*]], i64 [[OFFSET_IDX]], i32 0
-; CHECK-NEXT: [[TMP3:%.*]] = call i32 @llvm.vscale.i32()
-; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i32 [[TMP3]], 3
-; CHECK-NEXT: [[TMP5:%.*]] = sub nsw i32 2, [[TMP4]]
-; CHECK-NEXT: [[TMP6:%.*]] = sext i32 [[TMP5]] to i64
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP6]]
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP7]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_ST2:%.*]], ptr [[A:%.*]], i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT: [[TMP5:%.*]] = call i32 @llvm.vscale.i32()
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i32 [[TMP5]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = sub nsw i32 2, [[TMP6]]
+; CHECK-NEXT: [[TMP8:%.*]] = sext i32 [[TMP7]] to i64
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i64 [[TMP8]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP9]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP8:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[REVERSE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP8]])
-; CHECK-NEXT: [[TMP9:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
-; CHECK-NEXT: [[REVERSE1:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP9]])
-; CHECK-NEXT: [[TMP10:%.*]] = add nsw <vscale x 4 x i32> [[REVERSE]], [[VEC_IND]]
-; CHECK-NEXT: [[TMP11:%.*]] = sub nsw <vscale x 4 x i32> [[REVERSE1]], [[VEC_IND]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_ST2]], ptr [[B:%.*]], i64 [[OFFSET_IDX]], i32 1
-; CHECK-NEXT: [[TMP13:%.*]] = call i32 @llvm.vscale.i32()
-; CHECK-NEXT: [[TMP14:%.*]] = shl nuw nsw i32 [[TMP13]], 3
-; CHECK-NEXT: [[TMP15:%.*]] = sub nsw i32 1, [[TMP14]]
-; CHECK-NEXT: [[TMP16:%.*]] = sext i32 [[TMP15]] to i64
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i64 [[TMP16]]
-; CHECK-NEXT: [[REVERSE2:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP10]])
-; CHECK-NEXT: [[REVERSE3:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP11]])
+; CHECK-NEXT: [[TMP10:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[REVERSE:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP10]])
+; CHECK-NEXT: [[TMP11:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
+; CHECK-NEXT: [[REVERSE1:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP11]])
+; CHECK-NEXT: [[TMP12:%.*]] = add nsw <vscale x 4 x i32> [[REVERSE]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP13:%.*]] = sub nsw <vscale x 4 x i32> [[REVERSE1]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT_ST2]], ptr [[B:%.*]], i64 [[OFFSET_IDX]], i32 1
+; CHECK-NEXT: [[TMP15:%.*]] = call i32 @llvm.vscale.i32()
+; CHECK-NEXT: [[TMP16:%.*]] = shl nuw nsw i32 [[TMP15]], 3
+; CHECK-NEXT: [[TMP17:%.*]] = sub nsw i32 1, [[TMP16]]
+; CHECK-NEXT: [[TMP18:%.*]] = sext i32 [[TMP17]] to i64
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i64 [[TMP18]]
+; CHECK-NEXT: [[REVERSE2:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP12]])
+; CHECK-NEXT: [[REVERSE3:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP13]])
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[REVERSE2]], <vscale x 4 x i32> [[REVERSE3]])
-; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP17]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP19]]
+; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP19]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i32> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; CHECK-NEXT: br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
@@ -475,21 +475,21 @@ define void @even_load_static_tc(i32* noalias nocapture readonly %A, i32* noalia
; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
; CHECK-NEXT: [[N_VEC:%.*]] = sub nuw nsw i64 512, [[TMP1]]
; CHECK-NEXT: [[IND_END:%.*]] = shl nuw nsw i64 [[N_VEC]], 1
-; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[TMP2]], 2
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[TMP4:%.*]] = shl nsw <vscale x 4 x i32> [[TMP3]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP5:%.*]] = and i64 [[INDEX]], 9223372036854775804
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP5]]
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP4]], ptr [[TMP6]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
+; CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw <vscale x 4 x i32> [[TMP5]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP7:%.*]] = and i64 [[INDEX]], 9223372036854775804
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP7]]
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP6]], ptr [[TMP8]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: middle.block:
@@ -561,21 +561,21 @@ define void @even_load_dynamic_tc(i32* noalias nocapture readonly %A, i32* noali
; CHECK-NEXT: [[TMP9:%.*]] = select i1 [[TMP8]], i64 [[TMP6]], i64 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP4]], [[TMP9]]
; CHECK-NEXT: [[IND_END:%.*]] = shl i64 [[N_VEC]], 1
-; CHECK-NEXT: [[TMP15:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP16:%.*]] = shl nuw nsw i64 [[TMP15]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP10]], 2
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP10]], align 4
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP12]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP11:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[TMP12:%.*]] = shl nsw <vscale x 4 x i32> [[TMP11]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP13:%.*]] = and i64 [[INDEX]], 9223372036854775804
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP13]]
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP12]], ptr [[TMP14]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP16]]
+; CHECK-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw <vscale x 4 x i32> [[TMP13]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP15:%.*]] = and i64 [[INDEX]], 9223372036854775804
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP15]]
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP14]], ptr [[TMP16]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK: middle.block:
@@ -640,12 +640,12 @@ define void @load_gap_reverse(%pair* noalias nocapture readonly %P1, %pair* noal
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i64 [[TMP6]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[INDUCTION:%.*]] = sub <vscale x 4 x i64> shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1023, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer), [[TMP0]]
-; CHECK-NEXT: [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[DOTNEG:%.*]] = mul nsw i64 [[TMP1]], -4
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[INDUCTION:%.*]] = sub <vscale x 4 x i64> shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1023, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer), [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[DOTNEG:%.*]] = mul nsw i64 [[TMP3]], -4
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[DOTNEG]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[X:%.*]], i64 0
@@ -654,14 +654,14 @@ define void @load_gap_reverse(%pair* noalias nocapture readonly %P1, %pair* noal
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = add nsw <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[VEC_IND]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[PAIR:%.*]], ptr [[P1:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P2:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 1
-; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i64> @llvm.masked.gather.nxv4i64.nxv4p0(<vscale x 4 x ptr> [[TMP4]], i32 8, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i64> poison)
-; CHECK-NEXT: [[TMP5:%.*]] = sub nsw <vscale x 4 x i64> [[WIDE_MASKED_GATHER]], [[VEC_IND]]
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i64.nxv4p0(<vscale x 4 x i64> [[TMP2]], <vscale x 4 x ptr> [[TMP3]], i32 8, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i64.nxv4p0(<vscale x 4 x i64> [[TMP5]], <vscale x 4 x ptr> [[TMP4]], i32 8, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP7]]
+; CHECK-NEXT: [[TMP4:%.*]] = add nsw <vscale x 4 x i64> [[BROADCAST_SPLAT]], [[VEC_IND]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[PAIR:%.*]], ptr [[P1:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P2:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 1
+; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i64> @llvm.masked.gather.nxv4i64.nxv4p0(<vscale x 4 x ptr> [[TMP6]], i32 8, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i64> poison)
+; CHECK-NEXT: [[TMP7:%.*]] = sub nsw <vscale x 4 x i64> [[WIDE_MASKED_GATHER]], [[VEC_IND]]
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i64.nxv4p0(<vscale x 4 x i64> [[TMP4]], <vscale x 4 x ptr> [[TMP5]], i32 8, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i64.nxv4p0(<vscale x 4 x i64> [[TMP7]], <vscale x 4 x ptr> [[TMP6]], i32 8, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
@@ -709,28 +709,28 @@ define void @mixed_load2_store2(i32* noalias nocapture readonly %A, i32* noalias
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP10]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP0]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
-; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 1
-; CHECK-NEXT: [[TMP4:%.*]] = mul nsw <vscale x 4 x i32> [[TMP2]], [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
+; CHECK-NEXT: [[TMP5:%.*]] = or disjoint i64 [[OFFSET_IDX]], 1
+; CHECK-NEXT: [[TMP6:%.*]] = mul nsw <vscale x 4 x i32> [[TMP4]], [[TMP3]]
; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 0
-; CHECK-NEXT: [[TMP6:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 1
-; CHECK-NEXT: [[TMP7:%.*]] = add nsw <vscale x 4 x i32> [[TMP6]], [[TMP5]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP8]], i64 -1
-; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP7]])
-; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP9]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP11]]
+; CHECK-NEXT: [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 0
+; CHECK-NEXT: [[TMP8:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 1
+; CHECK-NEXT: [[TMP9:%.*]] = add nsw <vscale x 4 x i32> [[TMP8]], [[TMP7]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i64 -4
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP9]])
+; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP11]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 512
; CHECK-NEXT: br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
; CHECK: middle.block:
@@ -802,27 +802,27 @@ define void @int_float_struct(%struct.IntFloat* nocapture readonly %p) #0 {
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i64 [[TMP6]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x float> [ insertelement (<vscale x 4 x float> zeroinitializer, float undef, i32 0), [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 4 x i32> [ insertelement (<vscale x 4 x i32> zeroinitializer, i32 undef, i32 0), [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_INTFLOAT:%.*]], ptr [[P:%.*]], i64 [[INDEX]], i32 0
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP0]], align 4
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x float> [ insertelement (<vscale x 4 x float> zeroinitializer, float undef, i32 0), [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 4 x i32> [ insertelement (<vscale x 4 x i32> zeroinitializer, i32 undef, i32 0), [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_INTFLOAT:%.*]], ptr [[P:%.*]], i64 [[INDEX]], i32 0
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[TMP2:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
-; CHECK-NEXT: [[TMP3:%.*]] = bitcast <vscale x 4 x i32> [[TMP2]] to <vscale x 4 x float>
-; CHECK-NEXT: [[TMP4]] = add <vscale x 4 x i32> [[TMP1]], [[VEC_PHI1]]
-; CHECK-NEXT: [[TMP5]] = fadd fast <vscale x 4 x float> [[VEC_PHI]], [[TMP3]]
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP7]]
+; CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP4:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
+; CHECK-NEXT: [[TMP5:%.*]] = bitcast <vscale x 4 x i32> [[TMP4]] to <vscale x 4 x float>
+; CHECK-NEXT: [[TMP6]] = add <vscale x 4 x i32> [[TMP3]], [[VEC_PHI1]]
+; CHECK-NEXT: [[TMP7]] = fadd fast <vscale x 4 x float> [[VEC_PHI]], [[TMP5]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
; CHECK: middle.block:
-; CHECK-NEXT: [[TMP9:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP4]])
-; CHECK-NEXT: [[TMP10:%.*]] = call fast float @llvm.vector.reduce.fadd.nxv4f32(float -0.000000e+00, <vscale x 4 x float> [[TMP5]])
+; CHECK-NEXT: [[TMP9:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP6]])
+; CHECK-NEXT: [[TMP10:%.*]] = call fast float @llvm.vector.reduce.fadd.nxv4f32(float -0.000000e+00, <vscale x 4 x float> [[TMP7]])
; CHECK-NEXT: br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
@@ -892,28 +892,28 @@ define void @PR27626_0(%pair.i32 *%p, i32 %z, i64 %n) #1 {
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP6:%.*]] = select i1 [[TMP5]], i64 [[TMP3]], i64 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub nsw i64 [[SMAX]], [[TMP6]]
-; CHECK-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[TMP14]], 2
-; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[TMP8]], 2
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP9]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP10]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP11]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[Z:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP7]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]], i32 1
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP10]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[TMP12:%.*]] = extractelement <vscale x 4 x ptr> [[TMP10]], i64 0
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP12]], align 4
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]], i32 1
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP12]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[TMP14:%.*]] = extractelement <vscale x 4 x ptr> [[TMP12]], i64 0
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP14]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP13]], <vscale x 4 x ptr> [[TMP11]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP15]]
+; CHECK-NEXT: [[TMP15:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP15]], <vscale x 4 x ptr> [[TMP13]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
@@ -979,35 +979,35 @@ define i32 @PR27626_1(%pair.i32 *%p, i64 %n) #1 {
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP6:%.*]] = select i1 [[TMP5]], i64 [[TMP3]], i64 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub nsw i64 [[SMAX]], [[TMP6]]
-; CHECK-NEXT: [[TMP16:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP17:%.*]] = shl nuw nsw i64 [[TMP16]], 2
-; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[TMP8]], 2
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP9]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP10]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP11]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP7]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]], i32 0
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]], i32 1
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP10]], align 4
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP17:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]], i32 0
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]], i32 1
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP12]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP12:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP12]], <vscale x 4 x ptr> [[TMP11]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[TMP13:%.*]] = extractelement <vscale x 4 x ptr> [[TMP11]], i64 0
-; CHECK-NEXT: [[WIDE_VEC1:%.*]] = load <vscale x 8 x i32>, ptr [[TMP13]], align 4
+; CHECK-NEXT: [[TMP14:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP14]], <vscale x 4 x ptr> [[TMP13]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[TMP15:%.*]] = extractelement <vscale x 4 x ptr> [[TMP13]], i64 0
+; CHECK-NEXT: [[WIDE_VEC1:%.*]] = load <vscale x 8 x i32>, ptr [[TMP15]], align 4
; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC1]])
-; CHECK-NEXT: [[TMP14:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 0
-; CHECK-NEXT: [[TMP15]] = add <vscale x 4 x i32> [[TMP14]], [[VEC_PHI]]
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP17]]
+; CHECK-NEXT: [[TMP16:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 0
+; CHECK-NEXT: [[TMP17]] = add <vscale x 4 x i32> [[TMP16]], [[VEC_PHI]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
; CHECK: middle.block:
-; CHECK-NEXT: [[TMP19:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP15]])
+; CHECK-NEXT: [[TMP19:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP17]])
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
@@ -1074,28 +1074,28 @@ define void @PR27626_2(%pair.i32 *%p, i64 %n, i32 %z) #1 {
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP6:%.*]] = select i1 [[TMP5]], i64 [[TMP3]], i64 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub nsw i64 [[SMAX]], [[TMP6]]
-; CHECK-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[TMP14]], 2
-; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[TMP8]], 2
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP9]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP10]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP11]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[Z:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP7]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 -1, i32 0
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]], i32 1
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP10]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP11]], align 4
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], <vscale x 4 x i64> [[VEC_IND]], i32 0
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 -8
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]], i32 1
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP12]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP13]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP13]], <vscale x 4 x ptr> [[TMP12]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP15]]
+; CHECK-NEXT: [[TMP15:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP15]], <vscale x 4 x ptr> [[TMP14]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
@@ -1107,7 +1107,7 @@ define void @PR27626_2(%pair.i32 *%p, i64 %n, i32 %z) #1 {
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]], i32 0
-; CHECK-NEXT: [[P_I_MINUS_1_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 -1, i32 0
+; CHECK-NEXT: [[P_I_MINUS_1_X:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 -8
; CHECK-NEXT: [[P_I_Y:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]], i32 1
; CHECK-NEXT: store i32 [[Z]], ptr [[P_I_X]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[P_I_MINUS_1_X]], align 4
@@ -1164,36 +1164,36 @@ define i32 @PR27626_3(%pair.i32 *%p, i64 %n, i32 %z) #1 {
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[N_MOD_VF]], 0
; CHECK-NEXT: [[TMP6:%.*]] = select i1 [[TMP5]], i64 [[TMP3]], i64 [[N_MOD_VF]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub nsw i64 [[SMAX]], [[TMP6]]
-; CHECK-NEXT: [[TMP17:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP18:%.*]] = shl nuw nsw i64 [[TMP17]], 2
-; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[TMP8]], 2
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP9]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP10]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP11]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP7]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP16:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]], i32 0
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]], i32 1
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[TMP10]], i32 1
-; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP11]], align 4
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP12:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]], i32 0
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]], i32 1
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[TMP12]], i32 1
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP13]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP14:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP14]], <vscale x 4 x ptr> [[TMP13]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[WIDE_VEC1:%.*]] = load <vscale x 8 x i32>, ptr [[TMP12]], align 4
+; CHECK-NEXT: [[TMP16:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP16]], <vscale x 4 x ptr> [[TMP15]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[WIDE_VEC1:%.*]] = load <vscale x 8 x i32>, ptr [[TMP14]], align 4
; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC1]])
-; CHECK-NEXT: [[TMP15:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 0
-; CHECK-NEXT: [[TMP16]] = add <vscale x 4 x i32> [[TMP15]], [[VEC_PHI]]
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP18]]
+; CHECK-NEXT: [[TMP17:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 0
+; CHECK-NEXT: [[TMP18]] = add <vscale x 4 x i32> [[TMP17]], [[VEC_PHI]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
; CHECK: middle.block:
-; CHECK-NEXT: [[TMP20:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP16]])
+; CHECK-NEXT: [[TMP20:%.*]] = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> [[TMP18]])
; CHECK-NEXT: br label [[SCALAR_PH]]
; CHECK: scalar.ph:
; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
@@ -1267,13 +1267,13 @@ define void @PR27626_4(i32 *%a, i32 %x, i32 %y, i32 %z, i64 %n) #1 {
; CHECK-NEXT: [[DOTNEG:%.*]] = mul nsw i64 [[TMP5]], -4
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], [[DOTNEG]]
; CHECK-NEXT: [[IND_END:%.*]] = shl nuw i64 [[N_VEC]], 1
-; CHECK-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[TMP14]], 2
-; CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP7:%.*]] = shl <vscale x 4 x i64> [[TMP6]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[TMP8]], 3
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP9]], i64 0
+; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i64 [[TMP6]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP9:%.*]] = shl <vscale x 4 x i64> [[TMP8]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP10:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP10]], 3
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP11]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[X:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
@@ -1284,16 +1284,16 @@ define void @PR27626_4(i32 *%a, i32 %x, i32 %y, i32 %z, i64 %n) #1 {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP7]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP10:%.*]] = or disjoint i64 [[OFFSET_IDX]], 1
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], <vscale x 4 x i64> [[VEC_IND]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP10]]
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP11]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP12]], i64 -1
+; CHECK-NEXT: [[TMP12:%.*]] = or disjoint i64 [[OFFSET_IDX]], 1
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP12]]
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP13]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP14]], i64 -4
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.experimental.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[BROADCAST_SPLAT2]], <vscale x 4 x i32> [[BROADCAST_SPLAT4]])
-; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP13]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP15]]
+; CHECK-NEXT: store <vscale x 8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP15]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP7]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
@@ -1363,14 +1363,14 @@ define void @PR27626_5(i32 *%a, i32 %x, i32 %y, i32 %z, i64 %n) #1 {
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP2]], [[DOTNEG]]
; CHECK-NEXT: [[TMP6:%.*]] = shl nuw i64 [[N_VEC]], 1
; CHECK-NEXT: [[IND_END:%.*]] = or disjoint i64 [[TMP6]], 3
-; CHECK-NEXT: [[TMP16:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP17:%.*]] = shl nuw nsw i64 [[TMP16]], 2
-; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP8:%.*]] = shl <vscale x 4 x i64> [[TMP7]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[INDUCTION:%.*]] = add <vscale x 4 x i64> [[TMP8]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 3, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw i64 [[TMP9]], 3
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP10]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP10:%.*]] = shl <vscale x 4 x i64> [[TMP9]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[INDUCTION:%.*]] = add <vscale x 4 x i64> [[TMP10]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 3, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP11:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw i64 [[TMP11]], 3
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP12]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[X:%.*]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
@@ -1382,15 +1382,15 @@ define void @PR27626_5(i32 *%a, i32 %x, i32 %y, i32 %z, i64 %n) #1 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP11:%.*]] = add <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 -1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP12:%.*]] = add <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 -3, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], <vscale x 4 x i64> [[VEC_IND]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[A]], <vscale x 4 x i64> [[TMP11]]
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], <vscale x 4 x i64> [[TMP12]]
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP14]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT2]], <vscale x 4 x ptr> [[TMP15]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT4]], <vscale x 4 x ptr> [[TMP13]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP17]]
+; CHECK-NEXT: [[TMP13:%.*]] = add <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 -1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP14:%.*]] = add <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 -3, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[A]], <vscale x 4 x i64> [[TMP13]]
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], <vscale x 4 x i64> [[TMP14]]
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> [[TMP16]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT2]], <vscale x 4 x ptr> [[TMP17]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT4]], <vscale x 4 x ptr> [[TMP15]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
@@ -1404,9 +1404,9 @@ define void @PR27626_5(i32 *%a, i32 %x, i32 %y, i32 %z, i64 %n) #1 {
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[A_I:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I]]
; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[A_I_MINUS_1:%.*]] = getelementptr i32, ptr [[TMP19]], i64 -1
+; CHECK-NEXT: [[A_I_MINUS_1:%.*]] = getelementptr i8, ptr [[TMP19]], i64 -4
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[A_I_MINUS_3:%.*]] = getelementptr i32, ptr [[TMP20]], i64 -3
+; CHECK-NEXT: [[A_I_MINUS_3:%.*]] = getelementptr i8, ptr [[TMP20]], i64 -12
; CHECK-NEXT: store i32 [[X]], ptr [[A_I_MINUS_1]], align 4
; CHECK-NEXT: store i32 [[Y]], ptr [[A_I_MINUS_3]], align 4
; CHECK-NEXT: store i32 [[Z]], ptr [[A_I]], align 4
@@ -1473,38 +1473,38 @@ define void @PR34743(i16* %a, i32* %b, i64 %n) #1 {
; CHECK-NEXT: [[DOTNEG:%.*]] = mul nsw i64 [[TMP8]], -4
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP1]], [[DOTNEG]]
; CHECK-NEXT: [[IND_END:%.*]] = shl i64 [[N_VEC]], 1
-; CHECK-NEXT: [[TMP27:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP28:%.*]] = shl nuw nsw i64 [[TMP27]], 2
-; CHECK-NEXT: [[TMP9:%.*]] = call i32 @llvm.vscale.i32()
-; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw i32 [[TMP9]], 2
-; CHECK-NEXT: [[TMP11:%.*]] = add nsw i32 [[TMP10]], -1
-; CHECK-NEXT: [[VECTOR_RECUR_INIT:%.*]] = insertelement <vscale x 4 x i16> poison, i16 [[DOTPRE]], i32 [[TMP11]]
-; CHECK-NEXT: [[TMP12:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
-; CHECK-NEXT: [[TMP13:%.*]] = shl <vscale x 4 x i64> [[TMP12]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[TMP14]], 3
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP15]], i64 0
+; CHECK-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw i64 [[TMP9]], 2
+; CHECK-NEXT: [[TMP11:%.*]] = call i32 @llvm.vscale.i32()
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = add nsw i32 [[TMP12]], -1
+; CHECK-NEXT: [[VECTOR_RECUR_INIT:%.*]] = insertelement <vscale x 4 x i16> poison, i16 [[DOTPRE]], i32 [[TMP13]]
+; CHECK-NEXT: [[TMP14:%.*]] = call <vscale x 4 x i64> @llvm.experimental.stepvector.nxv4i64()
+; CHECK-NEXT: [[TMP15:%.*]] = shl <vscale x 4 x i64> [[TMP14]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP16:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP17:%.*]] = shl nuw nsw i64 [[TMP16]], 3
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP17]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <vscale x 4 x i64> [[DOTSPLATINSERT]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <vscale x 4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_MASKED_GATHER4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP13]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP16:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP17:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 2, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i16, ptr [[A]], <vscale x 4 x i64> [[TMP16]]
-; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP18]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison), !alias.scope [[META34:![0-9]+]]
-; CHECK-NEXT: [[TMP19:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER]] to <vscale x 4 x i32>
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i16, ptr [[A]], <vscale x 4 x i64> [[TMP17]]
-; CHECK-NEXT: [[WIDE_MASKED_GATHER4]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP20]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison), !alias.scope [[META34]]
-; CHECK-NEXT: [[TMP21:%.*]] = call <vscale x 4 x i16> @llvm.experimental.vector.splice.nxv4i16(<vscale x 4 x i16> [[VECTOR_RECUR]], <vscale x 4 x i16> [[WIDE_MASKED_GATHER4]], i32 -1)
-; CHECK-NEXT: [[TMP22:%.*]] = sext <vscale x 4 x i16> [[TMP21]] to <vscale x 4 x i32>
-; CHECK-NEXT: [[TMP23:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER4]] to <vscale x 4 x i32>
-; CHECK-NEXT: [[TMP24:%.*]] = mul nsw <vscale x 4 x i32> [[TMP22]], [[TMP19]]
-; CHECK-NEXT: [[TMP25:%.*]] = mul nsw <vscale x 4 x i32> [[TMP24]], [[TMP23]]
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP25]], ptr [[TMP26]], align 4, !alias.scope [[META37:![0-9]+]], !noalias [[META34]]
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP28]]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP15]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP18:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 1, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP19:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], shufflevector (<vscale x 4 x i64> insertelement (<vscale x 4 x i64> poison, i64 2, i64 0), <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i16, ptr [[A]], <vscale x 4 x i64> [[TMP18]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP20]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison), !alias.scope [[META34:![0-9]+]]
+; CHECK-NEXT: [[TMP21:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER]] to <vscale x 4 x i32>
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i16, ptr [[A]], <vscale x 4 x i64> [[TMP19]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER4]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> [[TMP22]], i32 4, <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x i16> poison), !alias.scope [[META34]]
+; CHECK-NEXT: [[TMP23:%.*]] = call <vscale x 4 x i16> @llvm.experimental.vector.splice.nxv4i16(<vscale x 4 x i16> [[VECTOR_RECUR]], <vscale x 4 x i16> [[WIDE_MASKED_GATHER4]], i32 -1)
+; CHECK-NEXT: [[TMP24:%.*]] = sext <vscale x 4 x i16> [[TMP23]] to <vscale x 4 x i32>
+; CHECK-NEXT: [[TMP25:%.*]] = sext <vscale x 4 x i16> [[WIDE_MASKED_GATHER4]] to <vscale x 4 x i32>
+; CHECK-NEXT: [[TMP26:%.*]] = mul nsw <vscale x 4 x i32> [[TMP24]], [[TMP21]]
+; CHECK-NEXT: [[TMP27:%.*]] = mul nsw <vscale x 4 x i32> [[TMP26]], [[TMP25]]
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP27]], ptr [[TMP28]], align 4, !alias.scope [[META37:![0-9]+]], !noalias [[META34]]
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP10]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
; CHECK-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP39:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll
index 1682d0740d0f5b..7226048c478d48 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll
@@ -27,43 +27,43 @@ define void @widen_ptr_phi_unrolled(ptr noalias nocapture %a, ptr noalias nocapt
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[DOTNEG]], [[N]]
; CHECK-NEXT: [[TMP3:%.*]] = shl i64 [[N_VEC]], 3
; CHECK-NEXT: [[IND_END:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP25:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP26:%.*]] = shl nuw nsw i64 [[TMP25]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i64 [[TMP4]], 3
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[INDEX]], 3
-; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[C]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP5]], 5
-; CHECK-NEXT: [[TMP7:%.*]] = shl i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[C]], i64 [[TMP6]]
-; CHECK-NEXT: [[NEXT_GEP2:%.*]] = getelementptr i8, ptr [[TMP8]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[C]], i64 [[TMP6]]
+; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 5
+; CHECK-NEXT: [[TMP9:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[C]], i64 [[TMP8]]
+; CHECK-NEXT: [[NEXT_GEP2:%.*]] = getelementptr i8, ptr [[TMP10]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[NEXT_GEP]], align 4
; CHECK-NEXT: [[WIDE_VEC3:%.*]] = load <vscale x 8 x i32>, ptr [[NEXT_GEP2]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
-; CHECK-NEXT: [[TMP9:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
-; CHECK-NEXT: [[TMP10:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
+; CHECK-NEXT: [[TMP11:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
+; CHECK-NEXT: [[TMP12:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
; CHECK-NEXT: [[STRIDED_VEC4:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.experimental.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC3]])
-; CHECK-NEXT: [[TMP11:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC4]], 0
-; CHECK-NEXT: [[TMP12:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC4]], 1
-; CHECK-NEXT: [[TMP13:%.*]] = add nsw <vscale x 4 x i32> [[TMP9]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP14:%.*]] = add nsw <vscale x 4 x i32> [[TMP11]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP16:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP17:%.*]] = shl nuw nsw i64 [[TMP16]], 2
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP15]], i64 [[TMP17]]
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP13]], ptr [[TMP15]], align 4
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP14]], ptr [[TMP18]], align 4
-; CHECK-NEXT: [[TMP19:%.*]] = add nsw <vscale x 4 x i32> [[TMP10]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP20:%.*]] = add nsw <vscale x 4 x i32> [[TMP12]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP22:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP23:%.*]] = shl nuw nsw i64 [[TMP22]], 2
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i64 [[TMP23]]
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP19]], ptr [[TMP21]], align 4
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP20]], ptr [[TMP24]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP26]]
+; CHECK-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC4]], 0
+; CHECK-NEXT: [[TMP14:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC4]], 1
+; CHECK-NEXT: [[TMP15:%.*]] = add nsw <vscale x 4 x i32> [[TMP11]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP16:%.*]] = add nsw <vscale x 4 x i32> [[TMP13]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP19:%.*]] = shl nuw nsw i64 [[TMP18]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[TMP17]], i64 [[TMP19]]
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP15]], ptr [[TMP17]], align 4
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP16]], ptr [[TMP20]], align 4
+; CHECK-NEXT: [[TMP21:%.*]] = add nsw <vscale x 4 x i32> [[TMP12]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP22:%.*]] = add nsw <vscale x 4 x i32> [[TMP14]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP24:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP25:%.*]] = shl nuw nsw i64 [[TMP24]], 2
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[TMP23]], i64 [[TMP25]]
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP21]], ptr [[TMP23]], align 4
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP22]], ptr [[TMP26]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-NEXT: [[TMP27:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP27]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
@@ -76,9 +76,9 @@ define void @widen_ptr_phi_unrolled(ptr noalias nocapture %a, ptr noalias nocapt
; CHECK: for.body:
; CHECK-NEXT: [[PTR_014:%.*]] = phi ptr [ [[INCDEC_PTR1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[I_013:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[INCDEC_PTR:%.*]] = getelementptr inbounds i32, ptr [[PTR_014]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR:%.*]] = getelementptr inbounds i8, ptr [[PTR_014]], i64 4
; CHECK-NEXT: [[TMP28:%.*]] = load i32, ptr [[PTR_014]], align 4
-; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i32, ptr [[PTR_014]], i64 2
+; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i8, ptr [[PTR_014]], i64 8
; CHECK-NEXT: [[TMP29:%.*]] = load i32, ptr [[INCDEC_PTR]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I_013]]
@@ -143,28 +143,28 @@ define void @widen_2ptrs_phi_unrolled(ptr noalias nocapture %dst, ptr noalias no
; CHECK-NEXT: [[IND_END:%.*]] = getelementptr i8, ptr [[SRC:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[N_VEC]], 2
; CHECK-NEXT: [[IND_END2:%.*]] = getelementptr i8, ptr [[DST:%.*]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP15:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP16:%.*]] = shl nuw nsw i64 [[TMP15]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP5]], 3
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP5:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP6:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[NEXT_GEP5:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP6]]
-; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 2
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[NEXT_GEP5:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw i64 [[TMP9]], 2
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[NEXT_GEP]], align 4
-; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <vscale x 4 x i32>, ptr [[TMP9]], align 4
-; CHECK-NEXT: [[TMP10:%.*]] = shl nsw <vscale x 4 x i32> [[WIDE_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP11:%.*]] = shl nsw <vscale x 4 x i32> [[WIDE_LOAD7]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP12:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP13:%.*]] = shl nuw nsw i64 [[TMP12]], 2
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[NEXT_GEP5]], i64 [[TMP13]]
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP10]], ptr [[NEXT_GEP5]], align 4
-; CHECK-NEXT: store <vscale x 4 x i32> [[TMP11]], ptr [[TMP14]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP16]]
+; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <vscale x 4 x i32>, ptr [[TMP11]], align 4
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw <vscale x 4 x i32> [[WIDE_LOAD]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw <vscale x 4 x i32> [[WIDE_LOAD7]], shufflevector (<vscale x 4 x i32> insertelement (<vscale x 4 x i32> poison, i32 1, i64 0), <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP14:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[TMP14]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[NEXT_GEP5]], i64 [[TMP15]]
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP12]], ptr [[NEXT_GEP5]], align 4
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP13]], ptr [[TMP16]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: middle.block:
@@ -182,8 +182,8 @@ define void @widen_2ptrs_phi_unrolled(ptr noalias nocapture %dst, ptr noalias no
; CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[S_010]], align 4
; CHECK-NEXT: [[MUL:%.*]] = shl nsw i32 [[TMP18]], 1
; CHECK-NEXT: store i32 [[MUL]], ptr [[D_09]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[D_09]], i64 1
-; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i32, ptr [[S_010]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[D_09]], i64 4
+; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i8, ptr [[S_010]], i64 4
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_011]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[N]]
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
@@ -234,30 +234,30 @@ define i32 @pointer_iv_mixed(ptr noalias %a, ptr noalias %b, i64 %n) #0 {
; CHECK-NEXT: [[IND_END:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[N_VEC]], 3
; CHECK-NEXT: [[IND_END2:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP12:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP13:%.*]] = shl nuw nsw i64 [[TMP12]], 1
+; CHECK-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP5]], 1
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[POINTER_PHI:%.*]] = phi ptr [ [[A]], [[VECTOR_PH]] ], [ [[PTR_IND:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP11:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP5]], 3
-; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.experimental.stepvector.nxv2i64()
-; CHECK-NEXT: [[VECTOR_GEP:%.*]] = shl <vscale x 2 x i64> [[TMP7]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 2, i64 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[POINTER_PHI]], <vscale x 2 x i64> [[VECTOR_GEP]]
-; CHECK-NEXT: [[TMP9:%.*]] = shl i64 [[INDEX]], 3
-; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP9]]
-; CHECK-NEXT: [[TMP10:%.*]] = extractelement <vscale x 2 x ptr> [[TMP8]], i64 0
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i32>, ptr [[TMP10]], align 8
-; CHECK-NEXT: [[TMP11]] = add <vscale x 2 x i32> [[WIDE_LOAD]], [[VEC_PHI]]
-; CHECK-NEXT: store <vscale x 2 x ptr> [[TMP8]], ptr [[NEXT_GEP]], align 8
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP13]]
-; CHECK-NEXT: [[PTR_IND]] = getelementptr i8, ptr [[POINTER_PHI]], i64 [[TMP6]]
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP7]], 3
+; CHECK-NEXT: [[TMP9:%.*]] = call <vscale x 2 x i64> @llvm.experimental.stepvector.nxv2i64()
+; CHECK-NEXT: [[VECTOR_GEP:%.*]] = shl <vscale x 2 x i64> [[TMP9]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 2, i64 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[POINTER_PHI]], <vscale x 2 x i64> [[VECTOR_GEP]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP12:%.*]] = extractelement <vscale x 2 x ptr> [[TMP10]], i64 0
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i32>, ptr [[TMP12]], align 8
+; CHECK-NEXT: [[TMP13]] = add <vscale x 2 x i32> [[WIDE_LOAD]], [[VEC_PHI]]
+; CHECK-NEXT: store <vscale x 2 x ptr> [[TMP10]], ptr [[NEXT_GEP]], align 8
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
+; CHECK-NEXT: [[PTR_IND]] = getelementptr i8, ptr [[POINTER_PHI]], i64 [[TMP8]]
; CHECK-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: middle.block:
-; CHECK-NEXT: [[TMP15:%.*]] = call i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32> [[TMP11]])
+; CHECK-NEXT: [[TMP15:%.*]] = call i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32> [[TMP13]])
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
@@ -274,8 +274,8 @@ define i32 @pointer_iv_mixed(ptr noalias %a, ptr noalias %b, i64 %n) #0 {
; CHECK-NEXT: [[VAR1:%.*]] = load i32, ptr [[P]], align 8
; CHECK-NEXT: [[VAR2]] = add i32 [[VAR1]], [[VAR0]]
; CHECK-NEXT: store ptr [[P]], ptr [[Q]], align 8
-; CHECK-NEXT: [[VAR3]] = getelementptr inbounds i32, ptr [[P]], i64 1
-; CHECK-NEXT: [[VAR4]] = getelementptr inbounds ptr, ptr [[Q]], i64 1
+; CHECK-NEXT: [[VAR3]] = getelementptr inbounds i8, ptr [[P]], i64 4
+; CHECK-NEXT: [[VAR4]] = getelementptr inbounds i8, ptr [[Q]], i64 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP8:![0-9]+]]
@@ -310,22 +310,22 @@ define void @phi_used_in_vector_compare_and_scalar_indvar_update_and_store(ptr %
; CHECK-NEXT: entry:
; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i64 [[TMP6]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 1
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[POINTER_PHI:%.*]] = phi ptr [ [[PTR:%.*]], [[VECTOR_PH]] ], [ [[PTR_IND:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[TMP0]], 2
-; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.experimental.stepvector.nxv2i64()
-; CHECK-NEXT: [[VECTOR_GEP:%.*]] = shl <vscale x 2 x i64> [[TMP2]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 1, i64 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[POINTER_PHI]], <vscale x 2 x i64> [[VECTOR_GEP]]
-; CHECK-NEXT: [[TMP4:%.*]] = icmp ne <vscale x 2 x ptr> [[TMP3]], zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = extractelement <vscale x 2 x ptr> [[TMP3]], i64 0
-; CHECK-NEXT: call void @llvm.masked.store.nxv2i16.p0(<vscale x 2 x i16> zeroinitializer, ptr [[TMP5]], i32 2, <vscale x 2 x i1> [[TMP4]])
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP7]]
-; CHECK-NEXT: [[PTR_IND]] = getelementptr i8, ptr [[POINTER_PHI]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[TMP2]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x i64> @llvm.experimental.stepvector.nxv2i64()
+; CHECK-NEXT: [[VECTOR_GEP:%.*]] = shl <vscale x 2 x i64> [[TMP4]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 1, i64 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[POINTER_PHI]], <vscale x 2 x i64> [[VECTOR_GEP]]
+; CHECK-NEXT: [[TMP6:%.*]] = icmp ne <vscale x 2 x ptr> [[TMP5]], zeroinitializer
+; CHECK-NEXT: [[TMP7:%.*]] = extractelement <vscale x 2 x ptr> [[TMP5]], i64 0
+; CHECK-NEXT: call void @llvm.masked.store.nxv2i16.p0(<vscale x 2 x i16> zeroinitializer, ptr [[TMP7]], i32 2, <vscale x 2 x i1> [[TMP6]])
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
+; CHECK-NEXT: [[PTR_IND]] = getelementptr i8, ptr [[POINTER_PHI]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; CHECK-NEXT: br i1 [[TMP8]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK: middle.block:
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll b/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll
index 347ec55f6a992c..d5ace655fdcc12 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll
@@ -34,8 +34,8 @@ define void @vector_reverse_mask_v4i1(ptr noalias %a, ptr noalias %cond, i64 %N)
; CHECK-NEXT: [[TMP0:%.*]] = xor i64 [[INDEX]], -1
; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[TMP0]], [[N]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[COND:%.*]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds double, ptr [[TMP2]], i64 -3
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[TMP2]], i64 -7
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 -24
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 -56
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[REVERSE:%.*]] = shufflevector <4 x double> [[WIDE_LOAD]], <4 x double> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x double>, ptr [[TMP4]], align 8
@@ -43,8 +43,8 @@ define void @vector_reverse_mask_v4i1(ptr noalias %a, ptr noalias %cond, i64 %N)
; CHECK-NEXT: [[TMP5:%.*]] = fcmp une <4 x double> [[REVERSE]], zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = fcmp une <4 x double> [[REVERSE2]], zeroinitializer
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr double, ptr [[A:%.*]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr double, ptr [[TMP7]], i64 -3
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr double, ptr [[TMP7]], i64 -7
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 -24
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP7]], i64 -56
; CHECK-NEXT: [[REVERSE3:%.*]] = shufflevector <4 x i1> [[TMP5]], <4 x i1> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
; CHECK-NEXT: [[REVERSE4:%.*]] = shufflevector <4 x i1> [[TMP6]], <4 x i1> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x double> @llvm.masked.load.v4f64.p0(ptr [[TMP8]], i32 8, <4 x i1> [[REVERSE3]], <4 x double> poison)
diff --git a/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll b/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll
index 4988ba0d7d37aa..8f768d959d8b1e 100644
--- a/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll
+++ b/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll
@@ -11,27 +11,27 @@ define half @vectorize_v2f16_loop(ptr addrspace(1) noalias %s) {
; GFX9-NEXT: br label [[VECTOR_BODY:%.*]]
; GFX9: vector.body:
; GFX9-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; GFX9-NEXT: [[VEC_PHI:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; GFX9-NEXT: [[VEC_PHI1:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; GFX9-NEXT: [[VEC_PHI:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
+; GFX9-NEXT: [[VEC_PHI1:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
; GFX9-NEXT: [[TMP0:%.*]] = getelementptr inbounds half, ptr addrspace(1) [[S:%.*]], i64 [[INDEX]]
-; GFX9-NEXT: [[TMP2:%.*]] = getelementptr inbounds half, ptr addrspace(1) [[TMP0]], i64 2
+; GFX9-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP0]], i64 4
; GFX9-NEXT: [[WIDE_LOAD:%.*]] = load <2 x half>, ptr addrspace(1) [[TMP0]], align 2
-; GFX9-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x half>, ptr addrspace(1) [[TMP2]], align 2
-; GFX9-NEXT: [[TMP4]] = fadd fast <2 x half> [[VEC_PHI]], [[WIDE_LOAD]]
-; GFX9-NEXT: [[TMP5]] = fadd fast <2 x half> [[VEC_PHI1]], [[WIDE_LOAD2]]
+; GFX9-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x half>, ptr addrspace(1) [[TMP1]], align 2
+; GFX9-NEXT: [[TMP2]] = fadd fast <2 x half> [[VEC_PHI]], [[WIDE_LOAD]]
+; GFX9-NEXT: [[TMP3]] = fadd fast <2 x half> [[VEC_PHI1]], [[WIDE_LOAD2]]
; GFX9-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; GFX9-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; GFX9-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; GFX9-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; GFX9-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; GFX9: middle.block:
-; GFX9-NEXT: [[BIN_RDX:%.*]] = fadd fast <2 x half> [[TMP5]], [[TMP4]]
-; GFX9-NEXT: [[TMP7:%.*]] = call fast half @llvm.vector.reduce.fadd.v2f16(half 0xH8000, <2 x half> [[BIN_RDX]])
+; GFX9-NEXT: [[BIN_RDX:%.*]] = fadd fast <2 x half> [[TMP3]], [[TMP2]]
+; GFX9-NEXT: [[TMP5:%.*]] = call fast half @llvm.vector.reduce.fadd.v2f16(half 0xH8000, <2 x half> [[BIN_RDX]])
; GFX9-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; GFX9: scalar.ph:
; GFX9-NEXT: br label [[FOR_BODY:%.*]]
; GFX9: for.body:
; GFX9-NEXT: br i1 poison, label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
; GFX9: for.end:
-; GFX9-NEXT: [[ADD_LCSSA:%.*]] = phi half [ poison, [[FOR_BODY]] ], [ [[TMP7]], [[MIDDLE_BLOCK]] ]
+; GFX9-NEXT: [[ADD_LCSSA:%.*]] = phi half [ poison, [[FOR_BODY]] ], [ [[TMP5]], [[MIDDLE_BLOCK]] ]
; GFX9-NEXT: ret half [[ADD_LCSSA]]
;
; VI-LABEL: @vectorize_v2f16_loop(
@@ -41,27 +41,27 @@ define half @vectorize_v2f16_loop(ptr addrspace(1) noalias %s) {
; VI-NEXT: br label [[VECTOR_BODY:%.*]]
; VI: vector.body:
; VI-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VI-NEXT: [[VEC_PHI:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; VI-NEXT: [[VEC_PHI1:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
+; VI-NEXT: [[VEC_PHI:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
+; VI-NEXT: [[VEC_PHI1:%.*]] = phi <2 x half> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
; VI-NEXT: [[TMP0:%.*]] = getelementptr inbounds half, ptr addrspace(1) [[S:%.*]], i64 [[INDEX]]
-; VI-NEXT: [[TMP2:%.*]] = getelementptr inbounds half, ptr addrspace(1) [[TMP0]], i64 2
+; VI-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP0]], i64 4
; VI-NEXT: [[WIDE_LOAD:%.*]] = load <2 x half>, ptr addrspace(1) [[TMP0]], align 2
-; VI-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x half>, ptr addrspace(1) [[TMP2]], align 2
-; VI-NEXT: [[TMP4]] = fadd fast <2 x half> [[VEC_PHI]], [[WIDE_LOAD]]
-; VI-NEXT: [[TMP5]] = fadd fast <2 x half> [[VEC_PHI1]], [[WIDE_LOAD2]]
+; VI-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x half>, ptr addrspace(1) [[TMP1]], align 2
+; VI-NEXT: [[TMP2]] = fadd fast <2 x half> [[VEC_PHI]], [[WIDE_LOAD]]
+; VI-NEXT: [[TMP3]] = fadd fast <2 x half> [[VEC_PHI1]], [[WIDE_LOAD2]]
; VI-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; VI-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; VI-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; VI-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; VI-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; VI: middle.block:
-; VI-NEXT: [[BIN_RDX:%.*]] = fadd fast <2 x half> [[TMP5]], [[TMP4]]
-; VI-NEXT: [[TMP7:%.*]] = call fast half @llvm.vector.reduce.fadd.v2f16(half 0xH8000, <2 x half> [[BIN_RDX]])
+; VI-NEXT: [[BIN_RDX:%.*]] = fadd fast <2 x half> [[TMP3]], [[TMP2]]
+; VI-NEXT: [[TMP5:%.*]] = call fast half @llvm.vector.reduce.fadd.v2f16(half 0xH8000, <2 x half> [[BIN_RDX]])
; VI-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
; VI: scalar.ph:
; VI-NEXT: br label [[FOR_BODY:%.*]]
; VI: for.body:
; VI-NEXT: br i1 poison, label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
; VI: for.end:
-; VI-NEXT: [[ADD_LCSSA:%.*]] = phi half [ poison, [[FOR_BODY]] ], [ [[TMP7]], [[MIDDLE_BLOCK]] ]
+; VI-NEXT: [[ADD_LCSSA:%.*]] = phi half [ poison, [[FOR_BODY]] ], [ [[TMP5]], [[MIDDLE_BLOCK]] ]
; VI-NEXT: ret half [[ADD_LCSSA]]
;
; CI-LABEL: @vectorize_v2f16_loop(
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll
index c1492ec5dbb749..45b84a0b5e856c 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll
@@ -60,7 +60,7 @@ define void @arm_abs_q7(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32 %
; CHECK-NEXT: store i8 [[COND11]], ptr [[PDST_ADDR_020]], align 1
; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_021]], -1
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0
-; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: while.end:
; CHECK-NEXT: ret void
;
@@ -131,7 +131,7 @@ define void @arm_abs_q15(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32
; CHECK-NEXT: store <8 x i16> [[TMP9]], ptr [[NEXT_GEP7]], align 2
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N_VEC]], [[BLOCKSIZE]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[WHILE_END]], label [[SCALAR_PH]]
@@ -144,18 +144,18 @@ define void @arm_abs_q15(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32
; CHECK-NEXT: [[PSRC_ADDR_023:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[BLKCNT_022:%.*]] = phi i32 [ [[DEC:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[PDST_ADDR_021:%.*]] = phi ptr [ [[INCDEC_PTR13:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL6]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i16, ptr [[PSRC_ADDR_023]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRC_ADDR_023]], i32 2
; CHECK-NEXT: [[TMP11:%.*]] = load i16, ptr [[PSRC_ADDR_023]], align 2
; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i16 [[TMP11]], 0
; CHECK-NEXT: [[CMP5:%.*]] = icmp eq i16 [[TMP11]], -32768
; CHECK-NEXT: [[SUB:%.*]] = sub i16 0, [[TMP11]]
; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP5]], i16 32767, i16 [[SUB]]
; CHECK-NEXT: [[COND11:%.*]] = select i1 [[CMP1]], i16 [[TMP11]], i16 [[COND]]
-; CHECK-NEXT: [[INCDEC_PTR13]] = getelementptr inbounds i16, ptr [[PDST_ADDR_021]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR13]] = getelementptr inbounds i8, ptr [[PDST_ADDR_021]], i32 2
; CHECK-NEXT: store i16 [[COND11]], ptr [[PDST_ADDR_021]], align 2
; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_022]], -1
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0
-; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: while.end:
; CHECK-NEXT: ret void
;
@@ -226,7 +226,7 @@ define void @arm_abs_q31(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32
; CHECK-NEXT: store <4 x i32> [[TMP9]], ptr [[NEXT_GEP7]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[N_VEC]], [[BLOCKSIZE]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[WHILE_END]], label [[SCALAR_PH]]
@@ -239,18 +239,18 @@ define void @arm_abs_q31(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32
; CHECK-NEXT: [[PSRC_ADDR_017:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[BLKCNT_016:%.*]] = phi i32 [ [[DEC:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[PDST_ADDR_015:%.*]] = phi ptr [ [[INCDEC_PTR7:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL6]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[PSRC_ADDR_017]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRC_ADDR_017]], i32 4
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[PSRC_ADDR_017]], align 4
; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i32 [[TMP11]], 0
; CHECK-NEXT: [[CMP2:%.*]] = icmp eq i32 [[TMP11]], -2147483648
; CHECK-NEXT: [[SUB:%.*]] = sub nsw i32 0, [[TMP11]]
; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP2]], i32 2147483647, i32 [[SUB]]
; CHECK-NEXT: [[COND6:%.*]] = select i1 [[CMP1]], i32 [[TMP11]], i32 [[COND]]
-; CHECK-NEXT: [[INCDEC_PTR7]] = getelementptr inbounds i32, ptr [[PDST_ADDR_015]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR7]] = getelementptr inbounds i8, ptr [[PDST_ADDR_015]], i32 4
; CHECK-NEXT: store i32 [[COND6]], ptr [[PDST_ADDR_015]], align 4
; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_016]], -1
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0
-; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: while.end:
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll
index 1dd0347e4d286f..18caa9cc16f350 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll
@@ -1338,7 +1338,7 @@ define i32 @reduction_interleave_group(i32 %n, ptr %arr) #0 {
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i32 [[INDEX]], 1
; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i32 [[OFFSET_IDX]], 1
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[ARR:%.*]], i32 [[TMP3]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 -1
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i32 -4
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll
index 5d85a4cd73fddf..938e1deab0b6c9 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll
@@ -65,9 +65,9 @@ define float @test(ptr nocapture readonly %pA, ptr nocapture readonly %pB, i32 %
; CHECK-NEXT: [[PB_ADDR_019:%.*]] = phi ptr [ [[INCDEC_PTR1:%.*]], [[IF_END]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[BLOCKSIZE_ADDR_018:%.*]] = phi i32 [ [[DEC:%.*]], [[IF_END]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[ACCUM_017:%.*]] = phi float [ [[ACCUM_1:%.*]], [[IF_END]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[PA_ADDR_020]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PA_ADDR_020]], i32 4
; CHECK-NEXT: [[TMP15:%.*]] = load float, ptr [[PA_ADDR_020]], align 4
-; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds float, ptr [[PB_ADDR_019]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i8, ptr [[PB_ADDR_019]], i32 4
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[PB_ADDR_019]], align 4
; CHECK-NEXT: [[CMP2:%.*]] = fcmp fast une float [[TMP15]], 0.000000e+00
; CHECK-NEXT: [[CMP3:%.*]] = fcmp fast une float [[TMP16]], 0.000000e+00
@@ -86,7 +86,7 @@ define float @test(ptr nocapture readonly %pA, ptr nocapture readonly %pB, i32 %
; CHECK-NEXT: [[ACCUM_1]] = phi float [ [[ADD4]], [[IF_THEN]] ], [ [[ACCUM_017]], [[WHILE_BODY]] ]
; CHECK-NEXT: [[DEC]] = add i32 [[BLOCKSIZE_ADDR_018]], -1
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0
-; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: while.end:
; CHECK-NEXT: [[ACCUM_0_LCSSA:%.*]] = phi float [ 0.000000e+00, [[ENTRY:%.*]] ], [ [[ACCUM_1]], [[IF_END]] ], [ [[TMP14]], [[MIDDLE_BLOCK]] ]
; CHECK-NEXT: ret float [[ACCUM_0_LCSSA]]
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll b/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll
index c3f307c234ed95..2269b774d9f31d 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll
@@ -69,10 +69,10 @@ define hidden void @pointer_phi_v4i32_add2(ptr noalias nocapture readonly %A, pt
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[A_ADDR_09]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i32, ptr [[A_ADDR_09]], i32 2
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 8
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP4]], [[Y]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_07]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[B_ADDR_07]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
@@ -123,10 +123,10 @@ define hidden void @pointer_phi_v4i32_add3(ptr noalias nocapture readonly %A, pt
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[A_ADDR_09]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i32, ptr [[A_ADDR_09]], i32 3
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 12
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP4]], [[Y]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_07]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[B_ADDR_07]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
@@ -219,10 +219,10 @@ define hidden void @pointer_phi_v8i16_add2(ptr noalias nocapture readonly %A, pt
; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_09:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[L1:%.*]] = load i16, ptr [[A_ADDR_011]], align 2
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i16, ptr [[A_ADDR_011]], i32 2
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_011]], i32 4
; CHECK-NEXT: [[CONV1:%.*]] = add i16 [[L1]], [[TMP0]]
; CHECK-NEXT: store i16 [[CONV1]], ptr [[B_ADDR_09]], align 2
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i16, ptr [[B_ADDR_09]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_09]], i32 2
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_010]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
@@ -258,10 +258,10 @@ define hidden void @pointer_phi_v8i16_add3(ptr noalias nocapture readonly %A, pt
; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_09:%.*]] = phi ptr [ [[B:%.*]], [[ENTRY]] ], [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[L1:%.*]] = load i16, ptr [[A_ADDR_011]], align 2
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i16, ptr [[A_ADDR_011]], i32 3
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_011]], i32 6
; CHECK-NEXT: [[CONV1:%.*]] = add i16 [[L1]], [[TMP0]]
; CHECK-NEXT: store i16 [[CONV1]], ptr [[B_ADDR_09]], align 2
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i16, ptr [[B_ADDR_09]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_09]], i32 2
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_010]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]]
@@ -502,10 +502,10 @@ define hidden void @pointer_phi_v4f32_add2(ptr noalias nocapture readonly %A, pt
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[A_ADDR_09]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_09]], i32 2
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 8
; CHECK-NEXT: [[ADD:%.*]] = fadd fast float [[TMP4]], [[Y]]
; CHECK-NEXT: store float [[ADD]], ptr [[B_ADDR_07]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[B_ADDR_07]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
@@ -556,10 +556,10 @@ define hidden void @pointer_phi_v4f32_add3(ptr noalias nocapture readonly %A, pt
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[A_ADDR_09]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds float, ptr [[A_ADDR_09]], i32 3
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 12
; CHECK-NEXT: [[ADD:%.*]] = fadd fast float [[TMP4]], [[Y]]
; CHECK-NEXT: store float [[ADD]], ptr [[B_ADDR_07]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds float, ptr [[B_ADDR_07]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
@@ -649,10 +649,10 @@ define hidden void @pointer_phi_v4half_add2(ptr noalias nocapture readonly %A, p
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = load half, ptr [[A_ADDR_09]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds half, ptr [[A_ADDR_09]], i32 2
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 4
; CHECK-NEXT: [[ADD:%.*]] = fadd fast half [[TMP4]], [[Y]]
; CHECK-NEXT: store half [[ADD]], ptr [[B_ADDR_07]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds half, ptr [[B_ADDR_07]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 2
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP21:![0-9]+]]
@@ -703,10 +703,10 @@ define hidden void @pointer_phi_v4half_add3(ptr noalias nocapture readonly %A, p
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = load half, ptr [[A_ADDR_09]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds half, ptr [[A_ADDR_09]], i32 3
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 6
; CHECK-NEXT: [[ADD:%.*]] = fadd fast half [[TMP4]], [[Y]]
; CHECK-NEXT: store half [[ADD]], ptr [[B_ADDR_07]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds half, ptr [[B_ADDR_07]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 2
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]]
@@ -753,7 +753,7 @@ define hidden void @pointer_phi_v4i32_uf2(ptr noalias nocapture readonly %A, ptr
; CHECK-NEXT: [[WIDE_MASKED_GATHER5:%.*]] = call <4 x i32> @llvm.masked.gather.v4i32.v4p0(<4 x ptr> [[TMP1]], i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i32> poison)
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_MASKED_GATHER]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[WIDE_MASKED_GATHER5]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i32 4
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i32 16
; CHECK-NEXT: store <4 x i32> [[TMP3]], ptr [[NEXT_GEP]], align 4
; CHECK-NEXT: store <4 x i32> [[TMP4]], ptr [[TMP5]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -767,10 +767,10 @@ define hidden void @pointer_phi_v4i32_uf2(ptr noalias nocapture readonly %A, ptr
; CHECK-NEXT: [[I_07:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 9992, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_06:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[A_ADDR_08]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i32, ptr [[A_ADDR_08]], i32 6
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_08]], i32 24
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[Y]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_06]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[B_ADDR_06]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_06]], i32 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_07]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 10000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP25:![0-9]+]]
@@ -824,9 +824,9 @@ define hidden void @pointer_phi_v4i32_uf4(ptr noalias nocapture readonly %A, ptr
; CHECK-NEXT: [[TMP6:%.*]] = add nsw <4 x i32> [[WIDE_MASKED_GATHER7]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP7:%.*]] = add nsw <4 x i32> [[WIDE_MASKED_GATHER8]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_MASKED_GATHER9]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i32 4
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i32 8
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i32 12
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i32 16
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i32 32
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i32 48
; CHECK-NEXT: store <4 x i32> [[TMP5]], ptr [[NEXT_GEP]], align 4
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr [[TMP9]], align 4
; CHECK-NEXT: store <4 x i32> [[TMP7]], ptr [[TMP10]], align 4
@@ -842,10 +842,10 @@ define hidden void @pointer_phi_v4i32_uf4(ptr noalias nocapture readonly %A, ptr
; CHECK-NEXT: [[I_07:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 9984, [[VECTOR_BODY]] ]
; CHECK-NEXT: [[B_ADDR_06:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[A_ADDR_08]], align 4
-; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i32, ptr [[A_ADDR_08]], i32 6
+; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_08]], i32 24
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[Y]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_06]], align 4
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i32, ptr [[B_ADDR_06]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_06]], i32 4
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_07]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 10000
; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP27:![0-9]+]]
@@ -886,18 +886,18 @@ define hidden void @mult_ptr_iv(ptr noalias nocapture readonly %x, ptr noalias n
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[POINTER_PHI]], <4 x i32> <i32 0, i32 3, i32 6, i32 9>
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[POINTER_PHI5]], <4 x i32> <i32 0, i32 3, i32 6, i32 9>
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, <4 x ptr> [[TMP0]], i32 1
-; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <4 x i8> @llvm.masked.gather.v4i8.v4p0(<4 x ptr> [[TMP0]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i8> poison), !alias.scope !28
+; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <4 x i8> @llvm.masked.gather.v4i8.v4p0(<4 x ptr> [[TMP0]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i8> poison), !alias.scope [[META28:![0-9]+]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, <4 x ptr> [[TMP0]], i32 2
-; CHECK-NEXT: [[WIDE_MASKED_GATHER7:%.*]] = call <4 x i8> @llvm.masked.gather.v4i8.v4p0(<4 x ptr> [[TMP2]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i8> poison), !alias.scope !28
-; CHECK-NEXT: [[WIDE_MASKED_GATHER8:%.*]] = call <4 x i8> @llvm.masked.gather.v4i8.v4p0(<4 x ptr> [[TMP3]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i8> poison), !alias.scope !28
+; CHECK-NEXT: [[WIDE_MASKED_GATHER7:%.*]] = call <4 x i8> @llvm.masked.gather.v4i8.v4p0(<4 x ptr> [[TMP2]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i8> poison), !alias.scope [[META28]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER8:%.*]] = call <4 x i8> @llvm.masked.gather.v4i8.v4p0(<4 x ptr> [[TMP3]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>, <4 x i8> poison), !alias.scope [[META28]]
; CHECK-NEXT: [[TMP4:%.*]] = mul <4 x i8> [[WIDE_MASKED_GATHER]], <i8 10, i8 10, i8 10, i8 10>
; CHECK-NEXT: [[TMP5:%.*]] = mul <4 x i8> [[WIDE_MASKED_GATHER]], [[WIDE_MASKED_GATHER7]]
; CHECK-NEXT: [[TMP6:%.*]] = mul <4 x i8> [[WIDE_MASKED_GATHER]], [[WIDE_MASKED_GATHER8]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, <4 x ptr> [[TMP1]], i32 1
-; CHECK-NEXT: call void @llvm.masked.scatter.v4i8.v4p0(<4 x i8> [[TMP4]], <4 x ptr> [[TMP1]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>), !alias.scope !31, !noalias !28
+; CHECK-NEXT: call void @llvm.masked.scatter.v4i8.v4p0(<4 x i8> [[TMP4]], <4 x ptr> [[TMP1]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>), !alias.scope [[META31:![0-9]+]], !noalias [[META28]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, <4 x ptr> [[TMP1]], i32 2
-; CHECK-NEXT: call void @llvm.masked.scatter.v4i8.v4p0(<4 x i8> [[TMP5]], <4 x ptr> [[TMP7]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>), !alias.scope !31, !noalias !28
-; CHECK-NEXT: call void @llvm.masked.scatter.v4i8.v4p0(<4 x i8> [[TMP6]], <4 x ptr> [[TMP8]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>), !alias.scope !31, !noalias !28
+; CHECK-NEXT: call void @llvm.masked.scatter.v4i8.v4p0(<4 x i8> [[TMP5]], <4 x ptr> [[TMP7]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>), !alias.scope [[META31]], !noalias [[META28]]
+; CHECK-NEXT: call void @llvm.masked.scatter.v4i8.v4p0(<4 x i8> [[TMP6]], <4 x ptr> [[TMP8]], i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>), !alias.scope [[META31]], !noalias [[META28]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT: [[PTR_IND]] = getelementptr i8, ptr [[POINTER_PHI]], i32 12
; CHECK-NEXT: [[PTR_IND6]] = getelementptr i8, ptr [[POINTER_PHI5]], i32 12
diff --git a/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll b/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
index c8c791b301633b..d7f2d8024d7c5a 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/float-induction-x86.ll
@@ -36,9 +36,9 @@ define void @fp_iv_loop1(ptr noalias nocapture %A, i32 %N) #0 {
; AUTO_VEC-NEXT: [[STEP_ADD2:%.*]] = fadd fast <8 x float> [[VEC_IND]], <float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00, float 8.000000e+00>
; AUTO_VEC-NEXT: [[STEP_ADD3:%.*]] = fadd fast <8 x float> [[VEC_IND]], <float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01, float 1.200000e+01>
; AUTO_VEC-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 8
-; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 16
-; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 24
+; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 32
+; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 64
+; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 96
; AUTO_VEC-NEXT: store <8 x float> [[VEC_IND]], ptr [[TMP1]], align 4
; AUTO_VEC-NEXT: store <8 x float> [[STEP_ADD]], ptr [[TMP2]], align 4
; AUTO_VEC-NEXT: store <8 x float> [[STEP_ADD2]], ptr [[TMP3]], align 4
@@ -211,9 +211,9 @@ define double @external_use_with_fast_math(ptr %a, i64 %n) {
; AUTO_VEC-NEXT: [[STEP_ADD2:%.*]] = fadd fast <4 x double> [[VEC_IND]], <double 2.400000e+01, double 2.400000e+01, double 2.400000e+01, double 2.400000e+01>
; AUTO_VEC-NEXT: [[STEP_ADD3:%.*]] = fadd fast <4 x double> [[VEC_IND]], <double 3.600000e+01, double 3.600000e+01, double 3.600000e+01, double 3.600000e+01>
; AUTO_VEC-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[A:%.*]], i64 [[INDEX]]
-; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr double, ptr [[TMP1]], i64 4
-; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr double, ptr [[TMP1]], i64 8
-; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr double, ptr [[TMP1]], i64 12
+; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP1]], i64 32
+; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 64
+; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[TMP1]], i64 96
; AUTO_VEC-NEXT: store <4 x double> [[VEC_IND]], ptr [[TMP1]], align 8
; AUTO_VEC-NEXT: store <4 x double> [[STEP_ADD]], ptr [[TMP2]], align 8
; AUTO_VEC-NEXT: store <4 x double> [[STEP_ADD2]], ptr [[TMP3]], align 8
@@ -377,9 +377,9 @@ define void @fadd_reassoc_FMF(ptr nocapture %p, i32 %N) {
; AUTO_VEC-NEXT: [[STEP_ADD2:%.*]] = fadd reassoc <8 x float> [[STEP_ADD]], <float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02>
; AUTO_VEC-NEXT: [[STEP_ADD3:%.*]] = fadd reassoc <8 x float> [[STEP_ADD2]], <float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02, float 3.360000e+02>
; AUTO_VEC-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[P:%.*]], i64 [[INDEX]]
-; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 8
-; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 16
-; AUTO_VEC-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 24
+; AUTO_VEC-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 32
+; AUTO_VEC-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 64
+; AUTO_VEC-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 96
; AUTO_VEC-NEXT: [[WIDE_LOAD:%.*]] = load <8 x float>, ptr [[TMP2]], align 4
; AUTO_VEC-NEXT: [[WIDE_LOAD5:%.*]] = load <8 x float>, ptr [[TMP3]], align 4
; AUTO_VEC-NEXT: [[WIDE_LOAD6:%.*]] = load <8 x float>, ptr [[TMP4]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll b/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll
index 55757157fce989..e64b02f00dfc18 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll
@@ -24,12 +24,12 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; SSE-NEXT: [[STRIDED_VEC2:%.*]] = shufflevector <8 x i32> [[WIDE_VEC1]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; SSE-NEXT: [[STRIDED_VEC3:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE-NEXT: [[STRIDED_VEC4:%.*]] = shufflevector <8 x i32> [[WIDE_VEC1]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
-; SSE-NEXT: [[TMP7:%.*]] = add nsw <4 x i32> [[STRIDED_VEC3]], [[STRIDED_VEC]]
-; SSE-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[STRIDED_VEC4]], [[STRIDED_VEC2]]
-; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; SSE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[TMP9]], i64 4
-; SSE-NEXT: store <4 x i32> [[TMP7]], ptr [[TMP9]], align 4
-; SSE-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP11]], align 4
+; SSE-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[STRIDED_VEC3]], [[STRIDED_VEC]]
+; SSE-NEXT: [[TMP6:%.*]] = add nsw <4 x i32> [[STRIDED_VEC4]], [[STRIDED_VEC2]]
+; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; SSE-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i64 16
+; SSE-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP7]], align 4
+; SSE-NEXT: store <4 x i32> [[TMP6]], ptr [[TMP8]], align 4
; SSE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; SSE-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; SSE-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
@@ -72,18 +72,18 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; AVX1-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <8 x i32> [[WIDE_VEC1]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <8 x i32> [[WIDE_VEC2]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <8 x i32> [[WIDE_VEC3]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
-; AVX1-NEXT: [[TMP15:%.*]] = add nsw <4 x i32> [[STRIDED_VEC7]], [[STRIDED_VEC]]
-; AVX1-NEXT: [[TMP16:%.*]] = add nsw <4 x i32> [[STRIDED_VEC8]], [[STRIDED_VEC4]]
-; AVX1-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[STRIDED_VEC9]], [[STRIDED_VEC5]]
-; AVX1-NEXT: [[TMP18:%.*]] = add nsw <4 x i32> [[STRIDED_VEC10]], [[STRIDED_VEC6]]
-; AVX1-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; AVX1-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 4
-; AVX1-NEXT: [[TMP23:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 8
-; AVX1-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 12
-; AVX1-NEXT: store <4 x i32> [[TMP15]], ptr [[TMP19]], align 4
-; AVX1-NEXT: store <4 x i32> [[TMP16]], ptr [[TMP21]], align 4
-; AVX1-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP23]], align 4
-; AVX1-NEXT: store <4 x i32> [[TMP18]], ptr [[TMP25]], align 4
+; AVX1-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[STRIDED_VEC7]], [[STRIDED_VEC]]
+; AVX1-NEXT: [[TMP12:%.*]] = add nsw <4 x i32> [[STRIDED_VEC8]], [[STRIDED_VEC4]]
+; AVX1-NEXT: [[TMP13:%.*]] = add nsw <4 x i32> [[STRIDED_VEC9]], [[STRIDED_VEC5]]
+; AVX1-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[STRIDED_VEC10]], [[STRIDED_VEC6]]
+; AVX1-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; AVX1-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP15]], i64 16
+; AVX1-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[TMP15]], i64 32
+; AVX1-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP15]], i64 48
+; AVX1-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP15]], align 4
+; AVX1-NEXT: store <4 x i32> [[TMP12]], ptr [[TMP16]], align 4
+; AVX1-NEXT: store <4 x i32> [[TMP13]], ptr [[TMP17]], align 4
+; AVX1-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP18]], align 4
; AVX1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; AVX1-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; AVX1-NEXT: br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
@@ -126,18 +126,18 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; AVX2-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <16 x i32> [[WIDE_VEC1]], <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
; AVX2-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <16 x i32> [[WIDE_VEC2]], <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
; AVX2-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <16 x i32> [[WIDE_VEC3]], <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
-; AVX2-NEXT: [[TMP15:%.*]] = add nsw <8 x i32> [[STRIDED_VEC7]], [[STRIDED_VEC]]
-; AVX2-NEXT: [[TMP16:%.*]] = add nsw <8 x i32> [[STRIDED_VEC8]], [[STRIDED_VEC4]]
-; AVX2-NEXT: [[TMP17:%.*]] = add nsw <8 x i32> [[STRIDED_VEC9]], [[STRIDED_VEC5]]
-; AVX2-NEXT: [[TMP18:%.*]] = add nsw <8 x i32> [[STRIDED_VEC10]], [[STRIDED_VEC6]]
-; AVX2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; AVX2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 8
-; AVX2-NEXT: [[TMP23:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 16
-; AVX2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 24
-; AVX2-NEXT: store <8 x i32> [[TMP15]], ptr [[TMP19]], align 4
-; AVX2-NEXT: store <8 x i32> [[TMP16]], ptr [[TMP21]], align 4
-; AVX2-NEXT: store <8 x i32> [[TMP17]], ptr [[TMP23]], align 4
-; AVX2-NEXT: store <8 x i32> [[TMP18]], ptr [[TMP25]], align 4
+; AVX2-NEXT: [[TMP11:%.*]] = add nsw <8 x i32> [[STRIDED_VEC7]], [[STRIDED_VEC]]
+; AVX2-NEXT: [[TMP12:%.*]] = add nsw <8 x i32> [[STRIDED_VEC8]], [[STRIDED_VEC4]]
+; AVX2-NEXT: [[TMP13:%.*]] = add nsw <8 x i32> [[STRIDED_VEC9]], [[STRIDED_VEC5]]
+; AVX2-NEXT: [[TMP14:%.*]] = add nsw <8 x i32> [[STRIDED_VEC10]], [[STRIDED_VEC6]]
+; AVX2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP15]], i64 32
+; AVX2-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[TMP15]], i64 64
+; AVX2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP15]], i64 96
+; AVX2-NEXT: store <8 x i32> [[TMP11]], ptr [[TMP15]], align 4
+; AVX2-NEXT: store <8 x i32> [[TMP12]], ptr [[TMP16]], align 4
+; AVX2-NEXT: store <8 x i32> [[TMP13]], ptr [[TMP17]], align 4
+; AVX2-NEXT: store <8 x i32> [[TMP14]], ptr [[TMP18]], align 4
; AVX2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 32
; AVX2-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; AVX2-NEXT: br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/intrinsiccost.ll b/llvm/test/Transforms/LoopVectorize/X86/intrinsiccost.ll
index 8527fb433c638d..23b653bbda380b 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/intrinsiccost.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/intrinsiccost.ll
@@ -7,11 +7,6 @@ target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"
; CHECK-COST-LABEL: uaddsat
-; CHECK-COST: Found an estimated cost of 2 for VF 1 For instruction: %1 = tail call i16 @llvm.uadd.sat.i16(i16 %0, i16 %offset)
-; CHECK-COST: Found an estimated cost of 1 for VF 2 For instruction: %1 = tail call i16 @llvm.uadd.sat.i16(i16 %0, i16 %offset)
-; CHECK-COST: Found an estimated cost of 1 for VF 4 For instruction: %1 = tail call i16 @llvm.uadd.sat.i16(i16 %0, i16 %offset)
-; CHECK-COST: Found an estimated cost of 1 for VF 8 For instruction: %1 = tail call i16 @llvm.uadd.sat.i16(i16 %0, i16 %offset)
-; CHECK-COST: Found an estimated cost of 4 for VF 16 For instruction: %1 = tail call i16 @llvm.uadd.sat.i16(i16 %0, i16 %offset)
define void @uaddsat(ptr nocapture readonly %pSrc, i16 signext %offset, ptr nocapture noalias %pDst, i32 %blockSize) #0 {
; CHECK-LABEL: @uaddsat(
@@ -36,9 +31,9 @@ define void @uaddsat(ptr nocapture readonly %pSrc, i16 signext %offset, ptr noca
; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[PSRC:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 1
; CHECK-NEXT: [[NEXT_GEP5:%.*]] = getelementptr i8, ptr [[PDST:%.*]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i16, ptr [[NEXT_GEP]], i64 16
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i16, ptr [[NEXT_GEP]], i64 32
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i16, ptr [[NEXT_GEP]], i64 48
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i64 32
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i64 64
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i64 96
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i16>, ptr [[NEXT_GEP]], align 2
; CHECK-NEXT: [[WIDE_LOAD9:%.*]] = load <16 x i16>, ptr [[TMP3]], align 2
; CHECK-NEXT: [[WIDE_LOAD10:%.*]] = load <16 x i16>, ptr [[TMP4]], align 2
@@ -47,9 +42,9 @@ define void @uaddsat(ptr nocapture readonly %pSrc, i16 signext %offset, ptr noca
; CHECK-NEXT: [[TMP7:%.*]] = call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> [[WIDE_LOAD9]], <16 x i16> [[BROADCAST_SPLAT]])
; CHECK-NEXT: [[TMP8:%.*]] = call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> [[WIDE_LOAD10]], <16 x i16> [[BROADCAST_SPLAT]])
; CHECK-NEXT: [[TMP9:%.*]] = call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> [[WIDE_LOAD11]], <16 x i16> [[BROADCAST_SPLAT]])
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i16, ptr [[NEXT_GEP5]], i64 16
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i16, ptr [[NEXT_GEP5]], i64 32
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i16, ptr [[NEXT_GEP5]], i64 48
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[NEXT_GEP5]], i64 32
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[NEXT_GEP5]], i64 64
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[NEXT_GEP5]], i64 96
; CHECK-NEXT: store <16 x i16> [[TMP6]], ptr [[NEXT_GEP5]], align 2
; CHECK-NEXT: store <16 x i16> [[TMP7]], ptr [[TMP10]], align 2
; CHECK-NEXT: store <16 x i16> [[TMP8]], ptr [[TMP11]], align 2
@@ -106,10 +101,10 @@ define void @uaddsat(ptr nocapture readonly %pSrc, i16 signext %offset, ptr noca
; CHECK-NEXT: [[BLKCNT_09:%.*]] = phi i32 [ [[DEC:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL19]], [[VEC_EPILOG_SCALAR_PH]] ]
; CHECK-NEXT: [[PSRC_ADDR_08:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL22]], [[VEC_EPILOG_SCALAR_PH]] ]
; CHECK-NEXT: [[PDST_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR3:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL25]], [[VEC_EPILOG_SCALAR_PH]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i16, ptr [[PSRC_ADDR_08]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRC_ADDR_08]], i64 2
; CHECK-NEXT: [[TMP22:%.*]] = load i16, ptr [[PSRC_ADDR_08]], align 2
; CHECK-NEXT: [[TMP23:%.*]] = tail call i16 @llvm.uadd.sat.i16(i16 [[TMP22]], i16 [[OFFSET]])
-; CHECK-NEXT: [[INCDEC_PTR3]] = getelementptr inbounds i16, ptr [[PDST_ADDR_07]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR3]] = getelementptr inbounds i8, ptr [[PDST_ADDR_07]], i64 2
; CHECK-NEXT: store i16 [[TMP23]], ptr [[PDST_ADDR_07]], align 2
; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_09]], -1
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0
@@ -139,12 +134,6 @@ while.end: ; preds = %while.body, %entry
}
; CHECK-COST-LABEL: fshl
-; CHECK-COST: Found an estimated cost of 2 for VF 1 For instruction: %1 = tail call i8 @llvm.fshl.i8(i8 %0, i8 %0, i8 %offset)
-; CHECK-COST: Found an estimated cost of 1 for VF 2 For instruction: %1 = tail call i8 @llvm.fshl.i8(i8 %0, i8 %0, i8 %offset)
-; CHECK-COST: Found an estimated cost of 1 for VF 4 For instruction: %1 = tail call i8 @llvm.fshl.i8(i8 %0, i8 %0, i8 %offset)
-; CHECK-COST: Found an estimated cost of 1 for VF 8 For instruction: %1 = tail call i8 @llvm.fshl.i8(i8 %0, i8 %0, i8 %offset)
-; CHECK-COST: Found an estimated cost of 1 for VF 16 For instruction: %1 = tail call i8 @llvm.fshl.i8(i8 %0, i8 %0, i8 %offset)
-; CHECK-COST: Found an estimated cost of 4 for VF 32 For instruction: %1 = tail call i8 @llvm.fshl.i8(i8 %0, i8 %0, i8 %offset)
define void @fshl(ptr nocapture readonly %pSrc, i8 signext %offset, ptr nocapture noalias %pDst, i32 %blockSize) #0 {
; CHECK-LABEL: @fshl(
diff --git a/llvm/test/Transforms/LoopVectorize/X86/invariant-store-vectorization.ll b/llvm/test/Transforms/LoopVectorize/X86/invariant-store-vectorization.ll
index 292ab4e4b2c4d0..4c7670c4dc045d 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/invariant-store-vectorization.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/invariant-store-vectorization.ll
@@ -36,18 +36,18 @@ define i32 @inv_val_store_to_inv_address_with_reduction(ptr %a, i64 %n, ptr %b)
; CHECK-NEXT: [[VEC_PHI5:%.*]] = phi <16 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI6:%.*]] = phi <16 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 16
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 32
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 48
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP1]], align 8, !alias.scope !0
-; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <16 x i32>, ptr [[TMP2]], align 8, !alias.scope !0
-; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <16 x i32>, ptr [[TMP3]], align 8, !alias.scope !0
-; CHECK-NEXT: [[WIDE_LOAD9:%.*]] = load <16 x i32>, ptr [[TMP4]], align 8, !alias.scope !0
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 64
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 128
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 192
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP1]], align 8, !alias.scope [[META0:![0-9]+]]
+; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <16 x i32>, ptr [[TMP2]], align 8, !alias.scope [[META0]]
+; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <16 x i32>, ptr [[TMP3]], align 8, !alias.scope [[META0]]
+; CHECK-NEXT: [[WIDE_LOAD9:%.*]] = load <16 x i32>, ptr [[TMP4]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[TMP5]] = add <16 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP6]] = add <16 x i32> [[VEC_PHI4]], [[WIDE_LOAD7]]
; CHECK-NEXT: [[TMP7]] = add <16 x i32> [[VEC_PHI5]], [[WIDE_LOAD8]]
; CHECK-NEXT: [[TMP8]] = add <16 x i32> [[VEC_PHI6]], [[WIDE_LOAD9]]
-; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4, !alias.scope !3, !noalias !0
+; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 64
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
@@ -72,9 +72,9 @@ define i32 @inv_val_store_to_inv_address_with_reduction(ptr %a, i64 %n, ptr %b)
; CHECK-NEXT: [[INDEX15:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT18:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI16:%.*]] = phi <8 x i32> [ [[TMP11]], [[VEC_EPILOG_PH]] ], [ [[TMP13:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX15]]
-; CHECK-NEXT: [[WIDE_LOAD17:%.*]] = load <8 x i32>, ptr [[TMP12]], align 8, !alias.scope !8
+; CHECK-NEXT: [[WIDE_LOAD17:%.*]] = load <8 x i32>, ptr [[TMP12]], align 8, !alias.scope [[META8:![0-9]+]]
; CHECK-NEXT: [[TMP13]] = add <8 x i32> [[VEC_PHI16]], [[WIDE_LOAD17]]
-; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4, !alias.scope !11, !noalias !8
+; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4, !alias.scope [[META11:![0-9]+]], !noalias [[META8]]
; CHECK-NEXT: [[INDEX_NEXT18]] = add nuw i64 [[INDEX15]], 8
; CHECK-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT18]], [[N_VEC13]]
; CHECK-NEXT: br i1 [[TMP14]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
@@ -153,10 +153,10 @@ define void @inv_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b, i3
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP1]], align 8, !alias.scope !15, !noalias !18
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP1]], align 8, !alias.scope [[META15:![0-9]+]], !noalias [[META18:![0-9]+]]
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <16 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: store <16 x i32> [[BROADCAST_SPLAT5]], ptr [[TMP1]], align 4, !alias.scope !15, !noalias !18
-; CHECK-NEXT: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> [[BROADCAST_SPLAT5]], <16 x ptr> [[BROADCAST_SPLAT7]], i32 4, <16 x i1> [[TMP2]]), !alias.scope !18
+; CHECK-NEXT: store <16 x i32> [[BROADCAST_SPLAT5]], ptr [[TMP1]], align 4, !alias.scope [[META15]], !noalias [[META18]]
+; CHECK-NEXT: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> [[BROADCAST_SPLAT5]], <16 x ptr> [[BROADCAST_SPLAT7]], i32 4, <16 x i1> [[TMP2]]), !alias.scope [[META18]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
@@ -180,10 +180,10 @@ define void @inv_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b, i3
; CHECK: vec.epilog.vector.body:
; CHECK-NEXT: [[INDEX11:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT19:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX11]]
-; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <8 x i32>, ptr [[TMP4]], align 8, !alias.scope !21, !noalias !24
+; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <8 x i32>, ptr [[TMP4]], align 8, !alias.scope [[META21:![0-9]+]], !noalias [[META24:![0-9]+]]
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq <8 x i32> [[WIDE_LOAD12]], [[BROADCAST_SPLAT14]]
-; CHECK-NEXT: store <8 x i32> [[BROADCAST_SPLAT16]], ptr [[TMP4]], align 4, !alias.scope !21, !noalias !24
-; CHECK-NEXT: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> [[BROADCAST_SPLAT16]], <8 x ptr> [[BROADCAST_SPLAT18]], i32 4, <8 x i1> [[TMP5]]), !alias.scope !24
+; CHECK-NEXT: store <8 x i32> [[BROADCAST_SPLAT16]], ptr [[TMP4]], align 4, !alias.scope [[META21]], !noalias [[META24]]
+; CHECK-NEXT: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> [[BROADCAST_SPLAT16]], <8 x ptr> [[BROADCAST_SPLAT18]], i32 4, <8 x i1> [[TMP5]]), !alias.scope [[META24]]
; CHECK-NEXT: [[INDEX_NEXT19]] = add nuw i64 [[INDEX11]], 8
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT19]], [[N_VEC9]]
; CHECK-NEXT: br i1 [[TMP6]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
@@ -275,12 +275,12 @@ define void @variant_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP1]], align 8, !alias.scope !28, !noalias !31
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i32>, ptr [[TMP1]], align 8, !alias.scope [[META28:![0-9]+]], !noalias [[META31:![0-9]+]]
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <16 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: store <16 x i32> [[BROADCAST_SPLAT13]], ptr [[TMP1]], align 4, !alias.scope !28, !noalias !31
+; CHECK-NEXT: store <16 x i32> [[BROADCAST_SPLAT13]], ptr [[TMP1]], align 4, !alias.scope [[META28]], !noalias [[META31]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i32, ptr [[C]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr [[TMP3]], i32 8, <16 x i1> [[TMP2]], <16 x i32> poison), !alias.scope !34
-; CHECK-NEXT: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> [[WIDE_MASKED_LOAD]], <16 x ptr> [[BROADCAST_SPLAT15]], i32 4, <16 x i1> [[TMP2]]), !alias.scope !35, !noalias !34
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <16 x i32> @llvm.masked.load.v16i32.p0(ptr [[TMP3]], i32 8, <16 x i1> [[TMP2]], <16 x i32> poison), !alias.scope [[META34:![0-9]+]]
+; CHECK-NEXT: call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> [[WIDE_MASKED_LOAD]], <16 x ptr> [[BROADCAST_SPLAT15]], i32 4, <16 x i1> [[TMP2]]), !alias.scope [[META35:![0-9]+]], !noalias [[META34]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP4]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
@@ -304,12 +304,12 @@ define void @variant_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b
; CHECK: vec.epilog.vector.body:
; CHECK-NEXT: [[INDEX19:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT28:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX19]]
-; CHECK-NEXT: [[WIDE_LOAD20:%.*]] = load <8 x i32>, ptr [[TMP5]], align 8, !alias.scope !37, !noalias !40
+; CHECK-NEXT: [[WIDE_LOAD20:%.*]] = load <8 x i32>, ptr [[TMP5]], align 8, !alias.scope [[META37:![0-9]+]], !noalias [[META40:![0-9]+]]
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq <8 x i32> [[WIDE_LOAD20]], [[BROADCAST_SPLAT22]]
-; CHECK-NEXT: store <8 x i32> [[BROADCAST_SPLAT24]], ptr [[TMP5]], align 4, !alias.scope !37, !noalias !40
+; CHECK-NEXT: store <8 x i32> [[BROADCAST_SPLAT24]], ptr [[TMP5]], align 4, !alias.scope [[META37]], !noalias [[META40]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[C]], i64 [[INDEX19]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD25:%.*]] = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr [[TMP7]], i32 8, <8 x i1> [[TMP6]], <8 x i32> poison), !alias.scope !43
-; CHECK-NEXT: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> [[WIDE_MASKED_LOAD25]], <8 x ptr> [[BROADCAST_SPLAT27]], i32 4, <8 x i1> [[TMP6]]), !alias.scope !44, !noalias !43
+; CHECK-NEXT: [[WIDE_MASKED_LOAD25:%.*]] = call <8 x i32> @llvm.masked.load.v8i32.p0(ptr [[TMP7]], i32 8, <8 x i1> [[TMP6]], <8 x i32> poison), !alias.scope [[META43:![0-9]+]]
+; CHECK-NEXT: call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> [[WIDE_MASKED_LOAD25]], <8 x ptr> [[BROADCAST_SPLAT27]], i32 4, <8 x i1> [[TMP6]]), !alias.scope [[META44:![0-9]+]], !noalias [[META43]]
; CHECK-NEXT: [[INDEX_NEXT28]] = add nuw i64 [[INDEX19]], 8
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT28]], [[N_VEC17]]
; CHECK-NEXT: br i1 [[TMP8]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP45:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll b/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
index a1dd890212b768..cc36c2ba6f7586 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
@@ -24,80 +24,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; O1-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O1-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O1-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O1-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O1-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O1-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O1-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O1-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O1-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O1-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O1-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O1-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O1-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O1-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O1-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O1-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O1-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O1-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O1-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O1-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O1-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O1-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O1-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O1-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O1-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O1-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O1-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O1-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O1-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O1-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O1-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O1-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O1-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O1-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O1-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O1-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O1-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O1-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O1-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O1-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O1-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O1-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O1-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O1-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O1-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O1-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O1-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O1-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O1-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O1-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O1-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O1-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O1-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O1-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O1-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O1-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O1-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O1-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O1-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O1-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O1-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O1-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O1-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O1-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O1-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O1-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O1-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O1-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O1-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O1-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O1-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O1-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O1-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O1-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O1-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O1-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O1-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O1-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O1-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O1-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O1-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O1-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O1-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O1-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O1-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O1-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O1-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O1-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O1-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O1-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O1-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O1-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O1-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O1-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O1-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O1-NEXT: ret i32 [[TMP46]]
@@ -109,80 +109,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; O2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O2-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O2-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O2-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O2-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O2-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O2-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O2-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O2-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O2-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O2-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O2-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O2-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O2-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O2-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O2-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O2-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O2-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O2-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O2-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O2-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O2-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O2-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O2-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O2-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O2-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O2-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O2-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O2-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O2-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O2-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O2-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O2-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O2-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O2-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O2-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O2-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O2-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O2-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O2-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O2-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O2-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O2-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O2-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O2-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O2-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O2-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O2-NEXT: ret i32 [[TMP46]]
@@ -194,80 +194,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; O3-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O3-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O3-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O3-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O3-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O3-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O3-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O3-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O3-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O3-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O3-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O3-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O3-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O3-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O3-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O3-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O3-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O3-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O3-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O3-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O3-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O3-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O3-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O3-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O3-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O3-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O3-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O3-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O3-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O3-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O3-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O3-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O3-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O3-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O3-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O3-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O3-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O3-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O3-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O3-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O3-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O3-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O3-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O3-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O3-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O3-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O3-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O3-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O3-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O3-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O3-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O3-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O3-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O3-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O3-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O3-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O3-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O3-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O3-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O3-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O3-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O3-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O3-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O3-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O3-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O3-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O3-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O3-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O3-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O3-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O3-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O3-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O3-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O3-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O3-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O3-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O3-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O3-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O3-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O3-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O3-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O3-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O3-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O3-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O3-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O3-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O3-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O3-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O3-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O3-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O3-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O3-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O3-NEXT: ret i32 [[TMP46]]
@@ -279,80 +279,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; O3DEFAULT-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O3DEFAULT-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O3DEFAULT-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O3DEFAULT-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O3DEFAULT-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O3DEFAULT-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O3DEFAULT-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O3DEFAULT-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O3DEFAULT-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O3DEFAULT-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O3DEFAULT-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O3DEFAULT-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O3DEFAULT-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O3DEFAULT-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O3DEFAULT-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O3DEFAULT-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O3DEFAULT-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O3DEFAULT-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O3DEFAULT-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O3DEFAULT-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O3DEFAULT-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O3DEFAULT-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O3DEFAULT-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O3DEFAULT-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O3DEFAULT-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O3DEFAULT-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O3DEFAULT-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O3DEFAULT-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O3DEFAULT-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O3DEFAULT-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O3DEFAULT-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O3DEFAULT-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O3DEFAULT-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O3DEFAULT-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O3DEFAULT-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O3DEFAULT-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O3DEFAULT-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O3DEFAULT-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O3DEFAULT-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O3DEFAULT-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O3DEFAULT-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O3DEFAULT-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O3DEFAULT-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O3DEFAULT-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O3DEFAULT-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O3DEFAULT-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O3DEFAULT-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O3DEFAULT-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O3DEFAULT-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O3DEFAULT-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O3DEFAULT-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O3DEFAULT-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O3DEFAULT-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O3DEFAULT-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O3DEFAULT-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O3DEFAULT-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O3DEFAULT-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O3DEFAULT-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O3DEFAULT-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O3DEFAULT-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O3DEFAULT-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O3DEFAULT-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O3DEFAULT-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O3DEFAULT-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O3DEFAULT-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O3DEFAULT-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O3DEFAULT-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O3DEFAULT-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O3DEFAULT-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O3DEFAULT-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O3DEFAULT-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O3DEFAULT-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O3DEFAULT-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O3DEFAULT-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O3DEFAULT-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O3DEFAULT-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O3DEFAULT-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O3DEFAULT-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O3DEFAULT-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O3DEFAULT-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O3DEFAULT-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O3DEFAULT-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O3DEFAULT-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O3DEFAULT-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O3DEFAULT-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O3DEFAULT-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O3DEFAULT-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O3DEFAULT-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O3DEFAULT-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O3DEFAULT-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O3DEFAULT-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O3DEFAULT-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O3DEFAULT-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O3DEFAULT-NEXT: ret i32 [[TMP46]]
@@ -364,80 +364,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; Os-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; Os-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; Os-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; Os-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; Os-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; Os-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; Os-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; Os-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; Os-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; Os-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; Os-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; Os-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; Os-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; Os-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; Os-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; Os-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; Os-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; Os-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; Os-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; Os-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; Os-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; Os-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; Os-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; Os-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; Os-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; Os-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; Os-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; Os-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; Os-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; Os-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; Os-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; Os-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; Os-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; Os-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; Os-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; Os-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; Os-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; Os-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; Os-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; Os-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; Os-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; Os-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; Os-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; Os-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; Os-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; Os-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; Os-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; Os-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; Os-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; Os-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; Os-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; Os-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; Os-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; Os-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; Os-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; Os-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; Os-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; Os-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; Os-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; Os-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; Os-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; Os-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; Os-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; Os-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; Os-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; Os-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; Os-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; Os-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; Os-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; Os-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; Os-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; Os-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; Os-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; Os-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; Os-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; Os-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; Os-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; Os-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; Os-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; Os-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; Os-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; Os-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; Os-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; Os-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; Os-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; Os-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; Os-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; Os-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; Os-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; Os-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; Os-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; Os-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; Os-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; Os-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; Os-NEXT: ret i32 [[TMP46]]
@@ -449,80 +449,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; Oz-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; Oz-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; Oz-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; Oz-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; Oz-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; Oz-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; Oz-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; Oz-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; Oz-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; Oz-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; Oz-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; Oz-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; Oz-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; Oz-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; Oz-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; Oz-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; Oz-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; Oz-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; Oz-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; Oz-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; Oz-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; Oz-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; Oz-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; Oz-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; Oz-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; Oz-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; Oz-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; Oz-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; Oz-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; Oz-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; Oz-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; Oz-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; Oz-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; Oz-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; Oz-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; Oz-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; Oz-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; Oz-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; Oz-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; Oz-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; Oz-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; Oz-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; Oz-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; Oz-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; Oz-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; Oz-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; Oz-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; Oz-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; Oz-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; Oz-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; Oz-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; Oz-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; Oz-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; Oz-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; Oz-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; Oz-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; Oz-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; Oz-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; Oz-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; Oz-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; Oz-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; Oz-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; Oz-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; Oz-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; Oz-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; Oz-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; Oz-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; Oz-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; Oz-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; Oz-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; Oz-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; Oz-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; Oz-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; Oz-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; Oz-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; Oz-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; Oz-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; Oz-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; Oz-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; Oz-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; Oz-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; Oz-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; Oz-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; Oz-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; Oz-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; Oz-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; Oz-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; Oz-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; Oz-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; Oz-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; Oz-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; Oz-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; Oz-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; Oz-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; Oz-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; Oz-NEXT: ret i32 [[TMP46]]
@@ -534,80 +534,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; O1VEC2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O1VEC2-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O1VEC2-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O1VEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O1VEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O1VEC2-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O1VEC2-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O1VEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O1VEC2-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O1VEC2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O1VEC2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O1VEC2-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O1VEC2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O1VEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O1VEC2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O1VEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O1VEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O1VEC2-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O1VEC2-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O1VEC2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O1VEC2-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O1VEC2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O1VEC2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O1VEC2-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O1VEC2-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O1VEC2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O1VEC2-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O1VEC2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O1VEC2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O1VEC2-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O1VEC2-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O1VEC2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O1VEC2-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O1VEC2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O1VEC2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O1VEC2-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O1VEC2-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O1VEC2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O1VEC2-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O1VEC2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O1VEC2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O1VEC2-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O1VEC2-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O1VEC2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O1VEC2-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O1VEC2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O1VEC2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O1VEC2-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O1VEC2-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O1VEC2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O1VEC2-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O1VEC2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O1VEC2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O1VEC2-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O1VEC2-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O1VEC2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O1VEC2-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O1VEC2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O1VEC2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O1VEC2-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O1VEC2-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O1VEC2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O1VEC2-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O1VEC2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O1VEC2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O1VEC2-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O1VEC2-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O1VEC2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O1VEC2-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O1VEC2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O1VEC2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O1VEC2-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O1VEC2-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O1VEC2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O1VEC2-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O1VEC2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O1VEC2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O1VEC2-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O1VEC2-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O1VEC2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O1VEC2-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O1VEC2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O1VEC2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O1VEC2-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O1VEC2-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O1VEC2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O1VEC2-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O1VEC2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O1VEC2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O1VEC2-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O1VEC2-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O1VEC2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O1VEC2-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O1VEC2-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O1VEC2-NEXT: ret i32 [[TMP46]]
@@ -619,80 +619,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; OzVEC2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; OzVEC2-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; OzVEC2-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; OzVEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; OzVEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; OzVEC2-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; OzVEC2-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; OzVEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; OzVEC2-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; OzVEC2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; OzVEC2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; OzVEC2-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; OzVEC2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; OzVEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; OzVEC2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; OzVEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; OzVEC2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; OzVEC2-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; OzVEC2-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; OzVEC2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; OzVEC2-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; OzVEC2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; OzVEC2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; OzVEC2-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; OzVEC2-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; OzVEC2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; OzVEC2-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; OzVEC2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; OzVEC2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; OzVEC2-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; OzVEC2-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; OzVEC2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; OzVEC2-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; OzVEC2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; OzVEC2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; OzVEC2-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; OzVEC2-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; OzVEC2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; OzVEC2-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; OzVEC2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; OzVEC2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; OzVEC2-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; OzVEC2-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; OzVEC2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; OzVEC2-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; OzVEC2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; OzVEC2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; OzVEC2-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; OzVEC2-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; OzVEC2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; OzVEC2-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; OzVEC2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; OzVEC2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; OzVEC2-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; OzVEC2-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; OzVEC2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; OzVEC2-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; OzVEC2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; OzVEC2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; OzVEC2-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; OzVEC2-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; OzVEC2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; OzVEC2-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; OzVEC2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; OzVEC2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; OzVEC2-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; OzVEC2-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; OzVEC2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; OzVEC2-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; OzVEC2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; OzVEC2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; OzVEC2-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; OzVEC2-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; OzVEC2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; OzVEC2-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; OzVEC2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; OzVEC2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; OzVEC2-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; OzVEC2-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; OzVEC2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; OzVEC2-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; OzVEC2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; OzVEC2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; OzVEC2-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; OzVEC2-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; OzVEC2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; OzVEC2-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; OzVEC2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; OzVEC2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; OzVEC2-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; OzVEC2-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; OzVEC2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; OzVEC2-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; OzVEC2-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; OzVEC2-NEXT: ret i32 [[TMP46]]
@@ -704,80 +704,80 @@ define i32 @enabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b,
; O3DIS-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O3DIS-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O3DIS-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O3DIS-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O3DIS-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O3DIS-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O3DIS-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O3DIS-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O3DIS-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O3DIS-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O3DIS-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O3DIS-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O3DIS-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O3DIS-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O3DIS-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O3DIS-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O3DIS-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O3DIS-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O3DIS-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O3DIS-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O3DIS-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O3DIS-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O3DIS-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O3DIS-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O3DIS-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O3DIS-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O3DIS-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O3DIS-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O3DIS-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O3DIS-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O3DIS-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O3DIS-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O3DIS-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O3DIS-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O3DIS-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O3DIS-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O3DIS-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O3DIS-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O3DIS-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O3DIS-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O3DIS-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O3DIS-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O3DIS-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O3DIS-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O3DIS-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O3DIS-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O3DIS-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O3DIS-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O3DIS-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O3DIS-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O3DIS-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O3DIS-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O3DIS-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O3DIS-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O3DIS-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O3DIS-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O3DIS-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O3DIS-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O3DIS-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O3DIS-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O3DIS-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O3DIS-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O3DIS-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O3DIS-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O3DIS-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O3DIS-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O3DIS-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O3DIS-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O3DIS-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O3DIS-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O3DIS-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O3DIS-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O3DIS-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O3DIS-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O3DIS-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O3DIS-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O3DIS-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O3DIS-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O3DIS-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O3DIS-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O3DIS-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O3DIS-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O3DIS-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O3DIS-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O3DIS-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O3DIS-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O3DIS-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O3DIS-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O3DIS-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O3DIS-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O3DIS-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O3DIS-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O3DIS-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O3DIS-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O3DIS-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O3DIS-NEXT: ret i32 [[TMP46]]
@@ -826,80 +826,80 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O2-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O2-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O2-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O2-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O2-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O2-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O2-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O2-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O2-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O2-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O2-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O2-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O2-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O2-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O2-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O2-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O2-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O2-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O2-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O2-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O2-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O2-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O2-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O2-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O2-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O2-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O2-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O2-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O2-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O2-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O2-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O2-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O2-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O2-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O2-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O2-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O2-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O2-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O2-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O2-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O2-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O2-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O2-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O2-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O2-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O2-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O2-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O2-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O2-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O2-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O2-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O2-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O2-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O2-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O2-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O2-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O2-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O2-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O2-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O2-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O2-NEXT: ret i32 [[TMP46]]
@@ -911,80 +911,80 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O3-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O3-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O3-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O3-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O3-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O3-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O3-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O3-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O3-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O3-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O3-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O3-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O3-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O3-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O3-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O3-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O3-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O3-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O3-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O3-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O3-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O3-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O3-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O3-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O3-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O3-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O3-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O3-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O3-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O3-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O3-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O3-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O3-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O3-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O3-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O3-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O3-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O3-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O3-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O3-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O3-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O3-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O3-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O3-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O3-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O3-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O3-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O3-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O3-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O3-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O3-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O3-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O3-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O3-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O3-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O3-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O3-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O3-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O3-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O3-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O3-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O3-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O3-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O3-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O3-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O3-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O3-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O3-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O3-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O3-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O3-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O3-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O3-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O3-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O3-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O3-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O3-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O3-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O3-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O3-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O3-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O3-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O3-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O3-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O3-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O3-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O3-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O3-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O3-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O3-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O3-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O3-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O3-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O3-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O3-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O3-NEXT: ret i32 [[TMP46]]
@@ -996,80 +996,80 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O3DEFAULT-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; O3DEFAULT-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; O3DEFAULT-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; O3DEFAULT-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; O3DEFAULT-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O3DEFAULT-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O3DEFAULT-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O3DEFAULT-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; O3DEFAULT-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; O3DEFAULT-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; O3DEFAULT-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; O3DEFAULT-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O3DEFAULT-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O3DEFAULT-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; O3DEFAULT-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; O3DEFAULT-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; O3DEFAULT-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; O3DEFAULT-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O3DEFAULT-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O3DEFAULT-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; O3DEFAULT-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; O3DEFAULT-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; O3DEFAULT-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; O3DEFAULT-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O3DEFAULT-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O3DEFAULT-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; O3DEFAULT-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; O3DEFAULT-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; O3DEFAULT-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; O3DEFAULT-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O3DEFAULT-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O3DEFAULT-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; O3DEFAULT-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; O3DEFAULT-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; O3DEFAULT-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; O3DEFAULT-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O3DEFAULT-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O3DEFAULT-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; O3DEFAULT-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; O3DEFAULT-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; O3DEFAULT-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; O3DEFAULT-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O3DEFAULT-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O3DEFAULT-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; O3DEFAULT-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; O3DEFAULT-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; O3DEFAULT-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; O3DEFAULT-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O3DEFAULT-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O3DEFAULT-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; O3DEFAULT-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; O3DEFAULT-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; O3DEFAULT-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; O3DEFAULT-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O3DEFAULT-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O3DEFAULT-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; O3DEFAULT-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; O3DEFAULT-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; O3DEFAULT-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; O3DEFAULT-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O3DEFAULT-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O3DEFAULT-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; O3DEFAULT-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; O3DEFAULT-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; O3DEFAULT-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; O3DEFAULT-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O3DEFAULT-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O3DEFAULT-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; O3DEFAULT-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; O3DEFAULT-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; O3DEFAULT-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; O3DEFAULT-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; O3DEFAULT-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; O3DEFAULT-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; O3DEFAULT-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; O3DEFAULT-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; O3DEFAULT-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; O3DEFAULT-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; O3DEFAULT-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; O3DEFAULT-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; O3DEFAULT-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; O3DEFAULT-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; O3DEFAULT-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; O3DEFAULT-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; O3DEFAULT-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; O3DEFAULT-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; O3DEFAULT-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; O3DEFAULT-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; O3DEFAULT-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; O3DEFAULT-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; O3DEFAULT-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; O3DEFAULT-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; O3DEFAULT-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; O3DEFAULT-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; O3DEFAULT-NEXT: ret i32 [[TMP46]]
@@ -1081,80 +1081,80 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; Os-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[B:%.*]], align 4
; Os-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; Os-NEXT: store <4 x i32> [[TMP0]], ptr [[A:%.*]], align 4
-; Os-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
+; Os-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
; Os-NEXT: [[WIDE_LOAD_1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; Os-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; Os-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; Os-NEXT: store <4 x i32> [[TMP2]], ptr [[TMP3]], align 4
-; Os-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
+; Os-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
; Os-NEXT: [[WIDE_LOAD_2:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; Os-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; Os-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; Os-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP6]], align 4
-; Os-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
+; Os-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
; Os-NEXT: [[WIDE_LOAD_3:%.*]] = load <4 x i32>, ptr [[TMP7]], align 4
; Os-NEXT: [[TMP8:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; Os-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; Os-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
-; Os-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
+; Os-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
; Os-NEXT: [[WIDE_LOAD_4:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; Os-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_4]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; Os-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; Os-NEXT: store <4 x i32> [[TMP11]], ptr [[TMP12]], align 4
-; Os-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
+; Os-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
; Os-NEXT: [[WIDE_LOAD_5:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
; Os-NEXT: [[TMP14:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_5]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; Os-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; Os-NEXT: store <4 x i32> [[TMP14]], ptr [[TMP15]], align 4
-; Os-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
+; Os-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
; Os-NEXT: [[WIDE_LOAD_6:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4
; Os-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_6]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; Os-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; Os-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP18]], align 4
-; Os-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
+; Os-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
; Os-NEXT: [[WIDE_LOAD_7:%.*]] = load <4 x i32>, ptr [[TMP19]], align 4
; Os-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_7]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; Os-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; Os-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP21]], align 4
-; Os-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
+; Os-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
; Os-NEXT: [[WIDE_LOAD_8:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4
; Os-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_8]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; Os-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; Os-NEXT: store <4 x i32> [[TMP23]], ptr [[TMP24]], align 4
-; Os-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
+; Os-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
; Os-NEXT: [[WIDE_LOAD_9:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4
; Os-NEXT: [[TMP26:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_9]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; Os-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; Os-NEXT: store <4 x i32> [[TMP26]], ptr [[TMP27]], align 4
-; Os-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
+; Os-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
; Os-NEXT: [[WIDE_LOAD_10:%.*]] = load <4 x i32>, ptr [[TMP28]], align 4
; Os-NEXT: [[TMP29:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_10]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; Os-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; Os-NEXT: store <4 x i32> [[TMP29]], ptr [[TMP30]], align 4
-; Os-NEXT: [[TMP31:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
+; Os-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
; Os-NEXT: [[WIDE_LOAD_11:%.*]] = load <4 x i32>, ptr [[TMP31]], align 4
; Os-NEXT: [[TMP32:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_11]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; Os-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; Os-NEXT: store <4 x i32> [[TMP32]], ptr [[TMP33]], align 4
-; Os-NEXT: [[TMP34:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 48
+; Os-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 192
; Os-NEXT: [[WIDE_LOAD_12:%.*]] = load <4 x i32>, ptr [[TMP34]], align 4
; Os-NEXT: [[TMP35:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_12]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 48
+; Os-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 192
; Os-NEXT: store <4 x i32> [[TMP35]], ptr [[TMP36]], align 4
-; Os-NEXT: [[TMP37:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 52
+; Os-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 208
; Os-NEXT: [[WIDE_LOAD_13:%.*]] = load <4 x i32>, ptr [[TMP37]], align 4
; Os-NEXT: [[TMP38:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_13]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP39:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 52
+; Os-NEXT: [[TMP39:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 208
; Os-NEXT: store <4 x i32> [[TMP38]], ptr [[TMP39]], align 4
-; Os-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 56
+; Os-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 224
; Os-NEXT: [[WIDE_LOAD_14:%.*]] = load <4 x i32>, ptr [[TMP40]], align 4
; Os-NEXT: [[TMP41:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_14]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP42:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 56
+; Os-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 224
; Os-NEXT: store <4 x i32> [[TMP41]], ptr [[TMP42]], align 4
-; Os-NEXT: [[TMP43:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 60
+; Os-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 240
; Os-NEXT: [[WIDE_LOAD_15:%.*]] = load <4 x i32>, ptr [[TMP43]], align 4
; Os-NEXT: [[TMP44:%.*]] = add nsw <4 x i32> [[WIDE_LOAD_15]], [[BROADCAST_SPLAT]]
-; Os-NEXT: [[TMP45:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 60
+; Os-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 240
; Os-NEXT: store <4 x i32> [[TMP44]], ptr [[TMP45]], align 4
; Os-NEXT: [[TMP46:%.*]] = load i32, ptr [[A]], align 4
; Os-NEXT: ret i32 [[TMP46]]
@@ -1349,58 +1349,58 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O3DEFAULT-NEXT: [[TMP2:%.*]] = shufflevector <4 x i32> [[TMP1]], <4 x i32> poison, <4 x i32> zeroinitializer
; O3DEFAULT-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[TMP0]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP3]], ptr [[A:%.*]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 4
-; O3DEFAULT-NEXT: [[ARRAYIDX2_4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 4
+; O3DEFAULT-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 16
+; O3DEFAULT-NEXT: [[ARRAYIDX2_4:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 16
; O3DEFAULT-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_4]], align 4
; O3DEFAULT-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[TMP4]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP5]], ptr [[ARRAYIDX2_4]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_8:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 8
-; O3DEFAULT-NEXT: [[ARRAYIDX2_8:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 8
+; O3DEFAULT-NEXT: [[ARRAYIDX_8:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 32
+; O3DEFAULT-NEXT: [[ARRAYIDX2_8:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 32
; O3DEFAULT-NEXT: [[TMP6:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_8]], align 4
; O3DEFAULT-NEXT: [[TMP7:%.*]] = add nsw <4 x i32> [[TMP6]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP7]], ptr [[ARRAYIDX2_8]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_12:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 12
-; O3DEFAULT-NEXT: [[ARRAYIDX2_12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 12
+; O3DEFAULT-NEXT: [[ARRAYIDX_12:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 48
+; O3DEFAULT-NEXT: [[ARRAYIDX2_12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 48
; O3DEFAULT-NEXT: [[TMP8:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_12]], align 4
; O3DEFAULT-NEXT: [[TMP9:%.*]] = add nsw <4 x i32> [[TMP8]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP9]], ptr [[ARRAYIDX2_12]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 16
-; O3DEFAULT-NEXT: [[ARRAYIDX2_16:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 16
+; O3DEFAULT-NEXT: [[ARRAYIDX_16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 64
+; O3DEFAULT-NEXT: [[ARRAYIDX2_16:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 64
; O3DEFAULT-NEXT: [[TMP10:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_16]], align 4
; O3DEFAULT-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[TMP10]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP11]], ptr [[ARRAYIDX2_16]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_20:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 20
-; O3DEFAULT-NEXT: [[ARRAYIDX2_20:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 20
+; O3DEFAULT-NEXT: [[ARRAYIDX_20:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 80
+; O3DEFAULT-NEXT: [[ARRAYIDX2_20:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 80
; O3DEFAULT-NEXT: [[TMP12:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_20]], align 4
; O3DEFAULT-NEXT: [[TMP13:%.*]] = add nsw <4 x i32> [[TMP12]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP13]], ptr [[ARRAYIDX2_20]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_24:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 24
-; O3DEFAULT-NEXT: [[ARRAYIDX2_24:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 24
+; O3DEFAULT-NEXT: [[ARRAYIDX_24:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 96
+; O3DEFAULT-NEXT: [[ARRAYIDX2_24:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 96
; O3DEFAULT-NEXT: [[TMP14:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_24]], align 4
; O3DEFAULT-NEXT: [[TMP15:%.*]] = add nsw <4 x i32> [[TMP14]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP15]], ptr [[ARRAYIDX2_24]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_28:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 28
-; O3DEFAULT-NEXT: [[ARRAYIDX2_28:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 28
+; O3DEFAULT-NEXT: [[ARRAYIDX_28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 112
+; O3DEFAULT-NEXT: [[ARRAYIDX2_28:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 112
; O3DEFAULT-NEXT: [[TMP16:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_28]], align 4
; O3DEFAULT-NEXT: [[TMP17:%.*]] = add nsw <4 x i32> [[TMP16]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP17]], ptr [[ARRAYIDX2_28]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_32:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 32
-; O3DEFAULT-NEXT: [[ARRAYIDX2_32:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 32
+; O3DEFAULT-NEXT: [[ARRAYIDX_32:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 128
+; O3DEFAULT-NEXT: [[ARRAYIDX2_32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 128
; O3DEFAULT-NEXT: [[TMP18:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_32]], align 4
; O3DEFAULT-NEXT: [[TMP19:%.*]] = add nsw <4 x i32> [[TMP18]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP19]], ptr [[ARRAYIDX2_32]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_36:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 36
-; O3DEFAULT-NEXT: [[ARRAYIDX2_36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 36
+; O3DEFAULT-NEXT: [[ARRAYIDX_36:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 144
+; O3DEFAULT-NEXT: [[ARRAYIDX2_36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 144
; O3DEFAULT-NEXT: [[TMP20:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_36]], align 4
; O3DEFAULT-NEXT: [[TMP21:%.*]] = add nsw <4 x i32> [[TMP20]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP21]], ptr [[ARRAYIDX2_36]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_40:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 40
-; O3DEFAULT-NEXT: [[ARRAYIDX2_40:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 40
+; O3DEFAULT-NEXT: [[ARRAYIDX_40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 160
+; O3DEFAULT-NEXT: [[ARRAYIDX2_40:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 160
; O3DEFAULT-NEXT: [[TMP22:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_40]], align 4
; O3DEFAULT-NEXT: [[TMP23:%.*]] = add nsw <4 x i32> [[TMP22]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP23]], ptr [[ARRAYIDX2_40]], align 4
-; O3DEFAULT-NEXT: [[ARRAYIDX_44:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 44
-; O3DEFAULT-NEXT: [[ARRAYIDX2_44:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 44
+; O3DEFAULT-NEXT: [[ARRAYIDX_44:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 176
+; O3DEFAULT-NEXT: [[ARRAYIDX2_44:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 176
; O3DEFAULT-NEXT: [[TMP24:%.*]] = load <4 x i32>, ptr [[ARRAYIDX_44]], align 4
; O3DEFAULT-NEXT: [[TMP25:%.*]] = add nsw <4 x i32> [[TMP24]], [[TMP2]]
; O3DEFAULT-NEXT: store <4 x i32> [[TMP25]], ptr [[ARRAYIDX2_44]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/X86/pr23997.ll b/llvm/test/Transforms/LoopVectorize/X86/pr23997.ll
index 4472ae4fecd384..0b16d80a4adbc5 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/pr23997.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/pr23997.ll
@@ -30,21 +30,21 @@ define void @foo(ptr addrspace(1) align 8 dereferenceable_or_null(16), ptr addrs
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[DOT12]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[TMP5]], i64 4
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[TMP5]], i64 8
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[TMP5]], i64 12
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP5]], align 8, !alias.scope !0
-; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP6]], align 8, !alias.scope !0
-; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP7]], align 8, !alias.scope !0
-; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP8]], align 8, !alias.scope !0
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP5]], i64 32
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP5]], i64 64
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP5]], i64 96
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP5]], align 8, !alias.scope [[META0:![0-9]+]]
+; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP6]], align 8, !alias.scope [[META0]]
+; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP7]], align 8, !alias.scope [[META0]]
+; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <4 x ptr addrspace(1)>, ptr addrspace(1) [[TMP8]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[DOT10]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[TMP9]], i64 4
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[TMP9]], i64 8
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds ptr addrspace(1), ptr addrspace(1) [[TMP9]], i64 12
-; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD]], ptr addrspace(1) [[TMP9]], align 8, !alias.scope !3, !noalias !0
-; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD3]], ptr addrspace(1) [[TMP10]], align 8, !alias.scope !3, !noalias !0
-; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD4]], ptr addrspace(1) [[TMP11]], align 8, !alias.scope !3, !noalias !0
-; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD5]], ptr addrspace(1) [[TMP12]], align 8, !alias.scope !3, !noalias !0
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP9]], i64 32
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP9]], i64 64
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[TMP9]], i64 96
+; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD]], ptr addrspace(1) [[TMP9]], align 8, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
+; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD3]], ptr addrspace(1) [[TMP10]], align 8, !alias.scope [[META3]], !noalias [[META0]]
+; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD4]], ptr addrspace(1) [[TMP11]], align 8, !alias.scope [[META3]], !noalias [[META0]]
+; CHECK-NEXT: store <4 x ptr addrspace(1)> [[WIDE_LOAD5]], ptr addrspace(1) [[TMP12]], align 8, !alias.scope [[META3]], !noalias [[META0]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; CHECK-NEXT: [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/small-size.ll b/llvm/test/Transforms/LoopVectorize/X86/small-size.ll
index fcd41fd188ce6f..be83329d30fef9 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/small-size.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/small-size.ll
@@ -369,11 +369,11 @@ define void @example23(ptr nocapture %src, ptr nocapture %dst) optsize {
; CHECK-NEXT: [[DOT04:%.*]] = phi ptr [ [[SRC:%.*]], [[TMP0:%.*]] ], [ [[TMP2:%.*]], [[TMP1]] ]
; CHECK-NEXT: [[DOT013:%.*]] = phi ptr [ [[DST:%.*]], [[TMP0]] ], [ [[TMP6:%.*]], [[TMP1]] ]
; CHECK-NEXT: [[I_02:%.*]] = phi i32 [ 0, [[TMP0]] ], [ [[TMP7:%.*]], [[TMP1]] ]
-; CHECK-NEXT: [[TMP2]] = getelementptr inbounds i16, ptr [[DOT04]], i64 1
+; CHECK-NEXT: [[TMP2]] = getelementptr inbounds i8, ptr [[DOT04]], i64 2
; CHECK-NEXT: [[TMP3:%.*]] = load i16, ptr [[DOT04]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = zext i16 [[TMP3]] to i32
; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i32 [[TMP4]], 7
-; CHECK-NEXT: [[TMP6]] = getelementptr inbounds i32, ptr [[DOT013]], i64 1
+; CHECK-NEXT: [[TMP6]] = getelementptr inbounds i8, ptr [[DOT013]], i64 4
; CHECK-NEXT: store i32 [[TMP5]], ptr [[DOT013]], align 4
; CHECK-NEXT: [[TMP7]] = add nuw nsw i32 [[I_02]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[TMP7]], 256
@@ -561,11 +561,11 @@ define i64 @example23d(ptr noalias nocapture %src, ptr noalias nocapture %dst) o
; CHECK-NEXT: [[DOT04:%.*]] = phi ptr [ [[SRC:%.*]], [[TMP0:%.*]] ], [ [[TMP2:%.*]], [[TMP1]] ]
; CHECK-NEXT: [[DOT013:%.*]] = phi ptr [ [[DST:%.*]], [[TMP0]] ], [ [[TMP6:%.*]], [[TMP1]] ]
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[TMP0]] ], [ [[TMP7:%.*]], [[TMP1]] ]
-; CHECK-NEXT: [[TMP2]] = getelementptr inbounds i16, ptr [[DOT04]], i64 1
+; CHECK-NEXT: [[TMP2]] = getelementptr inbounds i8, ptr [[DOT04]], i64 2
; CHECK-NEXT: [[TMP3:%.*]] = load i16, ptr [[DOT04]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = zext i16 [[TMP3]] to i32
; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i32 [[TMP4]], 7
-; CHECK-NEXT: [[TMP6]] = getelementptr inbounds i32, ptr [[DOT013]], i64 1
+; CHECK-NEXT: [[TMP6]] = getelementptr inbounds i8, ptr [[DOT013]], i64 4
; CHECK-NEXT: store i32 [[TMP5]], ptr [[DOT013]], align 4
; CHECK-NEXT: [[TMP7]] = add nuw nsw i64 [[I_02]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[TMP7]], 257
diff --git a/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll b/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll
index ffe4e13b187a21..fabe2eb8062bbe 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-store-accesses-with-gaps.ll
@@ -72,7 +72,7 @@ define dso_local void @test1(ptr noalias nocapture %points, ptr noalias nocaptur
;
; ENABLED_MASKED_STRIDED-LABEL: @test1(
; ENABLED_MASKED_STRIDED-NEXT: entry:
-; ENABLED_MASKED_STRIDED-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr i16, ptr [[POINTS:%.*]], i64 -1
+; ENABLED_MASKED_STRIDED-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr i8, ptr [[POINTS:%.*]], i64 -2
; ENABLED_MASKED_STRIDED-NEXT: br label [[VECTOR_BODY:%.*]]
; ENABLED_MASKED_STRIDED: vector.body:
; ENABLED_MASKED_STRIDED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -239,7 +239,7 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; ENABLED_MASKED_STRIDED-NEXT: [[TRIP_COUNT_MINUS_1:%.*]] = add nsw i64 [[WIDE_TRIP_COUNT]], -1
; ENABLED_MASKED_STRIDED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[TRIP_COUNT_MINUS_1]], i64 0
; ENABLED_MASKED_STRIDED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer
-; ENABLED_MASKED_STRIDED-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr i16, ptr [[POINTS:%.*]], i64 -1
+; ENABLED_MASKED_STRIDED-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr i8, ptr [[POINTS:%.*]], i64 -2
; ENABLED_MASKED_STRIDED-NEXT: br label [[VECTOR_BODY:%.*]]
; ENABLED_MASKED_STRIDED: vector.body:
; ENABLED_MASKED_STRIDED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
diff --git a/llvm/test/Transforms/LoopVectorize/consecutive-ptr-uniforms.ll b/llvm/test/Transforms/LoopVectorize/consecutive-ptr-uniforms.ll
index 5f2f83e19ab5be..c7eaac315fd861 100644
--- a/llvm/test/Transforms/LoopVectorize/consecutive-ptr-uniforms.ll
+++ b/llvm/test/Transforms/LoopVectorize/consecutive-ptr-uniforms.ll
@@ -49,7 +49,7 @@ for.end:
; CHECK: %offset.idx = sub i64 %n, %index
; CHECK-NOT: getelementptr
; CHECK: %[[G0:.+]] = getelementptr inbounds i32, ptr %a, i64 %offset.idx
-; CHECK: getelementptr inbounds i32, ptr %[[G0]], i64 -3
+; CHECK: getelementptr inbounds i8, ptr %[[G0]], i64 -12
; CHECK-NOT: getelementptr
; CHECK: br i1 {{.*}}, label %middle.block, label %vector.body
;
@@ -162,7 +162,7 @@ for.end:
; INTER: %offset.idx = sub i64 %n, %index
; INTER-NOT: getelementptr
; INTER: %[[G0:.+]] = getelementptr inbounds %pair, ptr %p, i64 %offset.idx, i32 0
-; INTER: getelementptr inbounds i32, ptr %[[G0]], i64 -6
+; INTER: getelementptr inbounds i8, ptr %[[G0]], i64 -24
; INTER-NOT: getelementptr
; INTER: br i1 {{.*}}, label %middle.block, label %vector.body
;
diff --git a/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll b/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll
index 6de93c381d0ed5..fe6d9b3ec690ec 100644
--- a/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll
+++ b/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll
@@ -38,7 +38,7 @@ define void @inv_store_last_lane(ptr noalias nocapture %a, ptr noalias nocapture
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[EXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: [[MUL_LCSSA:%.*]] = phi i32 [ [[MUL]], [[FOR_BODY]] ], [ [[TMP4]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, ptr [[INV:%.*]], i64 42
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i8, ptr [[INV:%.*]], i64 168
; CHECK-NEXT: store i32 [[MUL_LCSSA]], ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LoopVectorize/float-induction.ll b/llvm/test/Transforms/LoopVectorize/float-induction.ll
index 24c52b70495257..6fcb1913b9bd7a 100644
--- a/llvm/test/Transforms/LoopVectorize/float-induction.ll
+++ b/llvm/test/Transforms/LoopVectorize/float-induction.ll
@@ -102,7 +102,7 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fsub fast <4 x float> [[VEC_IND]], [[DOTSPLAT5]]
; VEC4_INTERL2-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[TMP4]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP4]], align 4
; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP5]], align 4
; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -349,7 +349,7 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fsub reassoc <4 x float> [[VEC_IND]], [[DOTSPLAT5]]
; VEC4_INTERL2-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[TMP4]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP4]], align 4
; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP5]], align 4
; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -586,7 +586,7 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fadd fast <4 x float> [[VEC_IND]], <float 2.000000e+00, float 2.000000e+00, float 2.000000e+00, float 2.000000e+00>
; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP2]], align 4
; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP3]], align 4
; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -859,7 +859,7 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC4_INTERL2-NEXT: [[VEC_IND10:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT13:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[STEP_ADD11:%.*]] = fadd fast <4 x float> [[VEC_IND10]], [[DOTSPLAT9]]
; VEC4_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[TMP6]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP6]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND10]], ptr [[TMP6]], align 4
; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD11]], ptr [[TMP7]], align 4
; VEC4_INTERL2-NEXT: [[TMP8:%.*]] = fadd fast <4 x float> [[VEC_IND10]], [[BROADCAST_SPLAT]]
@@ -869,11 +869,11 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC4_INTERL2-NEXT: [[TMP12:%.*]] = fadd fast <4 x float> [[TMP10]], [[TMP8]]
; VEC4_INTERL2-NEXT: [[TMP13:%.*]] = fadd fast <4 x float> [[TMP11]], [[TMP9]]
; VEC4_INTERL2-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = getelementptr inbounds float, ptr [[TMP14]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP14]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> [[TMP12]], ptr [[TMP14]], align 4
; VEC4_INTERL2-NEXT: store <4 x float> [[TMP13]], ptr [[TMP15]], align 4
; VEC4_INTERL2-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[C:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds float, ptr [[TMP16]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[TMP16]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> [[TMP10]], ptr [[TMP16]], align 4
; VEC4_INTERL2-NEXT: store <4 x float> [[TMP11]], ptr [[TMP17]], align 4
; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -1167,7 +1167,7 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ <float 1.000000e+00, float 1.500000e+00, float 2.000000e+00, float 2.500000e+00>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[STEP_ADD:%.*]] = fadd fast <4 x float> [[VEC_IND]], <float 2.000000e+00, float 2.000000e+00, float 2.000000e+00, float 2.000000e+00>
; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP2]], align 4
; VEC4_INTERL2-NEXT: store <4 x float> [[STEP_ADD]], ptr [[TMP3]], align 4
; VEC4_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -1403,7 +1403,7 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float
; VEC4_INTERL2-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 4
; VEC4_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 4
+; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; VEC4_INTERL2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP1]], align 4
; VEC4_INTERL2-NEXT: [[WIDE_LOAD3:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = fcmp fast oeq <4 x float> [[WIDE_LOAD]], zeroinitializer
diff --git a/llvm/test/Transforms/LoopVectorize/induction.ll b/llvm/test/Transforms/LoopVectorize/induction.ll
index d73193392e3937..16225b9d72a703 100644
--- a/llvm/test/Transforms/LoopVectorize/induction.ll
+++ b/llvm/test/Transforms/LoopVectorize/induction.ll
@@ -112,7 +112,7 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; UNROLL-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 190, i32 191>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; UNROLL-NEXT: [[STEP_ADD:%.*]] = add <2 x i32> [[VEC_IND]], <i32 2, i32 2>
; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 2
+; UNROLL-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i64 8
; UNROLL-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP3]], align 4
; UNROLL-NEXT: store <2 x i32> [[STEP_ADD]], ptr [[TMP4]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -205,7 +205,7 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; INTERLEAVE-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 190, i32 191, i32 192, i32 193>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4>
; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP3]], i64 4
+; INTERLEAVE-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP3]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[STEP_ADD]], ptr [[TMP4]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -410,12 +410,12 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; UNROLL-NEXT: [[TMP5:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[TMP5]], i64 [[OFFSET]]
-; UNROLL-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[TMP6]], i64 2
+; UNROLL-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP6]], i64 8
; UNROLL-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP6]], align 4, !alias.scope [[META4:![0-9]+]], !noalias [[META7:![0-9]+]]
; UNROLL-NEXT: [[WIDE_LOAD4:%.*]] = load <2 x float>, ptr [[TMP7]], align 4, !alias.scope [[META4]], !noalias [[META7]]
; UNROLL-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
; UNROLL-NEXT: [[TMP9:%.*]] = getelementptr float, ptr [[TMP8]], i64 [[OFFSET2]]
-; UNROLL-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP9]], i64 2
+; UNROLL-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP9]], i64 8
; UNROLL-NEXT: [[WIDE_LOAD5:%.*]] = load <2 x float>, ptr [[TMP9]], align 4, !alias.scope [[META7]]
; UNROLL-NEXT: [[WIDE_LOAD6:%.*]] = load <2 x float>, ptr [[TMP10]], align 4, !alias.scope [[META7]]
; UNROLL-NEXT: [[TMP11:%.*]] = fmul fast <2 x float> [[BROADCAST_SPLAT]], [[WIDE_LOAD5]]
@@ -553,12 +553,12 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[TMP5:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[TMP5]], i64 [[OFFSET]]
-; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[TMP6]], i64 4
+; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP6]], i64 16
; INTERLEAVE-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP6]], align 4, !alias.scope [[META4:![0-9]+]], !noalias [[META7:![0-9]+]]
; INTERLEAVE-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x float>, ptr [[TMP7]], align 4, !alias.scope [[META4]], !noalias [[META7]]
; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
; INTERLEAVE-NEXT: [[TMP9:%.*]] = getelementptr float, ptr [[TMP8]], i64 [[OFFSET2]]
-; INTERLEAVE-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP9]], i64 4
+; INTERLEAVE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP9]], i64 16
; INTERLEAVE-NEXT: [[WIDE_LOAD5:%.*]] = load <4 x float>, ptr [[TMP9]], align 4, !alias.scope [[META7]]
; INTERLEAVE-NEXT: [[WIDE_LOAD6:%.*]] = load <4 x float>, ptr [[TMP10]], align 4, !alias.scope [[META7]]
; INTERLEAVE-NEXT: [[TMP11:%.*]] = fmul fast <4 x float> [[BROADCAST_SPLAT]], [[WIDE_LOAD5]]
@@ -718,7 +718,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; UNROLL-NEXT: [[VEC_PHI:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; UNROLL-NEXT: [[VEC_PHI1:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
; UNROLL-NEXT: [[TMP0:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 2
+; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; UNROLL-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP0]], align 8
; UNROLL-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x i64>, ptr [[TMP1]], align 8
; UNROLL-NEXT: [[TMP2]] = add <2 x i64> [[WIDE_LOAD]], [[VEC_PHI]]
@@ -809,7 +809,7 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; INTERLEAVE-NEXT: [[VEC_PHI:%.*]] = phi <4 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[TMP0:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 4
+; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 32
; INTERLEAVE-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP0]], align 8
; INTERLEAVE-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i64>, ptr [[TMP1]], align 8
; INTERLEAVE-NEXT: [[TMP2]] = add <4 x i64> [[WIDE_LOAD]], [[VEC_PHI]]
@@ -2117,7 +2117,7 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; UNROLL-NEXT: [[TMP0:%.*]] = or disjoint i32 [[INDEX]], 2
; UNROLL-NEXT: [[TMP1:%.*]] = sext i32 [[INDEX]] to i64
; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]]
-; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 2
+; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 8
; UNROLL-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
; UNROLL-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x i32>, ptr [[TMP3]], align 4
; UNROLL-NEXT: br i1 [[C]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
@@ -2305,7 +2305,7 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; INTERLEAVE-NEXT: [[TMP0:%.*]] = or disjoint i32 [[INDEX]], 4
; INTERLEAVE-NEXT: [[TMP1:%.*]] = sext i32 [[INDEX]] to i64
; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]]
-; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 4
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16
; INTERLEAVE-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; INTERLEAVE-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
; INTERLEAVE-NEXT: br i1 [[C]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
@@ -3607,7 +3607,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]]
; UNROLL-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
-; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP11]], i64 2
+; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 8
; UNROLL-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP11]], align 4
; UNROLL-NEXT: store <2 x i32> [[STEP_ADD]], ptr [[TMP12]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -3751,7 +3751,7 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]]
; INTERLEAVE-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
-; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP11]], i64 4
+; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP11]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[STEP_ADD]], ptr [[TMP12]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -3995,7 +3995,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]]
; UNROLL-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
-; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP11]], i64 2
+; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 8
; UNROLL-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP11]], align 4
; UNROLL-NEXT: store <2 x i32> [[STEP_ADD]], ptr [[TMP12]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -4145,7 +4145,7 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[DOTCAST5]], [[T]]
; INTERLEAVE-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
-; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP11]], i64 4
+; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP11]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[STEP_ADD]], ptr [[TMP12]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -4293,7 +4293,7 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; UNROLL-NEXT: [[STEP_ADD:%.*]] = add <2 x i32> [[VEC_IND]], <i32 2, i32 2>
; UNROLL-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
-; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 2
+; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 8
; UNROLL-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP1]], align 4
; UNROLL-NEXT: store <2 x i32> [[STEP_ADD]], ptr [[TMP2]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -4370,7 +4370,7 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4>
; INTERLEAVE-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
-; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP1]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[STEP_ADD]], ptr [[TMP2]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -4511,7 +4511,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) {
; UNROLL-NEXT: [[SEXT:%.*]] = shl i64 [[OFFSET_IDX]], 32
; UNROLL-NEXT: [[TMP0:%.*]] = ashr exact i64 [[SEXT]], 32
; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
-; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 2
+; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 8
; UNROLL-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP1]], align 4
; UNROLL-NEXT: store <2 x i32> [[STEP_ADD]], ptr [[TMP2]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[OFFSET_IDX]], 4
@@ -4603,7 +4603,7 @@ define void @trunciv(ptr nocapture %a, i32 %start, i64 %k) {
; INTERLEAVE-NEXT: [[SEXT:%.*]] = shl i64 [[OFFSET_IDX]], 32
; INTERLEAVE-NEXT: [[TMP0:%.*]] = ashr exact i64 [[SEXT]], 32
; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
-; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP1]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[STEP_ADD]], ptr [[TMP2]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[OFFSET_IDX]], 8
@@ -4748,7 +4748,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[INDEX]], [[I]]
; UNROLL-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64
; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]]
-; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 2
+; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 8
; UNROLL-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP2]], align 4
; UNROLL-NEXT: store <2 x i32> [[STEP_ADD]], ptr [[TMP3]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -4837,7 +4837,7 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[INDEX]], [[I]]
; INTERLEAVE-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64
; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]]
-; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 4
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP2]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[STEP_ADD]], ptr [[TMP3]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -4970,7 +4970,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; UNROLL-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 0, i32 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; UNROLL-NEXT: [[STEP_ADD:%.*]] = add <2 x i32> [[VEC_IND]], <i32 4, i32 4>
; UNROLL-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 2
+; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; UNROLL-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP0]], align 4
; UNROLL-NEXT: store <2 x i32> [[STEP_ADD]], ptr [[TMP1]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -5057,7 +5057,7 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; INTERLEAVE-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 2, i32 4, i32 6>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], <i32 8, i32 8, i32 8, i32 8>
; INTERLEAVE-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 4
+; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP0]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[STEP_ADD]], ptr [[TMP1]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
@@ -6050,7 +6050,7 @@ define void @pr52460_first_order_recurrence_truncated_iv(ptr noalias %src, ptr %
; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[DST:%.*]], i64 [[TMP5]]
; UNROLL-NEXT: [[TMP7:%.*]] = add <2 x i32> [[VEC_IND]], [[TMP3]]
; UNROLL-NEXT: [[TMP8:%.*]] = add <2 x i32> [[STEP_ADD]], [[TMP4]]
-; UNROLL-NEXT: [[TMP9:%.*]] = getelementptr i32, ptr [[TMP6]], i64 2
+; UNROLL-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP6]], i64 8
; UNROLL-NEXT: store <2 x i32> [[TMP7]], ptr [[TMP6]], align 4
; UNROLL-NEXT: store <2 x i32> [[TMP8]], ptr [[TMP9]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[OFFSET_IDX]], 4
@@ -6145,7 +6145,7 @@ define void @pr52460_first_order_recurrence_truncated_iv(ptr noalias %src, ptr %
; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[DST:%.*]], i64 [[TMP5]]
; INTERLEAVE-NEXT: [[TMP7:%.*]] = add <4 x i32> [[VEC_IND]], [[TMP3]]
; INTERLEAVE-NEXT: [[TMP8:%.*]] = add <4 x i32> [[STEP_ADD]], [[TMP4]]
-; INTERLEAVE-NEXT: [[TMP9:%.*]] = getelementptr i32, ptr [[TMP6]], i64 4
+; INTERLEAVE-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP6]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[TMP7]], ptr [[TMP6]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[TMP8]], ptr [[TMP9]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[OFFSET_IDX]], 8
@@ -6396,7 +6396,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; UNROLL-NEXT: [[TMP17:%.*]] = shufflevector <2 x i32> [[VECTOR_RECUR]], <2 x i32> [[VEC_IND]], <2 x i32> <i32 1, i32 2>
; UNROLL-NEXT: [[TMP18:%.*]] = shufflevector <2 x i32> [[VEC_IND]], <2 x i32> [[STEP_ADD]], <2 x i32> <i32 1, i32 2>
; UNROLL-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 2
+; UNROLL-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP19]], i64 8
; UNROLL-NEXT: store <2 x i32> [[TMP17]], ptr [[TMP19]], align 4
; UNROLL-NEXT: store <2 x i32> [[TMP18]], ptr [[TMP20]], align 4
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -6554,7 +6554,7 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; INTERLEAVE-NEXT: [[TMP17:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[VEC_IND]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
; INTERLEAVE-NEXT: [[TMP18:%.*]] = shufflevector <4 x i32> [[VEC_IND]], <4 x i32> [[STEP_ADD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
; INTERLEAVE-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 4
+; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP19]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[TMP17]], ptr [[TMP19]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[TMP18]], ptr [[TMP20]], align 4
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
index 187eefbe9b5956..44b1e3eef37337 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
@@ -42,7 +42,7 @@ define void @test_array_load2_store2(i32 %C, i32 %D) {
; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[STRIDED_VEC]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP3:%.*]] = mul nsw <4 x i32> [[STRIDED_VEC1]], [[BROADCAST_SPLAT3]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1024 x i32], ptr @CD, i64 0, i64 [[TMP1]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i64 -1
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 -4
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> [[TMP3]], <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP5]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -349,7 +349,7 @@ define void @test_reversed_load2_store2(ptr noalias nocapture readonly %A, ptr n
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 1023, i32 1022, i32 1021, i32 1020>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = sub i64 1023, [[INDEX]]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ST2:%.*]], ptr [[A:%.*]], i64 [[OFFSET_IDX]], i32 0
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 -6
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 -24
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT: [[REVERSE:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
@@ -358,7 +358,7 @@ define void @test_reversed_load2_store2(ptr noalias nocapture readonly %A, ptr n
; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[REVERSE]], [[VEC_IND]]
; CHECK-NEXT: [[TMP3:%.*]] = sub nsw <4 x i32> [[REVERSE2]], [[VEC_IND]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_ST2]], ptr [[B:%.*]], i64 [[OFFSET_IDX]], i32 1
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i64 -7
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 -28
; CHECK-NEXT: [[REVERSE3:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
; CHECK-NEXT: [[REVERSE4:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i32> [[REVERSE3]], <4 x i32> [[REVERSE4]], <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
@@ -675,7 +675,7 @@ define void @mixed_load2_store2(ptr noalias nocapture readonly %A, ptr noalias n
; CHECK-NEXT: [[STRIDED_VEC4:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[STRIDED_VEC4]], [[STRIDED_VEC3]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i64 -1
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 -4
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> [[TMP3]], <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP5]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -1093,7 +1093,7 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP2]], i32 0
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP3]], i32 0
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP4]], i32 0
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 -1, i32 0
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 -8
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]], i32 1
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP2]], i32 1
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP3]], i32 1
@@ -1122,7 +1122,7 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]], i32 0
-; CHECK-NEXT: [[P_I_MINUS_1_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 -1, i32 0
+; CHECK-NEXT: [[P_I_MINUS_1_X:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 -8
; CHECK-NEXT: [[P_I_Y:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]], i32 1
; CHECK-NEXT: store i32 [[Z]], ptr [[P_I_X]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = load i32, ptr [[P_I_MINUS_1_X]], align 4
@@ -1300,7 +1300,7 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: store i32 [[X]], ptr [[TMP8]], align 4
; CHECK-NEXT: store i32 [[X]], ptr [[TMP9]], align 4
; CHECK-NEXT: store i32 [[X]], ptr [[TMP10]], align 4
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP11]], i64 -1
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 -4
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLAT]], <4 x i32> [[BROADCAST_SPLAT2]], <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
; CHECK-NEXT: store <8 x i32> [[INTERLEAVED_VEC]], ptr [[TMP12]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -1383,7 +1383,7 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[TMP12]], i64 9
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP12]], i64 36
; CHECK-NEXT: [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i64 0
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
; CHECK-NEXT: [[TMP16:%.*]] = extractelement <4 x i64> [[TMP7]], i64 1
@@ -1426,9 +1426,9 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[A_I:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I]]
; CHECK-NEXT: [[TMP31:%.*]] = getelementptr i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[A_I_MINUS_1:%.*]] = getelementptr i32, ptr [[TMP31]], i64 -1
+; CHECK-NEXT: [[A_I_MINUS_1:%.*]] = getelementptr i8, ptr [[TMP31]], i64 -4
; CHECK-NEXT: [[TMP32:%.*]] = getelementptr i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[A_I_MINUS_3:%.*]] = getelementptr i32, ptr [[TMP32]], i64 -3
+; CHECK-NEXT: [[A_I_MINUS_3:%.*]] = getelementptr i8, ptr [[TMP32]], i64 -12
; CHECK-NEXT: store i32 [[X]], ptr [[A_I_MINUS_1]], align 4
; CHECK-NEXT: store i32 [[Y]], ptr [[A_I_MINUS_3]], align 4
; CHECK-NEXT: store i32 [[Z]], ptr [[A_I]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll b/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll
index e4c183906865a7..306ec125dc2023 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll
@@ -16,9 +16,9 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[VEC_PHI2:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP9:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI3:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP11:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 8
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 12
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 32
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 48
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/runtime-check.ll b/llvm/test/Transforms/LoopVectorize/runtime-check.ll
index 108580556f532a..d5df8afc80a792 100644
--- a/llvm/test/Transforms/LoopVectorize/runtime-check.ll
+++ b/llvm/test/Transforms/LoopVectorize/runtime-check.ll
@@ -235,7 +235,7 @@ define void @test_runtime_check2(ptr %a, float %b, i64 %offset, i64 %offset2, i6
; CHECK-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]]
; CHECK-NEXT: store float [[AD]], ptr [[ARR_IDX]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr float, ptr [[C:%.*]], i64 [[IV]]
-; CHECK-NEXT: [[C_IDX:%.*]] = getelementptr float, ptr [[TMP2]], i64 -1
+; CHECK-NEXT: [[C_IDX:%.*]] = getelementptr i8, ptr [[TMP2]], i64 -4
; CHECK-NEXT: [[LC:%.*]] = load float, ptr [[C_IDX]], align 4
; CHECK-NEXT: [[VC:%.*]] = fadd float [[LC]], 1.000000e+00
; CHECK-NEXT: [[C_IDX2:%.*]] = getelementptr inbounds float, ptr [[C]], i64 [[IV]]
diff --git a/llvm/test/Transforms/LoopVectorize/scalar_after_vectorization.ll b/llvm/test/Transforms/LoopVectorize/scalar_after_vectorization.ll
index 4a6523f2dabf55..f791b956bcf54f 100644
--- a/llvm/test/Transforms/LoopVectorize/scalar_after_vectorization.ll
+++ b/llvm/test/Transforms/LoopVectorize/scalar_after_vectorization.ll
@@ -11,7 +11,7 @@ target datalayout = "e-m:e-i64:64-i128:128-n32:64-S128"
; CHECK: %[[T2:.+]] = add nuw nsw i64 %offset.idx, %tmp0
; CHECK: %[[T3:.+]] = sub nsw i64 %[[T2]], %x
; CHECK: %[[T4:.+]] = getelementptr inbounds i32, ptr %a, i64 %[[T3]]
-; CHECK: %[[T6:.+]] = getelementptr inbounds i32, ptr %[[T4]], i64 4
+; CHECK: %[[T6:.+]] = getelementptr inbounds i8, ptr %[[T4]], i64 16
; CHECK: load <4 x i32>, ptr %[[T4]], align 4
; CHECK: load <4 x i32>, ptr %[[T6]], align 4
; CHECK: br {{.*}}, label %middle.block, label %vector.body
diff --git a/llvm/test/Transforms/LoopVectorize/vector-geps.ll b/llvm/test/Transforms/LoopVectorize/vector-geps.ll
index 4f4b4f6600da4f..bceb9f78b0e84f 100644
--- a/llvm/test/Transforms/LoopVectorize/vector-geps.ll
+++ b/llvm/test/Transforms/LoopVectorize/vector-geps.ll
@@ -21,8 +21,8 @@ define void @vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: store <4 x ptr> [[TMP0]], ptr [[TMP1]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
-; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
@@ -36,7 +36,7 @@ define void @vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: store ptr [[VAR0]], ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP2:![0-9]+]]
+; CHECK-NEXT: br i1 [[COND]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
@@ -68,14 +68,14 @@ define void @uniform_vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 4
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <4 x ptr> poison, ptr [[TMP0]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <4 x ptr> [[DOTSPLATINSERT]], <4 x ptr> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds ptr, ptr [[A:%.*]], i64 [[INDEX]]
; CHECK-NEXT: store <4 x ptr> [[DOTSPLAT]], ptr [[TMP1]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
@@ -84,7 +84,7 @@ define void @uniform_vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 1
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 4
; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds ptr, ptr [[A]], i64 [[I]]
; CHECK-NEXT: store ptr [[VAR0]], ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
diff --git a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-dominance.ll b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-dominance.ll
index 2b5ee9a76047f7..2af9909ee396b1 100644
--- a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-dominance.ll
+++ b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-dominance.ll
@@ -27,45 +27,45 @@ define void @multiply_can_hoist_cast(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <1 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[TMP4:%.*]] = fmul contract <1 x double> [[COL_LOAD]], [[COL_LOAD1]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <1 x double>, ptr [[TMP5]], align 8
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
; CHECK-NEXT: [[COL_LOAD3:%.*]] = load <1 x double>, ptr [[TMP6]], align 8
; CHECK-NEXT: [[TMP7:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD2]], <1 x double> [[COL_LOAD3]], <1 x double> [[TMP4]])
; CHECK-NEXT: store <1 x double> [[TMP7]], ptr [[C]], align 8
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr double, ptr [[A]], i64 1
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <1 x double>, ptr [[TMP8]], align 8
; CHECK-NEXT: [[COL_LOAD9:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = fmul contract <1 x double> [[COL_LOAD8]], [[COL_LOAD9]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: [[COL_LOAD13:%.*]] = load <1 x double>, ptr [[TMP10]], align 8
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
; CHECK-NEXT: [[COL_LOAD14:%.*]] = load <1 x double>, ptr [[TMP11]], align 8
; CHECK-NEXT: [[TMP12:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD13]], <1 x double> [[COL_LOAD14]], <1 x double> [[TMP9]])
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr double, ptr [[C]], i64 1
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[C]], i64 8
; CHECK-NEXT: store <1 x double> [[TMP12]], ptr [[TMP13]], align 8
; CHECK-NEXT: [[COL_LOAD19:%.*]] = load <1 x double>, ptr [[A]], align 8
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD20:%.*]] = load <1 x double>, ptr [[TMP14]], align 8
; CHECK-NEXT: [[TMP15:%.*]] = fmul contract <1 x double> [[COL_LOAD19]], [[COL_LOAD20]]
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD24:%.*]] = load <1 x double>, ptr [[TMP16]], align 8
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD25:%.*]] = load <1 x double>, ptr [[TMP17]], align 8
; CHECK-NEXT: [[TMP18:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD24]], <1 x double> [[COL_LOAD25]], <1 x double> [[TMP15]])
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <1 x double> [[TMP18]], ptr [[TMP19]], align 8
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr double, ptr [[A]], i64 1
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[COL_LOAD30:%.*]] = load <1 x double>, ptr [[TMP20]], align 8
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD31:%.*]] = load <1 x double>, ptr [[TMP21]], align 8
; CHECK-NEXT: [[TMP22:%.*]] = fmul contract <1 x double> [[COL_LOAD30]], [[COL_LOAD31]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: [[COL_LOAD35:%.*]] = load <1 x double>, ptr [[TMP23]], align 8
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD36:%.*]] = load <1 x double>, ptr [[TMP24]], align 8
; CHECK-NEXT: [[TMP25:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD35]], <1 x double> [[COL_LOAD36]], <1 x double> [[TMP22]])
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr double, ptr [[C]], i64 3
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[C]], i64 24
; CHECK-NEXT: store <1 x double> [[TMP25]], ptr [[TMP26]], align 8
; CHECK-NEXT: ret void
;
@@ -80,7 +80,7 @@ entry:
define void @multiply_can_hoist_multiple_insts(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-LABEL: @multiply_can_hoist_multiple_insts(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr [4 x double], ptr [[C:%.*]], i64 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 64
; CHECK-NEXT: [[STORE_BEGIN:%.*]] = ptrtoint ptr [[GEP]] to i64
; CHECK-NEXT: [[STORE_END:%.*]] = add nuw nsw i64 [[STORE_BEGIN]], 32
; CHECK-NEXT: [[LOAD_BEGIN:%.*]] = ptrtoint ptr [[B:%.*]] to i64
@@ -99,45 +99,45 @@ define void @multiply_can_hoist_multiple_insts(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <1 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[TMP4:%.*]] = fmul contract <1 x double> [[COL_LOAD]], [[COL_LOAD1]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <1 x double>, ptr [[TMP5]], align 8
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
; CHECK-NEXT: [[COL_LOAD3:%.*]] = load <1 x double>, ptr [[TMP6]], align 8
; CHECK-NEXT: [[TMP7:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD2]], <1 x double> [[COL_LOAD3]], <1 x double> [[TMP4]])
; CHECK-NEXT: store <1 x double> [[TMP7]], ptr [[GEP]], align 8
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr double, ptr [[A]], i64 1
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <1 x double>, ptr [[TMP8]], align 8
; CHECK-NEXT: [[COL_LOAD9:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = fmul contract <1 x double> [[COL_LOAD8]], [[COL_LOAD9]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: [[COL_LOAD13:%.*]] = load <1 x double>, ptr [[TMP10]], align 8
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
; CHECK-NEXT: [[COL_LOAD14:%.*]] = load <1 x double>, ptr [[TMP11]], align 8
; CHECK-NEXT: [[TMP12:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD13]], <1 x double> [[COL_LOAD14]], <1 x double> [[TMP9]])
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr [4 x double], ptr [[C]], i64 2, i64 1
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[C]], i64 72
; CHECK-NEXT: store <1 x double> [[TMP12]], ptr [[TMP13]], align 8
; CHECK-NEXT: [[COL_LOAD19:%.*]] = load <1 x double>, ptr [[A]], align 8
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD20:%.*]] = load <1 x double>, ptr [[TMP14]], align 8
; CHECK-NEXT: [[TMP15:%.*]] = fmul contract <1 x double> [[COL_LOAD19]], [[COL_LOAD20]]
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD24:%.*]] = load <1 x double>, ptr [[TMP16]], align 8
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD25:%.*]] = load <1 x double>, ptr [[TMP17]], align 8
; CHECK-NEXT: [[TMP18:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD24]], <1 x double> [[COL_LOAD25]], <1 x double> [[TMP15]])
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr [4 x double], ptr [[C]], i64 2, i64 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[C]], i64 80
; CHECK-NEXT: store <1 x double> [[TMP18]], ptr [[TMP19]], align 8
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr double, ptr [[A]], i64 1
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[COL_LOAD30:%.*]] = load <1 x double>, ptr [[TMP20]], align 8
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD31:%.*]] = load <1 x double>, ptr [[TMP21]], align 8
; CHECK-NEXT: [[TMP22:%.*]] = fmul contract <1 x double> [[COL_LOAD30]], [[COL_LOAD31]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: [[COL_LOAD35:%.*]] = load <1 x double>, ptr [[TMP23]], align 8
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD36:%.*]] = load <1 x double>, ptr [[TMP24]], align 8
; CHECK-NEXT: [[TMP25:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD35]], <1 x double> [[COL_LOAD36]], <1 x double> [[TMP22]])
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr [4 x double], ptr [[C]], i64 2, i64 3
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[C]], i64 88
; CHECK-NEXT: store <1 x double> [[TMP25]], ptr [[TMP26]], align 8
; CHECK-NEXT: ret void
;
@@ -154,7 +154,7 @@ entry:
define void @multiply_can_hoist_multiple_insts2(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-LABEL: @multiply_can_hoist_multiple_insts2(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[GEP_1:%.*]] = getelementptr <4 x double>, ptr [[C:%.*]], i64 42
+; CHECK-NEXT: [[GEP_1:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 1344
; CHECK-NEXT: [[STORE_BEGIN:%.*]] = ptrtoint ptr [[GEP_1]] to i64
; CHECK-NEXT: [[STORE_END:%.*]] = add nuw nsw i64 [[STORE_BEGIN]], 32
; CHECK-NEXT: [[LOAD_BEGIN:%.*]] = ptrtoint ptr [[B:%.*]] to i64
@@ -173,42 +173,42 @@ define void @multiply_can_hoist_multiple_insts2(ptr noalias %A, ptr %B, ptr %C)
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <1 x double>, ptr [[A:%.*]], align 8
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[TMP4:%.*]] = fmul contract <1 x double> [[COL_LOAD]], [[COL_LOAD1]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <1 x double>, ptr [[TMP5]], align 8
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
; CHECK-NEXT: [[COL_LOAD3:%.*]] = load <1 x double>, ptr [[TMP6]], align 8
; CHECK-NEXT: [[TMP7:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD2]], <1 x double> [[COL_LOAD3]], <1 x double> [[TMP4]])
; CHECK-NEXT: store <1 x double> [[TMP7]], ptr [[GEP_1]], align 8
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr double, ptr [[A]], i64 1
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <1 x double>, ptr [[TMP8]], align 8
; CHECK-NEXT: [[COL_LOAD9:%.*]] = load <1 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = fmul contract <1 x double> [[COL_LOAD8]], [[COL_LOAD9]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: [[COL_LOAD13:%.*]] = load <1 x double>, ptr [[TMP10]], align 8
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr double, ptr [[TMP3]], i64 1
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
; CHECK-NEXT: [[COL_LOAD14:%.*]] = load <1 x double>, ptr [[TMP11]], align 8
; CHECK-NEXT: [[TMP12:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD13]], <1 x double> [[COL_LOAD14]], <1 x double> [[TMP9]])
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[C]], i64 1352
; CHECK-NEXT: store <1 x double> [[TMP12]], ptr [[TMP13]], align 8
; CHECK-NEXT: [[COL_LOAD19:%.*]] = load <1 x double>, ptr [[A]], align 8
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD20:%.*]] = load <1 x double>, ptr [[TMP14]], align 8
; CHECK-NEXT: [[TMP15:%.*]] = fmul contract <1 x double> [[COL_LOAD19]], [[COL_LOAD20]]
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD24:%.*]] = load <1 x double>, ptr [[TMP16]], align 8
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD25:%.*]] = load <1 x double>, ptr [[TMP17]], align 8
; CHECK-NEXT: [[TMP18:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD24]], <1 x double> [[COL_LOAD25]], <1 x double> [[TMP15]])
; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[C]], i64 1360
; CHECK-NEXT: store <1 x double> [[TMP18]], ptr [[TMP19]], align 8
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr double, ptr [[A]], i64 1
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[COL_LOAD30:%.*]] = load <1 x double>, ptr [[TMP20]], align 8
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD31:%.*]] = load <1 x double>, ptr [[TMP21]], align 8
; CHECK-NEXT: [[TMP22:%.*]] = fmul contract <1 x double> [[COL_LOAD30]], [[COL_LOAD31]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: [[COL_LOAD35:%.*]] = load <1 x double>, ptr [[TMP23]], align 8
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD36:%.*]] = load <1 x double>, ptr [[TMP24]], align 8
; CHECK-NEXT: [[TMP25:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD35]], <1 x double> [[COL_LOAD36]], <1 x double> [[TMP22]])
; CHECK-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[C]], i64 1368
@@ -232,9 +232,9 @@ define void @multiply_dont_hoist_phi(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[NEXT:%.*]]
; CHECK: next:
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[A:%.*]], i64 2
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 16
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
-; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B:%.*]], i64 2
+; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 16
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT13:%.*]] = shufflevector <2 x double> [[COL_LOAD4]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A]], align 8
@@ -246,7 +246,7 @@ define void @multiply_dont_hoist_phi(ptr noalias %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP2:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
; CHECK-NEXT: [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT7]], <2 x double> [[TMP2]])
-; CHECK-NEXT: [[GEP_1:%.*]] = getelementptr <4 x double>, ptr [[C:%.*]], i64 26
+; CHECK-NEXT: [[GEP_1:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 832
; CHECK-NEXT: store <2 x double> [[TMP3]], ptr [[GEP_1]], align 8
; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr i8, ptr [[C]], i64 848
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[VEC_GEP14]], align 8
@@ -273,10 +273,10 @@ define void @multiply_dont_hoist_cast_due_to_operand(ptr noalias %A, ptr %B, ptr
; CHECK-LABEL: @multiply_dont_hoist_cast_due_to_operand(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B]], i64 2
+; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[B]], i64 16
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -288,7 +288,7 @@ define void @multiply_dont_hoist_cast_due_to_operand(ptr noalias %A, ptr %B, ptr
; CHECK-NEXT: [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP2]])
; CHECK-NEXT: [[C:%.*]] = load ptr, ptr [[C_PTR:%.*]], align 8
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[C]], align 8
-; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP3]], ptr [[VEC_GEP14]], align 8
; CHECK-NEXT: ret void
;
@@ -306,10 +306,10 @@ define void @multiply_dont_hoist_load(ptr noalias %A, ptr %B, ptr %C.ptr) {
; CHECK-LABEL: @multiply_dont_hoist_load(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B]], i64 2
+; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[B]], i64 16
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -321,7 +321,7 @@ define void @multiply_dont_hoist_load(ptr noalias %A, ptr %B, ptr %C.ptr) {
; CHECK-NEXT: [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP2]])
; CHECK-NEXT: [[C:%.*]] = load ptr, ptr [[C_PTR:%.*]], align 8
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[C]], align 8
-; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP3]], ptr [[VEC_GEP14]], align 8
; CHECK-NEXT: ret void
;
@@ -339,10 +339,10 @@ define void @multiply_dont_hoist_call(ptr noalias %A, ptr %B) {
; CHECK-LABEL: @multiply_dont_hoist_call(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr double, ptr [[B]], i64 2
+; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[B]], i64 16
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -354,7 +354,7 @@ define void @multiply_dont_hoist_call(ptr noalias %A, ptr %B) {
; CHECK-NEXT: [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP2]])
; CHECK-NEXT: [[C:%.*]] = call ptr @get_address()
; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[C]], align 8
-; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[VEC_GEP14:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP3]], ptr [[VEC_GEP14]], align 8
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll
index e0dbb21cac08b4..1cc886e68d8e61 100644
--- a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll
+++ b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll
@@ -30,13 +30,13 @@ define void @multiply_noalias_4x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[A:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr double, ptr [[TMP1]], i64 [[ROWS_IV]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[TMP2]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[TMP2]], i64 4
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 32
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = shl i64 [[COLS_IV]], 2
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr double, ptr [[B:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr double, ptr [[TMP4]], i64 [[INNER_IV]]
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[TMP5]], align 8
-; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr double, ptr [[TMP5]], i64 4
+; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[TMP5]], i64 32
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD]], <2 x double> [[SPLAT_SPLAT]], <2 x double> [[RESULT_VEC_0]])
@@ -58,7 +58,7 @@ define void @multiply_noalias_4x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr double, ptr [[C:%.*]], i64 [[TMP10]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr double, ptr [[TMP11]], i64 [[ROWS_IV]]
; CHECK-NEXT: store <2 x double> [[TMP7]], ptr [[TMP12]], align 8
-; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr double, ptr [[TMP12]], i64 4
+; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i8, ptr [[TMP12]], i64 32
; CHECK-NEXT: store <2 x double> [[TMP9]], ptr [[VEC_GEP16]], align 8
; CHECK-NEXT: br i1 [[ROWS_COND_NOT]], label [[COLS_LATCH]], label [[ROWS_HEADER]]
; CHECK: cols.latch:
@@ -106,13 +106,13 @@ define void @multiply_noalias_2x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i64, ptr [[A:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i64, ptr [[TMP1]], i64 [[ROWS_IV]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i64>, ptr [[TMP2]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i64, ptr [[TMP2]], i64 2
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 16
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i64>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = shl i64 [[COLS_IV]], 2
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i64, ptr [[B:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i64, ptr [[TMP4]], i64 [[INNER_IV]]
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
-; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i64, ptr [[TMP5]], i64 4
+; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[TMP5]], i64 32
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x i64>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x i64> [[COL_LOAD2]], <2 x i64> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = mul <2 x i64> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -138,7 +138,7 @@ define void @multiply_noalias_2x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i64, ptr [[C:%.*]], i64 [[TMP14]]
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i64, ptr [[TMP15]], i64 [[ROWS_IV]]
; CHECK-NEXT: store <2 x i64> [[TMP9]], ptr [[TMP16]], align 8
-; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i64, ptr [[TMP16]], i64 2
+; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i8, ptr [[TMP16]], i64 16
; CHECK-NEXT: store <2 x i64> [[TMP13]], ptr [[VEC_GEP16]], align 8
; CHECK-NEXT: br i1 [[ROWS_COND_NOT]], label [[COLS_LATCH]], label [[ROWS_HEADER]]
; CHECK: cols.latch:
@@ -192,13 +192,13 @@ define void @multiply_noalias_4x2_2x8(ptr noalias %A, ptr noalias %B, ptr noalia
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i64, ptr [[A:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i64, ptr [[TMP1]], i64 [[ROWS_IV]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i64>, ptr [[TMP2]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i64, ptr [[TMP2]], i64 4
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 32
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i64>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = shl i64 [[COLS_IV]], 1
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i64, ptr [[B:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i64, ptr [[TMP4]], i64 [[INNER_IV]]
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
-; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i64, ptr [[TMP5]], i64 2
+; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[TMP5]], i64 16
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x i64>, ptr [[VEC_GEP3]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x i64> [[COL_LOAD2]], <2 x i64> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = mul <2 x i64> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -224,7 +224,7 @@ define void @multiply_noalias_4x2_2x8(ptr noalias %A, ptr noalias %B, ptr noalia
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i64, ptr [[C:%.*]], i64 [[TMP14]]
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i64, ptr [[TMP15]], i64 [[ROWS_IV]]
; CHECK-NEXT: store <2 x i64> [[TMP9]], ptr [[TMP16]], align 8
-; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i64, ptr [[TMP16]], i64 4
+; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i8, ptr [[TMP16]], i64 32
; CHECK-NEXT: store <2 x i64> [[TMP13]], ptr [[VEC_GEP16]], align 8
; CHECK-NEXT: br i1 [[ROWS_COND_NOT]], label [[COLS_LATCH]], label [[ROWS_HEADER]]
; CHECK: cols.latch:
@@ -309,13 +309,13 @@ define void @multiply_alias_2x2(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP9:%.*]] = add i64 [[TMP8]], [[ROWS_IV]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr float, ptr [[TMP3]], i64 [[TMP9]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x float>, ptr [[TMP10]], align 4
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr float, ptr [[TMP10]], i64 2
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP10]], i64 8
; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <2 x float>, ptr [[VEC_GEP]], align 4
; CHECK-NEXT: [[TMP11:%.*]] = shl i64 [[COLS_IV]], 1
; CHECK-NEXT: [[TMP12:%.*]] = add i64 [[TMP11]], [[INNER_IV]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr float, ptr [[TMP7]], i64 [[TMP12]]
; CHECK-NEXT: [[COL_LOAD9:%.*]] = load <2 x float>, ptr [[TMP13]], align 4
-; CHECK-NEXT: [[VEC_GEP10:%.*]] = getelementptr float, ptr [[TMP13]], i64 2
+; CHECK-NEXT: [[VEC_GEP10:%.*]] = getelementptr i8, ptr [[TMP13]], i64 8
; CHECK-NEXT: [[COL_LOAD11:%.*]] = load <2 x float>, ptr [[VEC_GEP10]], align 4
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x float> [[COL_LOAD9]], <2 x float> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP14:%.*]] = call contract <2 x float> @llvm.fmuladd.v2f32(<2 x float> [[COL_LOAD]], <2 x float> [[SPLAT_SPLAT]], <2 x float> [[RESULT_VEC_0]])
@@ -337,7 +337,7 @@ define void @multiply_alias_2x2(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP19:%.*]] = getelementptr float, ptr [[C]], i64 [[TMP18]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr float, ptr [[TMP19]], i64 [[ROWS_IV]]
; CHECK-NEXT: store <2 x float> [[TMP15]], ptr [[TMP20]], align 8
-; CHECK-NEXT: [[VEC_GEP23:%.*]] = getelementptr float, ptr [[TMP20]], i64 2
+; CHECK-NEXT: [[VEC_GEP23:%.*]] = getelementptr i8, ptr [[TMP20]], i64 8
; CHECK-NEXT: store <2 x float> [[TMP17]], ptr [[VEC_GEP23]], align 8
; CHECK-NEXT: br i1 [[ROWS_COND_NOT]], label [[COLS_LATCH]], label [[ROWS_HEADER]]
; CHECK: cols.latch:
diff --git a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-multiple-blocks.ll b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-multiple-blocks.ll
index f7ac2d321439a5..aab3907bd622bc 100644
--- a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-multiple-blocks.ll
+++ b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-multiple-blocks.ll
@@ -10,12 +10,12 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK-LABEL: @test(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[COL_LOAD133:%.*]] = load <3 x double>, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP134:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[VEC_GEP134:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: [[COL_LOAD135:%.*]] = load <3 x double>, ptr [[VEC_GEP134]], align 8
; CHECK-NEXT: [[COL_LOAD136:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP137:%.*]] = getelementptr double, ptr [[B]], i64 2
+; CHECK-NEXT: [[VEC_GEP137:%.*]] = getelementptr i8, ptr [[B]], i64 16
; CHECK-NEXT: [[COL_LOAD138:%.*]] = load <2 x double>, ptr [[VEC_GEP137]], align 8
-; CHECK-NEXT: [[VEC_GEP139:%.*]] = getelementptr double, ptr [[B]], i64 4
+; CHECK-NEXT: [[VEC_GEP139:%.*]] = getelementptr i8, ptr [[B]], i64 32
; CHECK-NEXT: [[COL_LOAD140:%.*]] = load <2 x double>, ptr [[VEC_GEP139]], align 8
; CHECK-NEXT: [[STORE_BEGIN:%.*]] = ptrtoint ptr [[C:%.*]] to i64
; CHECK-NEXT: [[STORE_END:%.*]] = add nuw nsw i64 [[STORE_BEGIN]], 72
@@ -48,10 +48,10 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK: no_alias3:
; CHECK-NEXT: [[TMP7:%.*]] = phi ptr [ [[B]], [[NO_ALIAS]] ], [ [[B]], [[ALIAS_CONT1]] ], [ [[TMP6]], [[COPY2]] ]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[TMP3]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[COL_LOAD9:%.*]] = load <2 x double>, ptr [[TMP7]], align 8
-; CHECK-NEXT: [[VEC_GEP10:%.*]] = getelementptr double, ptr [[TMP7]], i64 2
+; CHECK-NEXT: [[VEC_GEP10:%.*]] = getelementptr i8, ptr [[TMP7]], i64 16
; CHECK-NEXT: [[COL_LOAD11:%.*]] = load <2 x double>, ptr [[VEC_GEP10]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD9]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP8:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -62,14 +62,14 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK-NEXT: [[SPLAT_SPLAT20:%.*]] = shufflevector <2 x double> [[COL_LOAD11]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP11:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD8]], <2 x double> [[SPLAT_SPLAT20]], <2 x double> [[TMP10]])
; CHECK-NEXT: store <2 x double> [[TMP9]], ptr [[C]], align 8
-; CHECK-NEXT: [[VEC_GEP21:%.*]] = getelementptr double, ptr [[C]], i64 3
+; CHECK-NEXT: [[VEC_GEP21:%.*]] = getelementptr i8, ptr [[C]], i64 24
; CHECK-NEXT: store <2 x double> [[TMP11]], ptr [[VEC_GEP21]], align 8
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD22:%.*]] = load <1 x double>, ptr [[TMP12]], align 8
-; CHECK-NEXT: [[VEC_GEP23:%.*]] = getelementptr double, ptr [[TMP3]], i64 5
+; CHECK-NEXT: [[VEC_GEP23:%.*]] = getelementptr i8, ptr [[TMP3]], i64 40
; CHECK-NEXT: [[COL_LOAD24:%.*]] = load <1 x double>, ptr [[VEC_GEP23]], align 8
; CHECK-NEXT: [[COL_LOAD25:%.*]] = load <2 x double>, ptr [[TMP7]], align 8
-; CHECK-NEXT: [[VEC_GEP26:%.*]] = getelementptr double, ptr [[TMP7]], i64 2
+; CHECK-NEXT: [[VEC_GEP26:%.*]] = getelementptr i8, ptr [[TMP7]], i64 16
; CHECK-NEXT: [[COL_LOAD27:%.*]] = load <2 x double>, ptr [[VEC_GEP26]], align 8
; CHECK-NEXT: [[SPLAT_SPLATINSERT29:%.*]] = shufflevector <2 x double> [[COL_LOAD25]], <2 x double> poison, <1 x i32> zeroinitializer
; CHECK-NEXT: [[TMP13:%.*]] = fmul contract <1 x double> [[COL_LOAD22]], [[SPLAT_SPLATINSERT29]]
@@ -79,39 +79,39 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK-NEXT: [[TMP15:%.*]] = fmul contract <1 x double> [[COL_LOAD22]], [[SPLAT_SPLATINSERT35]]
; CHECK-NEXT: [[SPLAT_SPLATINSERT38:%.*]] = shufflevector <2 x double> [[COL_LOAD27]], <2 x double> poison, <1 x i32> <i32 1>
; CHECK-NEXT: [[TMP16:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD24]], <1 x double> [[SPLAT_SPLATINSERT38]], <1 x double> [[TMP15]])
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <1 x double> [[TMP14]], ptr [[TMP17]], align 8
-; CHECK-NEXT: [[VEC_GEP40:%.*]] = getelementptr double, ptr [[C]], i64 5
+; CHECK-NEXT: [[VEC_GEP40:%.*]] = getelementptr i8, ptr [[C]], i64 40
; CHECK-NEXT: store <1 x double> [[TMP16]], ptr [[VEC_GEP40]], align 8
; CHECK-NEXT: [[COL_LOAD41:%.*]] = load <2 x double>, ptr [[TMP3]], align 8
-; CHECK-NEXT: [[VEC_GEP42:%.*]] = getelementptr double, ptr [[TMP3]], i64 3
+; CHECK-NEXT: [[VEC_GEP42:%.*]] = getelementptr i8, ptr [[TMP3]], i64 24
; CHECK-NEXT: [[COL_LOAD43:%.*]] = load <2 x double>, ptr [[VEC_GEP42]], align 8
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr double, ptr [[TMP7]], i64 4
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[TMP7]], i64 32
; CHECK-NEXT: [[COL_LOAD44:%.*]] = load <2 x double>, ptr [[TMP18]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT47:%.*]] = shufflevector <2 x double> [[COL_LOAD44]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP19:%.*]] = fmul contract <2 x double> [[COL_LOAD41]], [[SPLAT_SPLAT47]]
; CHECK-NEXT: [[SPLAT_SPLAT50:%.*]] = shufflevector <2 x double> [[COL_LOAD44]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP20:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD43]], <2 x double> [[SPLAT_SPLAT50]], <2 x double> [[TMP19]])
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr double, ptr [[C]], i64 6
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[C]], i64 48
; CHECK-NEXT: store <2 x double> [[TMP20]], ptr [[TMP21]], align 8
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD51:%.*]] = load <1 x double>, ptr [[TMP22]], align 8
-; CHECK-NEXT: [[VEC_GEP52:%.*]] = getelementptr double, ptr [[TMP3]], i64 5
+; CHECK-NEXT: [[VEC_GEP52:%.*]] = getelementptr i8, ptr [[TMP3]], i64 40
; CHECK-NEXT: [[COL_LOAD53:%.*]] = load <1 x double>, ptr [[VEC_GEP52]], align 8
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr double, ptr [[TMP7]], i64 4
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[TMP7]], i64 32
; CHECK-NEXT: [[COL_LOAD54:%.*]] = load <2 x double>, ptr [[TMP23]], align 8
; CHECK-NEXT: [[SPLAT_SPLATINSERT56:%.*]] = shufflevector <2 x double> [[COL_LOAD54]], <2 x double> poison, <1 x i32> zeroinitializer
; CHECK-NEXT: [[TMP24:%.*]] = fmul contract <1 x double> [[COL_LOAD51]], [[SPLAT_SPLATINSERT56]]
; CHECK-NEXT: [[SPLAT_SPLATINSERT59:%.*]] = shufflevector <2 x double> [[COL_LOAD54]], <2 x double> poison, <1 x i32> <i32 1>
; CHECK-NEXT: [[TMP25:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD53]], <1 x double> [[SPLAT_SPLATINSERT59]], <1 x double> [[TMP24]])
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr double, ptr [[C]], i64 8
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[C]], i64 64
; CHECK-NEXT: store <1 x double> [[TMP25]], ptr [[TMP26]], align 8
; CHECK-NEXT: br i1 [[COND:%.*]], label [[TRUE:%.*]], label [[FALSE:%.*]]
; CHECK: true:
; CHECK-NEXT: [[TMP27:%.*]] = fadd contract <3 x double> [[COL_LOAD133]], [[COL_LOAD133]]
; CHECK-NEXT: [[TMP28:%.*]] = fadd contract <3 x double> [[COL_LOAD135]], [[COL_LOAD135]]
; CHECK-NEXT: store <3 x double> [[TMP27]], ptr [[A]], align 8
-; CHECK-NEXT: [[VEC_GEP143:%.*]] = getelementptr double, ptr [[A]], i64 3
+; CHECK-NEXT: [[VEC_GEP143:%.*]] = getelementptr i8, ptr [[A]], i64 24
; CHECK-NEXT: store <3 x double> [[TMP28]], ptr [[VEC_GEP143]], align 8
; CHECK-NEXT: br label [[END:%.*]]
; CHECK: false:
@@ -119,9 +119,9 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK-NEXT: [[TMP30:%.*]] = fadd contract <2 x double> [[COL_LOAD138]], [[COL_LOAD138]]
; CHECK-NEXT: [[TMP31:%.*]] = fadd contract <2 x double> [[COL_LOAD140]], [[COL_LOAD140]]
; CHECK-NEXT: store <2 x double> [[TMP29]], ptr [[B]], align 8
-; CHECK-NEXT: [[VEC_GEP141:%.*]] = getelementptr double, ptr [[B]], i64 2
+; CHECK-NEXT: [[VEC_GEP141:%.*]] = getelementptr i8, ptr [[B]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP30]], ptr [[VEC_GEP141]], align 8
-; CHECK-NEXT: [[VEC_GEP142:%.*]] = getelementptr double, ptr [[B]], i64 4
+; CHECK-NEXT: [[VEC_GEP142:%.*]] = getelementptr i8, ptr [[B]], i64 32
; CHECK-NEXT: store <2 x double> [[TMP31]], ptr [[VEC_GEP142]], align 8
; CHECK-NEXT: br label [[END]]
; CHECK: end:
@@ -156,10 +156,10 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK: no_alias70:
; CHECK-NEXT: [[TMP39:%.*]] = phi ptr [ [[B]], [[NO_ALIAS63]] ], [ [[B]], [[ALIAS_CONT68]] ], [ [[TMP38]], [[COPY69]] ]
; CHECK-NEXT: [[COL_LOAD75:%.*]] = load <2 x double>, ptr [[TMP35]], align 8
-; CHECK-NEXT: [[VEC_GEP76:%.*]] = getelementptr double, ptr [[TMP35]], i64 3
+; CHECK-NEXT: [[VEC_GEP76:%.*]] = getelementptr i8, ptr [[TMP35]], i64 24
; CHECK-NEXT: [[COL_LOAD77:%.*]] = load <2 x double>, ptr [[VEC_GEP76]], align 8
; CHECK-NEXT: [[COL_LOAD78:%.*]] = load <2 x double>, ptr [[TMP39]], align 8
-; CHECK-NEXT: [[VEC_GEP79:%.*]] = getelementptr double, ptr [[TMP39]], i64 2
+; CHECK-NEXT: [[VEC_GEP79:%.*]] = getelementptr i8, ptr [[TMP39]], i64 16
; CHECK-NEXT: [[COL_LOAD80:%.*]] = load <2 x double>, ptr [[VEC_GEP79]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT83:%.*]] = shufflevector <2 x double> [[COL_LOAD78]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP40:%.*]] = fmul contract <2 x double> [[COL_LOAD75]], [[SPLAT_SPLAT83]]
@@ -170,14 +170,14 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK-NEXT: [[SPLAT_SPLAT92:%.*]] = shufflevector <2 x double> [[COL_LOAD80]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP43:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD77]], <2 x double> [[SPLAT_SPLAT92]], <2 x double> [[TMP42]])
; CHECK-NEXT: store <2 x double> [[TMP41]], ptr [[C]], align 8
-; CHECK-NEXT: [[VEC_GEP93:%.*]] = getelementptr double, ptr [[C]], i64 3
+; CHECK-NEXT: [[VEC_GEP93:%.*]] = getelementptr i8, ptr [[C]], i64 24
; CHECK-NEXT: store <2 x double> [[TMP43]], ptr [[VEC_GEP93]], align 8
-; CHECK-NEXT: [[TMP44:%.*]] = getelementptr double, ptr [[TMP35]], i64 2
+; CHECK-NEXT: [[TMP44:%.*]] = getelementptr i8, ptr [[TMP35]], i64 16
; CHECK-NEXT: [[COL_LOAD94:%.*]] = load <1 x double>, ptr [[TMP44]], align 8
-; CHECK-NEXT: [[VEC_GEP95:%.*]] = getelementptr double, ptr [[TMP35]], i64 5
+; CHECK-NEXT: [[VEC_GEP95:%.*]] = getelementptr i8, ptr [[TMP35]], i64 40
; CHECK-NEXT: [[COL_LOAD96:%.*]] = load <1 x double>, ptr [[VEC_GEP95]], align 8
; CHECK-NEXT: [[COL_LOAD97:%.*]] = load <2 x double>, ptr [[TMP39]], align 8
-; CHECK-NEXT: [[VEC_GEP98:%.*]] = getelementptr double, ptr [[TMP39]], i64 2
+; CHECK-NEXT: [[VEC_GEP98:%.*]] = getelementptr i8, ptr [[TMP39]], i64 16
; CHECK-NEXT: [[COL_LOAD99:%.*]] = load <2 x double>, ptr [[VEC_GEP98]], align 8
; CHECK-NEXT: [[SPLAT_SPLATINSERT101:%.*]] = shufflevector <2 x double> [[COL_LOAD97]], <2 x double> poison, <1 x i32> zeroinitializer
; CHECK-NEXT: [[TMP45:%.*]] = fmul contract <1 x double> [[COL_LOAD94]], [[SPLAT_SPLATINSERT101]]
@@ -187,32 +187,32 @@ define void @test(ptr %A, ptr %B, ptr %C, i1 %cond) {
; CHECK-NEXT: [[TMP47:%.*]] = fmul contract <1 x double> [[COL_LOAD94]], [[SPLAT_SPLATINSERT107]]
; CHECK-NEXT: [[SPLAT_SPLATINSERT110:%.*]] = shufflevector <2 x double> [[COL_LOAD99]], <2 x double> poison, <1 x i32> <i32 1>
; CHECK-NEXT: [[TMP48:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD96]], <1 x double> [[SPLAT_SPLATINSERT110]], <1 x double> [[TMP47]])
-; CHECK-NEXT: [[TMP49:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[TMP49:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <1 x double> [[TMP46]], ptr [[TMP49]], align 8
-; CHECK-NEXT: [[VEC_GEP112:%.*]] = getelementptr double, ptr [[C]], i64 5
+; CHECK-NEXT: [[VEC_GEP112:%.*]] = getelementptr i8, ptr [[C]], i64 40
; CHECK-NEXT: store <1 x double> [[TMP48]], ptr [[VEC_GEP112]], align 8
; CHECK-NEXT: [[COL_LOAD113:%.*]] = load <2 x double>, ptr [[TMP35]], align 8
-; CHECK-NEXT: [[VEC_GEP114:%.*]] = getelementptr double, ptr [[TMP35]], i64 3
+; CHECK-NEXT: [[VEC_GEP114:%.*]] = getelementptr i8, ptr [[TMP35]], i64 24
; CHECK-NEXT: [[COL_LOAD115:%.*]] = load <2 x double>, ptr [[VEC_GEP114]], align 8
-; CHECK-NEXT: [[TMP50:%.*]] = getelementptr double, ptr [[TMP39]], i64 4
+; CHECK-NEXT: [[TMP50:%.*]] = getelementptr i8, ptr [[TMP39]], i64 32
; CHECK-NEXT: [[COL_LOAD116:%.*]] = load <2 x double>, ptr [[TMP50]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT119:%.*]] = shufflevector <2 x double> [[COL_LOAD116]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP51:%.*]] = fmul contract <2 x double> [[COL_LOAD113]], [[SPLAT_SPLAT119]]
; CHECK-NEXT: [[SPLAT_SPLAT122:%.*]] = shufflevector <2 x double> [[COL_LOAD116]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP52:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD115]], <2 x double> [[SPLAT_SPLAT122]], <2 x double> [[TMP51]])
-; CHECK-NEXT: [[TMP53:%.*]] = getelementptr double, ptr [[C]], i64 6
+; CHECK-NEXT: [[TMP53:%.*]] = getelementptr i8, ptr [[C]], i64 48
; CHECK-NEXT: store <2 x double> [[TMP52]], ptr [[TMP53]], align 8
-; CHECK-NEXT: [[TMP54:%.*]] = getelementptr double, ptr [[TMP35]], i64 2
+; CHECK-NEXT: [[TMP54:%.*]] = getelementptr i8, ptr [[TMP35]], i64 16
; CHECK-NEXT: [[COL_LOAD123:%.*]] = load <1 x double>, ptr [[TMP54]], align 8
-; CHECK-NEXT: [[VEC_GEP124:%.*]] = getelementptr double, ptr [[TMP35]], i64 5
+; CHECK-NEXT: [[VEC_GEP124:%.*]] = getelementptr i8, ptr [[TMP35]], i64 40
; CHECK-NEXT: [[COL_LOAD125:%.*]] = load <1 x double>, ptr [[VEC_GEP124]], align 8
-; CHECK-NEXT: [[TMP55:%.*]] = getelementptr double, ptr [[TMP39]], i64 4
+; CHECK-NEXT: [[TMP55:%.*]] = getelementptr i8, ptr [[TMP39]], i64 32
; CHECK-NEXT: [[COL_LOAD126:%.*]] = load <2 x double>, ptr [[TMP55]], align 8
; CHECK-NEXT: [[SPLAT_SPLATINSERT128:%.*]] = shufflevector <2 x double> [[COL_LOAD126]], <2 x double> poison, <1 x i32> zeroinitializer
; CHECK-NEXT: [[TMP56:%.*]] = fmul contract <1 x double> [[COL_LOAD123]], [[SPLAT_SPLATINSERT128]]
; CHECK-NEXT: [[SPLAT_SPLATINSERT131:%.*]] = shufflevector <2 x double> [[COL_LOAD126]], <2 x double> poison, <1 x i32> <i32 1>
; CHECK-NEXT: [[TMP57:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[COL_LOAD125]], <1 x double> [[SPLAT_SPLATINSERT131]], <1 x double> [[TMP56]])
-; CHECK-NEXT: [[TMP58:%.*]] = getelementptr double, ptr [[C]], i64 8
+; CHECK-NEXT: [[TMP58:%.*]] = getelementptr i8, ptr [[C]], i64 64
; CHECK-NEXT: store <1 x double> [[TMP57]], ptr [[TMP58]], align 8
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused.ll b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused.ll
index 81d1fc5861511d..155f7755c20952 100644
--- a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused.ll
+++ b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused.ll
@@ -42,10 +42,10 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK: no_alias3:
; CHECK-NEXT: [[TMP7:%.*]] = phi ptr [ [[B]], [[NO_ALIAS]] ], [ [[B]], [[ALIAS_CONT1]] ], [ [[TMP6]], [[COPY2]] ]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[TMP3]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[TMP3]], i64 4
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP3]], i64 32
; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[COL_LOAD9:%.*]] = load <2 x double>, ptr [[TMP7]], align 8
-; CHECK-NEXT: [[VEC_GEP10:%.*]] = getelementptr double, ptr [[TMP7]], i64 4
+; CHECK-NEXT: [[VEC_GEP10:%.*]] = getelementptr i8, ptr [[TMP7]], i64 32
; CHECK-NEXT: [[COL_LOAD11:%.*]] = load <2 x double>, ptr [[VEC_GEP10]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD9]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP8:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -55,13 +55,13 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP10:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT17]]
; CHECK-NEXT: [[SPLAT_SPLAT20:%.*]] = shufflevector <2 x double> [[COL_LOAD11]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP11:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD8]], <2 x double> [[SPLAT_SPLAT20]], <2 x double> [[TMP10]])
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr double, ptr [[TMP3]], i64 8
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP3]], i64 64
; CHECK-NEXT: [[COL_LOAD21:%.*]] = load <2 x double>, ptr [[TMP12]], align 8
-; CHECK-NEXT: [[VEC_GEP22:%.*]] = getelementptr double, ptr [[TMP3]], i64 12
+; CHECK-NEXT: [[VEC_GEP22:%.*]] = getelementptr i8, ptr [[TMP3]], i64 96
; CHECK-NEXT: [[COL_LOAD23:%.*]] = load <2 x double>, ptr [[VEC_GEP22]], align 8
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr double, ptr [[TMP7]], i64 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP7]], i64 16
; CHECK-NEXT: [[COL_LOAD24:%.*]] = load <2 x double>, ptr [[TMP13]], align 8
-; CHECK-NEXT: [[VEC_GEP25:%.*]] = getelementptr double, ptr [[TMP7]], i64 6
+; CHECK-NEXT: [[VEC_GEP25:%.*]] = getelementptr i8, ptr [[TMP7]], i64 48
; CHECK-NEXT: [[COL_LOAD26:%.*]] = load <2 x double>, ptr [[VEC_GEP25]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT30:%.*]] = shufflevector <2 x double> [[COL_LOAD24]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP14:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD21]], <2 x double> [[SPLAT_SPLAT30]], <2 x double> [[TMP9]])
@@ -72,14 +72,14 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[SPLAT_SPLAT40:%.*]] = shufflevector <2 x double> [[COL_LOAD26]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP17:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD23]], <2 x double> [[SPLAT_SPLAT40]], <2 x double> [[TMP16]])
; CHECK-NEXT: store <2 x double> [[TMP15]], ptr [[C]], align 8
-; CHECK-NEXT: [[VEC_GEP41:%.*]] = getelementptr double, ptr [[C]], i64 4
+; CHECK-NEXT: [[VEC_GEP41:%.*]] = getelementptr i8, ptr [[C]], i64 32
; CHECK-NEXT: store <2 x double> [[TMP17]], ptr [[VEC_GEP41]], align 8
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD42:%.*]] = load <2 x double>, ptr [[TMP18]], align 8
-; CHECK-NEXT: [[VEC_GEP43:%.*]] = getelementptr double, ptr [[TMP3]], i64 6
+; CHECK-NEXT: [[VEC_GEP43:%.*]] = getelementptr i8, ptr [[TMP3]], i64 48
; CHECK-NEXT: [[COL_LOAD44:%.*]] = load <2 x double>, ptr [[VEC_GEP43]], align 8
; CHECK-NEXT: [[COL_LOAD45:%.*]] = load <2 x double>, ptr [[TMP7]], align 8
-; CHECK-NEXT: [[VEC_GEP46:%.*]] = getelementptr double, ptr [[TMP7]], i64 4
+; CHECK-NEXT: [[VEC_GEP46:%.*]] = getelementptr i8, ptr [[TMP7]], i64 32
; CHECK-NEXT: [[COL_LOAD47:%.*]] = load <2 x double>, ptr [[VEC_GEP46]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT50:%.*]] = shufflevector <2 x double> [[COL_LOAD45]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP19:%.*]] = fmul contract <2 x double> [[COL_LOAD42]], [[SPLAT_SPLAT50]]
@@ -89,13 +89,13 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP21:%.*]] = fmul contract <2 x double> [[COL_LOAD42]], [[SPLAT_SPLAT56]]
; CHECK-NEXT: [[SPLAT_SPLAT59:%.*]] = shufflevector <2 x double> [[COL_LOAD47]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP22:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD44]], <2 x double> [[SPLAT_SPLAT59]], <2 x double> [[TMP21]])
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr double, ptr [[TMP3]], i64 10
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[TMP3]], i64 80
; CHECK-NEXT: [[COL_LOAD60:%.*]] = load <2 x double>, ptr [[TMP23]], align 8
-; CHECK-NEXT: [[VEC_GEP61:%.*]] = getelementptr double, ptr [[TMP3]], i64 14
+; CHECK-NEXT: [[VEC_GEP61:%.*]] = getelementptr i8, ptr [[TMP3]], i64 112
; CHECK-NEXT: [[COL_LOAD62:%.*]] = load <2 x double>, ptr [[VEC_GEP61]], align 8
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr double, ptr [[TMP7]], i64 2
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[TMP7]], i64 16
; CHECK-NEXT: [[COL_LOAD63:%.*]] = load <2 x double>, ptr [[TMP24]], align 8
-; CHECK-NEXT: [[VEC_GEP64:%.*]] = getelementptr double, ptr [[TMP7]], i64 6
+; CHECK-NEXT: [[VEC_GEP64:%.*]] = getelementptr i8, ptr [[TMP7]], i64 48
; CHECK-NEXT: [[COL_LOAD65:%.*]] = load <2 x double>, ptr [[VEC_GEP64]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT69:%.*]] = shufflevector <2 x double> [[COL_LOAD63]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP25:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD60]], <2 x double> [[SPLAT_SPLAT69]], <2 x double> [[TMP20]])
@@ -105,16 +105,16 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP27:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD60]], <2 x double> [[SPLAT_SPLAT76]], <2 x double> [[TMP22]])
; CHECK-NEXT: [[SPLAT_SPLAT79:%.*]] = shufflevector <2 x double> [[COL_LOAD65]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP28:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD62]], <2 x double> [[SPLAT_SPLAT79]], <2 x double> [[TMP27]])
-; CHECK-NEXT: [[TMP29:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[TMP29:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP26]], ptr [[TMP29]], align 8
-; CHECK-NEXT: [[VEC_GEP80:%.*]] = getelementptr double, ptr [[C]], i64 6
+; CHECK-NEXT: [[VEC_GEP80:%.*]] = getelementptr i8, ptr [[C]], i64 48
; CHECK-NEXT: store <2 x double> [[TMP28]], ptr [[VEC_GEP80]], align 8
; CHECK-NEXT: [[COL_LOAD81:%.*]] = load <2 x double>, ptr [[TMP3]], align 8
-; CHECK-NEXT: [[VEC_GEP82:%.*]] = getelementptr double, ptr [[TMP3]], i64 4
+; CHECK-NEXT: [[VEC_GEP82:%.*]] = getelementptr i8, ptr [[TMP3]], i64 32
; CHECK-NEXT: [[COL_LOAD83:%.*]] = load <2 x double>, ptr [[VEC_GEP82]], align 8
-; CHECK-NEXT: [[TMP30:%.*]] = getelementptr double, ptr [[TMP7]], i64 8
+; CHECK-NEXT: [[TMP30:%.*]] = getelementptr i8, ptr [[TMP7]], i64 64
; CHECK-NEXT: [[COL_LOAD84:%.*]] = load <2 x double>, ptr [[TMP30]], align 8
-; CHECK-NEXT: [[VEC_GEP85:%.*]] = getelementptr double, ptr [[TMP7]], i64 12
+; CHECK-NEXT: [[VEC_GEP85:%.*]] = getelementptr i8, ptr [[TMP7]], i64 96
; CHECK-NEXT: [[COL_LOAD86:%.*]] = load <2 x double>, ptr [[VEC_GEP85]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT89:%.*]] = shufflevector <2 x double> [[COL_LOAD84]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP31:%.*]] = fmul contract <2 x double> [[COL_LOAD81]], [[SPLAT_SPLAT89]]
@@ -124,13 +124,13 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP33:%.*]] = fmul contract <2 x double> [[COL_LOAD81]], [[SPLAT_SPLAT95]]
; CHECK-NEXT: [[SPLAT_SPLAT98:%.*]] = shufflevector <2 x double> [[COL_LOAD86]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP34:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD83]], <2 x double> [[SPLAT_SPLAT98]], <2 x double> [[TMP33]])
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr double, ptr [[TMP3]], i64 8
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr i8, ptr [[TMP3]], i64 64
; CHECK-NEXT: [[COL_LOAD99:%.*]] = load <2 x double>, ptr [[TMP35]], align 8
-; CHECK-NEXT: [[VEC_GEP100:%.*]] = getelementptr double, ptr [[TMP3]], i64 12
+; CHECK-NEXT: [[VEC_GEP100:%.*]] = getelementptr i8, ptr [[TMP3]], i64 96
; CHECK-NEXT: [[COL_LOAD101:%.*]] = load <2 x double>, ptr [[VEC_GEP100]], align 8
-; CHECK-NEXT: [[TMP36:%.*]] = getelementptr double, ptr [[TMP7]], i64 10
+; CHECK-NEXT: [[TMP36:%.*]] = getelementptr i8, ptr [[TMP7]], i64 80
; CHECK-NEXT: [[COL_LOAD102:%.*]] = load <2 x double>, ptr [[TMP36]], align 8
-; CHECK-NEXT: [[VEC_GEP103:%.*]] = getelementptr double, ptr [[TMP7]], i64 14
+; CHECK-NEXT: [[VEC_GEP103:%.*]] = getelementptr i8, ptr [[TMP7]], i64 112
; CHECK-NEXT: [[COL_LOAD104:%.*]] = load <2 x double>, ptr [[VEC_GEP103]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT108:%.*]] = shufflevector <2 x double> [[COL_LOAD102]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP37:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD99]], <2 x double> [[SPLAT_SPLAT108]], <2 x double> [[TMP32]])
@@ -140,17 +140,17 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP39:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD99]], <2 x double> [[SPLAT_SPLAT115]], <2 x double> [[TMP34]])
; CHECK-NEXT: [[SPLAT_SPLAT118:%.*]] = shufflevector <2 x double> [[COL_LOAD104]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP40:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD101]], <2 x double> [[SPLAT_SPLAT118]], <2 x double> [[TMP39]])
-; CHECK-NEXT: [[TMP41:%.*]] = getelementptr double, ptr [[C]], i64 8
+; CHECK-NEXT: [[TMP41:%.*]] = getelementptr i8, ptr [[C]], i64 64
; CHECK-NEXT: store <2 x double> [[TMP38]], ptr [[TMP41]], align 8
-; CHECK-NEXT: [[VEC_GEP119:%.*]] = getelementptr double, ptr [[C]], i64 12
+; CHECK-NEXT: [[VEC_GEP119:%.*]] = getelementptr i8, ptr [[C]], i64 96
; CHECK-NEXT: store <2 x double> [[TMP40]], ptr [[VEC_GEP119]], align 8
-; CHECK-NEXT: [[TMP42:%.*]] = getelementptr double, ptr [[TMP3]], i64 2
+; CHECK-NEXT: [[TMP42:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD120:%.*]] = load <2 x double>, ptr [[TMP42]], align 8
-; CHECK-NEXT: [[VEC_GEP121:%.*]] = getelementptr double, ptr [[TMP3]], i64 6
+; CHECK-NEXT: [[VEC_GEP121:%.*]] = getelementptr i8, ptr [[TMP3]], i64 48
; CHECK-NEXT: [[COL_LOAD122:%.*]] = load <2 x double>, ptr [[VEC_GEP121]], align 8
-; CHECK-NEXT: [[TMP43:%.*]] = getelementptr double, ptr [[TMP7]], i64 8
+; CHECK-NEXT: [[TMP43:%.*]] = getelementptr i8, ptr [[TMP7]], i64 64
; CHECK-NEXT: [[COL_LOAD123:%.*]] = load <2 x double>, ptr [[TMP43]], align 8
-; CHECK-NEXT: [[VEC_GEP124:%.*]] = getelementptr double, ptr [[TMP7]], i64 12
+; CHECK-NEXT: [[VEC_GEP124:%.*]] = getelementptr i8, ptr [[TMP7]], i64 96
; CHECK-NEXT: [[COL_LOAD125:%.*]] = load <2 x double>, ptr [[VEC_GEP124]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT128:%.*]] = shufflevector <2 x double> [[COL_LOAD123]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP44:%.*]] = fmul contract <2 x double> [[COL_LOAD120]], [[SPLAT_SPLAT128]]
@@ -160,13 +160,13 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP46:%.*]] = fmul contract <2 x double> [[COL_LOAD120]], [[SPLAT_SPLAT134]]
; CHECK-NEXT: [[SPLAT_SPLAT137:%.*]] = shufflevector <2 x double> [[COL_LOAD125]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP47:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD122]], <2 x double> [[SPLAT_SPLAT137]], <2 x double> [[TMP46]])
-; CHECK-NEXT: [[TMP48:%.*]] = getelementptr double, ptr [[TMP3]], i64 10
+; CHECK-NEXT: [[TMP48:%.*]] = getelementptr i8, ptr [[TMP3]], i64 80
; CHECK-NEXT: [[COL_LOAD138:%.*]] = load <2 x double>, ptr [[TMP48]], align 8
-; CHECK-NEXT: [[VEC_GEP139:%.*]] = getelementptr double, ptr [[TMP3]], i64 14
+; CHECK-NEXT: [[VEC_GEP139:%.*]] = getelementptr i8, ptr [[TMP3]], i64 112
; CHECK-NEXT: [[COL_LOAD140:%.*]] = load <2 x double>, ptr [[VEC_GEP139]], align 8
-; CHECK-NEXT: [[TMP49:%.*]] = getelementptr double, ptr [[TMP7]], i64 10
+; CHECK-NEXT: [[TMP49:%.*]] = getelementptr i8, ptr [[TMP7]], i64 80
; CHECK-NEXT: [[COL_LOAD141:%.*]] = load <2 x double>, ptr [[TMP49]], align 8
-; CHECK-NEXT: [[VEC_GEP142:%.*]] = getelementptr double, ptr [[TMP7]], i64 14
+; CHECK-NEXT: [[VEC_GEP142:%.*]] = getelementptr i8, ptr [[TMP7]], i64 112
; CHECK-NEXT: [[COL_LOAD143:%.*]] = load <2 x double>, ptr [[VEC_GEP142]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT147:%.*]] = shufflevector <2 x double> [[COL_LOAD141]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP50:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD138]], <2 x double> [[SPLAT_SPLAT147]], <2 x double> [[TMP45]])
@@ -176,9 +176,9 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[TMP52:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD138]], <2 x double> [[SPLAT_SPLAT154]], <2 x double> [[TMP47]])
; CHECK-NEXT: [[SPLAT_SPLAT157:%.*]] = shufflevector <2 x double> [[COL_LOAD143]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP53:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD140]], <2 x double> [[SPLAT_SPLAT157]], <2 x double> [[TMP52]])
-; CHECK-NEXT: [[TMP54:%.*]] = getelementptr double, ptr [[C]], i64 10
+; CHECK-NEXT: [[TMP54:%.*]] = getelementptr i8, ptr [[C]], i64 80
; CHECK-NEXT: store <2 x double> [[TMP51]], ptr [[TMP54]], align 8
-; CHECK-NEXT: [[VEC_GEP158:%.*]] = getelementptr double, ptr [[C]], i64 14
+; CHECK-NEXT: [[VEC_GEP158:%.*]] = getelementptr i8, ptr [[C]], i64 112
; CHECK-NEXT: store <2 x double> [[TMP53]], ptr [[VEC_GEP158]], align 8
; CHECK-NEXT: ret void
;
@@ -234,7 +234,7 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-LABEL: @multiply_reuse_load(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 4
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[A]], i64 32
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <2 x double> [[COL_LOAD]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -244,13 +244,13 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[TMP2:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT10]]
; CHECK-NEXT: [[SPLAT_SPLAT13:%.*]] = shufflevector <2 x double> [[COL_LOAD1]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP3:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD1]], <2 x double> [[SPLAT_SPLAT13]], <2 x double> [[TMP2]])
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr double, ptr [[A]], i64 8
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[A]], i64 64
; CHECK-NEXT: [[COL_LOAD14:%.*]] = load <2 x double>, ptr [[TMP4]], align 8
-; CHECK-NEXT: [[VEC_GEP15:%.*]] = getelementptr double, ptr [[A]], i64 12
+; CHECK-NEXT: [[VEC_GEP15:%.*]] = getelementptr i8, ptr [[A]], i64 96
; CHECK-NEXT: [[COL_LOAD16:%.*]] = load <2 x double>, ptr [[VEC_GEP15]], align 8
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD17:%.*]] = load <2 x double>, ptr [[TMP5]], align 8
-; CHECK-NEXT: [[VEC_GEP18:%.*]] = getelementptr double, ptr [[A]], i64 6
+; CHECK-NEXT: [[VEC_GEP18:%.*]] = getelementptr i8, ptr [[A]], i64 48
; CHECK-NEXT: [[COL_LOAD19:%.*]] = load <2 x double>, ptr [[VEC_GEP18]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT23:%.*]] = shufflevector <2 x double> [[COL_LOAD17]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP6:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD14]], <2 x double> [[SPLAT_SPLAT23]], <2 x double> [[TMP1]])
@@ -261,14 +261,14 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[SPLAT_SPLAT33:%.*]] = shufflevector <2 x double> [[COL_LOAD19]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP9:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD16]], <2 x double> [[SPLAT_SPLAT33]], <2 x double> [[TMP8]])
; CHECK-NEXT: store <2 x double> [[TMP7]], ptr [[C:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP34:%.*]] = getelementptr double, ptr [[C]], i64 4
+; CHECK-NEXT: [[VEC_GEP34:%.*]] = getelementptr i8, ptr [[C]], i64 32
; CHECK-NEXT: store <2 x double> [[TMP9]], ptr [[VEC_GEP34]], align 8
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD35:%.*]] = load <2 x double>, ptr [[TMP10]], align 8
-; CHECK-NEXT: [[VEC_GEP36:%.*]] = getelementptr double, ptr [[A]], i64 6
+; CHECK-NEXT: [[VEC_GEP36:%.*]] = getelementptr i8, ptr [[A]], i64 48
; CHECK-NEXT: [[COL_LOAD37:%.*]] = load <2 x double>, ptr [[VEC_GEP36]], align 8
; CHECK-NEXT: [[COL_LOAD38:%.*]] = load <2 x double>, ptr [[A]], align 8
-; CHECK-NEXT: [[VEC_GEP39:%.*]] = getelementptr double, ptr [[A]], i64 4
+; CHECK-NEXT: [[VEC_GEP39:%.*]] = getelementptr i8, ptr [[A]], i64 32
; CHECK-NEXT: [[COL_LOAD40:%.*]] = load <2 x double>, ptr [[VEC_GEP39]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT43:%.*]] = shufflevector <2 x double> [[COL_LOAD38]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP11:%.*]] = fmul contract <2 x double> [[COL_LOAD35]], [[SPLAT_SPLAT43]]
@@ -278,13 +278,13 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[TMP13:%.*]] = fmul contract <2 x double> [[COL_LOAD35]], [[SPLAT_SPLAT49]]
; CHECK-NEXT: [[SPLAT_SPLAT52:%.*]] = shufflevector <2 x double> [[COL_LOAD40]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP14:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD37]], <2 x double> [[SPLAT_SPLAT52]], <2 x double> [[TMP13]])
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr double, ptr [[A]], i64 10
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[A]], i64 80
; CHECK-NEXT: [[COL_LOAD53:%.*]] = load <2 x double>, ptr [[TMP15]], align 8
-; CHECK-NEXT: [[VEC_GEP54:%.*]] = getelementptr double, ptr [[A]], i64 14
+; CHECK-NEXT: [[VEC_GEP54:%.*]] = getelementptr i8, ptr [[A]], i64 112
; CHECK-NEXT: [[COL_LOAD55:%.*]] = load <2 x double>, ptr [[VEC_GEP54]], align 8
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD56:%.*]] = load <2 x double>, ptr [[TMP16]], align 8
-; CHECK-NEXT: [[VEC_GEP57:%.*]] = getelementptr double, ptr [[A]], i64 6
+; CHECK-NEXT: [[VEC_GEP57:%.*]] = getelementptr i8, ptr [[A]], i64 48
; CHECK-NEXT: [[COL_LOAD58:%.*]] = load <2 x double>, ptr [[VEC_GEP57]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT62:%.*]] = shufflevector <2 x double> [[COL_LOAD56]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP17:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD53]], <2 x double> [[SPLAT_SPLAT62]], <2 x double> [[TMP12]])
@@ -294,16 +294,16 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[TMP19:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD53]], <2 x double> [[SPLAT_SPLAT69]], <2 x double> [[TMP14]])
; CHECK-NEXT: [[SPLAT_SPLAT72:%.*]] = shufflevector <2 x double> [[COL_LOAD58]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP20:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD55]], <2 x double> [[SPLAT_SPLAT72]], <2 x double> [[TMP19]])
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP18]], ptr [[TMP21]], align 8
-; CHECK-NEXT: [[VEC_GEP73:%.*]] = getelementptr double, ptr [[C]], i64 6
+; CHECK-NEXT: [[VEC_GEP73:%.*]] = getelementptr i8, ptr [[C]], i64 48
; CHECK-NEXT: store <2 x double> [[TMP20]], ptr [[VEC_GEP73]], align 8
; CHECK-NEXT: [[COL_LOAD74:%.*]] = load <2 x double>, ptr [[A]], align 8
-; CHECK-NEXT: [[VEC_GEP75:%.*]] = getelementptr double, ptr [[A]], i64 4
+; CHECK-NEXT: [[VEC_GEP75:%.*]] = getelementptr i8, ptr [[A]], i64 32
; CHECK-NEXT: [[COL_LOAD76:%.*]] = load <2 x double>, ptr [[VEC_GEP75]], align 8
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr double, ptr [[A]], i64 8
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i8, ptr [[A]], i64 64
; CHECK-NEXT: [[COL_LOAD77:%.*]] = load <2 x double>, ptr [[TMP22]], align 8
-; CHECK-NEXT: [[VEC_GEP78:%.*]] = getelementptr double, ptr [[A]], i64 12
+; CHECK-NEXT: [[VEC_GEP78:%.*]] = getelementptr i8, ptr [[A]], i64 96
; CHECK-NEXT: [[COL_LOAD79:%.*]] = load <2 x double>, ptr [[VEC_GEP78]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT82:%.*]] = shufflevector <2 x double> [[COL_LOAD77]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP23:%.*]] = fmul contract <2 x double> [[COL_LOAD74]], [[SPLAT_SPLAT82]]
@@ -313,13 +313,13 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[TMP25:%.*]] = fmul contract <2 x double> [[COL_LOAD74]], [[SPLAT_SPLAT88]]
; CHECK-NEXT: [[SPLAT_SPLAT91:%.*]] = shufflevector <2 x double> [[COL_LOAD79]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP26:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD76]], <2 x double> [[SPLAT_SPLAT91]], <2 x double> [[TMP25]])
-; CHECK-NEXT: [[TMP27:%.*]] = getelementptr double, ptr [[A]], i64 8
+; CHECK-NEXT: [[TMP27:%.*]] = getelementptr i8, ptr [[A]], i64 64
; CHECK-NEXT: [[COL_LOAD92:%.*]] = load <2 x double>, ptr [[TMP27]], align 8
-; CHECK-NEXT: [[VEC_GEP93:%.*]] = getelementptr double, ptr [[A]], i64 12
+; CHECK-NEXT: [[VEC_GEP93:%.*]] = getelementptr i8, ptr [[A]], i64 96
; CHECK-NEXT: [[COL_LOAD94:%.*]] = load <2 x double>, ptr [[VEC_GEP93]], align 8
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr double, ptr [[A]], i64 10
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr i8, ptr [[A]], i64 80
; CHECK-NEXT: [[COL_LOAD95:%.*]] = load <2 x double>, ptr [[TMP28]], align 8
-; CHECK-NEXT: [[VEC_GEP96:%.*]] = getelementptr double, ptr [[A]], i64 14
+; CHECK-NEXT: [[VEC_GEP96:%.*]] = getelementptr i8, ptr [[A]], i64 112
; CHECK-NEXT: [[COL_LOAD97:%.*]] = load <2 x double>, ptr [[VEC_GEP96]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT101:%.*]] = shufflevector <2 x double> [[COL_LOAD95]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP29:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD92]], <2 x double> [[SPLAT_SPLAT101]], <2 x double> [[TMP24]])
@@ -329,17 +329,17 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[TMP31:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD92]], <2 x double> [[SPLAT_SPLAT108]], <2 x double> [[TMP26]])
; CHECK-NEXT: [[SPLAT_SPLAT111:%.*]] = shufflevector <2 x double> [[COL_LOAD97]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP32:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD94]], <2 x double> [[SPLAT_SPLAT111]], <2 x double> [[TMP31]])
-; CHECK-NEXT: [[TMP33:%.*]] = getelementptr double, ptr [[C]], i64 8
+; CHECK-NEXT: [[TMP33:%.*]] = getelementptr i8, ptr [[C]], i64 64
; CHECK-NEXT: store <2 x double> [[TMP30]], ptr [[TMP33]], align 8
-; CHECK-NEXT: [[VEC_GEP112:%.*]] = getelementptr double, ptr [[C]], i64 12
+; CHECK-NEXT: [[VEC_GEP112:%.*]] = getelementptr i8, ptr [[C]], i64 96
; CHECK-NEXT: store <2 x double> [[TMP32]], ptr [[VEC_GEP112]], align 8
-; CHECK-NEXT: [[TMP34:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD113:%.*]] = load <2 x double>, ptr [[TMP34]], align 8
-; CHECK-NEXT: [[VEC_GEP114:%.*]] = getelementptr double, ptr [[A]], i64 6
+; CHECK-NEXT: [[VEC_GEP114:%.*]] = getelementptr i8, ptr [[A]], i64 48
; CHECK-NEXT: [[COL_LOAD115:%.*]] = load <2 x double>, ptr [[VEC_GEP114]], align 8
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr double, ptr [[A]], i64 8
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr i8, ptr [[A]], i64 64
; CHECK-NEXT: [[COL_LOAD116:%.*]] = load <2 x double>, ptr [[TMP35]], align 8
-; CHECK-NEXT: [[VEC_GEP117:%.*]] = getelementptr double, ptr [[A]], i64 12
+; CHECK-NEXT: [[VEC_GEP117:%.*]] = getelementptr i8, ptr [[A]], i64 96
; CHECK-NEXT: [[COL_LOAD118:%.*]] = load <2 x double>, ptr [[VEC_GEP117]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT121:%.*]] = shufflevector <2 x double> [[COL_LOAD116]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP36:%.*]] = fmul contract <2 x double> [[COL_LOAD113]], [[SPLAT_SPLAT121]]
@@ -349,9 +349,9 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[TMP38:%.*]] = fmul contract <2 x double> [[COL_LOAD113]], [[SPLAT_SPLAT127]]
; CHECK-NEXT: [[SPLAT_SPLAT130:%.*]] = shufflevector <2 x double> [[COL_LOAD118]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP39:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD115]], <2 x double> [[SPLAT_SPLAT130]], <2 x double> [[TMP38]])
-; CHECK-NEXT: [[TMP40:%.*]] = getelementptr double, ptr [[A]], i64 10
+; CHECK-NEXT: [[TMP40:%.*]] = getelementptr i8, ptr [[A]], i64 80
; CHECK-NEXT: [[COL_LOAD131:%.*]] = load <2 x double>, ptr [[TMP40]], align 8
-; CHECK-NEXT: [[VEC_GEP132:%.*]] = getelementptr double, ptr [[A]], i64 14
+; CHECK-NEXT: [[VEC_GEP132:%.*]] = getelementptr i8, ptr [[A]], i64 112
; CHECK-NEXT: [[COL_LOAD133:%.*]] = load <2 x double>, ptr [[VEC_GEP132]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT140:%.*]] = shufflevector <2 x double> [[COL_LOAD131]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP41:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD131]], <2 x double> [[SPLAT_SPLAT140]], <2 x double> [[TMP37]])
@@ -361,9 +361,9 @@ define void @multiply_reuse_load(ptr noalias %A, ptr noalias %B, ptr noalias %C)
; CHECK-NEXT: [[TMP43:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD131]], <2 x double> [[SPLAT_SPLAT147]], <2 x double> [[TMP39]])
; CHECK-NEXT: [[SPLAT_SPLAT150:%.*]] = shufflevector <2 x double> [[COL_LOAD133]], <2 x double> poison, <2 x i32> <i32 1, i32 1>
; CHECK-NEXT: [[TMP44:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD133]], <2 x double> [[SPLAT_SPLAT150]], <2 x double> [[TMP43]])
-; CHECK-NEXT: [[TMP45:%.*]] = getelementptr double, ptr [[C]], i64 10
+; CHECK-NEXT: [[TMP45:%.*]] = getelementptr i8, ptr [[C]], i64 80
; CHECK-NEXT: store <2 x double> [[TMP42]], ptr [[TMP45]], align 8
-; CHECK-NEXT: [[VEC_GEP151:%.*]] = getelementptr double, ptr [[C]], i64 14
+; CHECK-NEXT: [[VEC_GEP151:%.*]] = getelementptr i8, ptr [[C]], i64 112
; CHECK-NEXT: store <2 x double> [[TMP44]], ptr [[VEC_GEP151]], align 8
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-minimal.ll b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-minimal.ll
index a02ef276e43e47..fa52ce39aad9be 100644
--- a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-minimal.ll
+++ b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-minimal.ll
@@ -14,14 +14,14 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-LABEL: @multiply(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr double, ptr [[A]], i64 2
+; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[A]], i64 16
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
-; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr double, ptr [[A]], i64 4
+; CHECK-NEXT: [[VEC_GEP2:%.*]] = getelementptr i8, ptr [[A]], i64 32
; CHECK-NEXT: [[COL_LOAD3:%.*]] = load <2 x double>, ptr [[VEC_GEP2]], align 8
-; CHECK-NEXT: [[VEC_GEP4:%.*]] = getelementptr double, ptr [[A]], i64 6
+; CHECK-NEXT: [[VEC_GEP4:%.*]] = getelementptr i8, ptr [[A]], i64 48
; CHECK-NEXT: [[COL_LOAD5:%.*]] = load <2 x double>, ptr [[VEC_GEP4]], align 8
; CHECK-NEXT: [[COL_LOAD6:%.*]] = load <4 x double>, ptr [[B:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP7:%.*]] = getelementptr double, ptr [[B]], i64 4
+; CHECK-NEXT: [[VEC_GEP7:%.*]] = getelementptr i8, ptr [[B]], i64 32
; CHECK-NEXT: [[COL_LOAD8:%.*]] = load <4 x double>, ptr [[VEC_GEP7]], align 8
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <4 x double> [[COL_LOAD6]], <4 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP0:%.*]] = fmul contract <2 x double> [[COL_LOAD]], [[SPLAT_SPLAT]]
@@ -40,7 +40,7 @@ define void @multiply(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[SPLAT_SPLAT29:%.*]] = shufflevector <4 x double> [[COL_LOAD8]], <4 x double> poison, <2 x i32> <i32 3, i32 3>
; CHECK-NEXT: [[TMP7:%.*]] = call contract <2 x double> @llvm.fmuladd.v2f64(<2 x double> [[COL_LOAD5]], <2 x double> [[SPLAT_SPLAT29]], <2 x double> [[TMP6]])
; CHECK-NEXT: store <2 x double> [[TMP3]], ptr [[C:%.*]], align 8
-; CHECK-NEXT: [[VEC_GEP30:%.*]] = getelementptr double, ptr [[C]], i64 2
+; CHECK-NEXT: [[VEC_GEP30:%.*]] = getelementptr i8, ptr [[C]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP7]], ptr [[VEC_GEP30]], align 8
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll
index b43fcc8548ce66..458c2d751e2f86 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/hoisting-sinking-required-for-vectorization.ll
@@ -49,7 +49,7 @@ define void @loop(ptr %X, ptr %Y) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[TMP1]], i64 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP1]], align 8
; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <2 x double>, ptr [[TMP2]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = fcmp olt <2 x double> [[WIDE_LOAD]], zeroinitializer
@@ -61,7 +61,7 @@ define void @loop(ptr %X, ptr %Y) {
; CHECK-NEXT: [[TMP9:%.*]] = select <2 x i1> [[TMP3]], <2 x double> zeroinitializer, <2 x double> [[TMP7]]
; CHECK-NEXT: [[TMP10:%.*]] = select <2 x i1> [[TMP4]], <2 x double> zeroinitializer, <2 x double> [[TMP8]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds double, ptr [[TMP11]], i64 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP9]], ptr [[TMP11]], align 8
; CHECK-NEXT: store <2 x double> [[TMP10]], ptr [[TMP12]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -149,27 +149,27 @@ define void @loop2(ptr %A, ptr %B, ptr %C, float %x) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 4
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4, !alias.scope !4
-; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4, !alias.scope !4
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4, !alias.scope [[META4:![0-9]+]]
+; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4, !alias.scope [[META4]]
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], <i32 20, i32 20, i32 20, i32 20>
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD7]], <i32 20, i32 20, i32 20, i32 20>
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[TMP4]], i64 4
-; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <4 x float>, ptr [[TMP4]], align 4, !alias.scope !7
-; CHECK-NEXT: [[WIDE_LOAD9:%.*]] = load <4 x float>, ptr [[TMP5]], align 4, !alias.scope !7
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 16
+; CHECK-NEXT: [[WIDE_LOAD8:%.*]] = load <4 x float>, ptr [[TMP4]], align 4, !alias.scope [[META7:![0-9]+]]
+; CHECK-NEXT: [[WIDE_LOAD9:%.*]] = load <4 x float>, ptr [[TMP5]], align 4, !alias.scope [[META7]]
; CHECK-NEXT: [[TMP6:%.*]] = fmul <4 x float> [[WIDE_LOAD8]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP7:%.*]] = fmul <4 x float> [[WIDE_LOAD9]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr float, ptr [[TMP8]], i64 4
-; CHECK-NEXT: [[WIDE_LOAD10:%.*]] = load <4 x float>, ptr [[TMP8]], align 4, !alias.scope !9, !noalias !11
-; CHECK-NEXT: [[WIDE_LOAD11:%.*]] = load <4 x float>, ptr [[TMP9]], align 4, !alias.scope !9, !noalias !11
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 16
+; CHECK-NEXT: [[WIDE_LOAD10:%.*]] = load <4 x float>, ptr [[TMP8]], align 4, !alias.scope [[META9:![0-9]+]], !noalias [[META11:![0-9]+]]
+; CHECK-NEXT: [[WIDE_LOAD11:%.*]] = load <4 x float>, ptr [[TMP9]], align 4, !alias.scope [[META9]], !noalias [[META11]]
; CHECK-NEXT: [[TMP10:%.*]] = select <4 x i1> [[TMP2]], <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, <4 x float> [[WIDE_LOAD10]]
; CHECK-NEXT: [[PREDPHI:%.*]] = fadd <4 x float> [[TMP6]], [[TMP10]]
; CHECK-NEXT: [[TMP11:%.*]] = select <4 x i1> [[TMP3]], <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, <4 x float> [[WIDE_LOAD11]]
; CHECK-NEXT: [[PREDPHI12:%.*]] = fadd <4 x float> [[TMP7]], [[TMP11]]
-; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP8]], align 4, !alias.scope !9, !noalias !11
-; CHECK-NEXT: store <4 x float> [[PREDPHI12]], ptr [[TMP9]], align 4, !alias.scope !9, !noalias !11
+; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP8]], align 4, !alias.scope [[META9]], !noalias [[META11]]
+; CHECK-NEXT: store <4 x float> [[PREDPHI12]], ptr [[TMP9]], align 4, !alias.scope [[META9]], !noalias [[META11]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10000
; CHECK-NEXT: br i1 [[TMP12]], label [[EXIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/peel-multiple-unreachable-exits-for-vectorization.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/peel-multiple-unreachable-exits-for-vectorization.ll
index e346bef27bbd0e..8fc5189e8bc79e 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/peel-multiple-unreachable-exits-for-vectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/peel-multiple-unreachable-exits-for-vectorization.ll
@@ -11,13 +11,13 @@ define i64 @sum_2_at_with_int_conversion(ptr %A, ptr %B, i64 %N) {
; CHECK-LABEL: @sum_2_at_with_int_conversion(
; CHECK-NEXT: at_with_int_conversion.exit11.peel:
; CHECK-NEXT: [[START_I:%.*]] = load ptr, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[GEP_END_I:%.*]] = getelementptr [[VEC:%.*]], ptr [[A]], i64 0, i32 1
+; CHECK-NEXT: [[GEP_END_I:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[END_I:%.*]] = load ptr, ptr [[GEP_END_I]], align 8
; CHECK-NEXT: [[START_INT_I:%.*]] = ptrtoint ptr [[START_I]] to i64
; CHECK-NEXT: [[END_INT_I:%.*]] = ptrtoint ptr [[END_I]] to i64
; CHECK-NEXT: [[SUB_I:%.*]] = sub i64 [[END_INT_I]], [[START_INT_I]]
; CHECK-NEXT: [[SMAX:%.*]] = tail call i64 @llvm.smax.i64(i64 [[N:%.*]], i64 0)
-; CHECK-NEXT: [[GEP_END_I2:%.*]] = getelementptr [[VEC]], ptr [[B:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[GEP_END_I2:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 8
; CHECK-NEXT: [[START_I1_PEEL:%.*]] = load ptr, ptr [[B]], align 8
; CHECK-NEXT: [[END_I3_PEEL:%.*]] = load ptr, ptr [[GEP_END_I2]], align 8
; CHECK-NEXT: [[START_INT_I4_PEEL:%.*]] = ptrtoint ptr [[START_I1_PEEL]] to i64
@@ -50,11 +50,11 @@ define i64 @sum_2_at_with_int_conversion(ptr %A, ptr %B, i64 %N) {
; CHECK-NEXT: [[VEC_PHI16:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = or disjoint i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i64, ptr [[START_I]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i64, ptr [[TMP6]], i64 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP6]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP6]], align 8
; CHECK-NEXT: [[WIDE_LOAD17:%.*]] = load <2 x i64>, ptr [[TMP7]], align 8
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i64, ptr [[START_I1_PEEL]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i64, ptr [[TMP8]], i64 2
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 16
; CHECK-NEXT: [[WIDE_LOAD18:%.*]] = load <2 x i64>, ptr [[TMP8]], align 8
; CHECK-NEXT: [[WIDE_LOAD19:%.*]] = load <2 x i64>, ptr [[TMP9]], align 8
; CHECK-NEXT: [[TMP10:%.*]] = add <2 x i64> [[WIDE_LOAD]], [[VEC_PHI]]
@@ -122,14 +122,14 @@ define i64 @sum_3_at_with_int_conversion(ptr %A, ptr %B, ptr %C, i64 %N) {
; CHECK-LABEL: @sum_3_at_with_int_conversion(
; CHECK-NEXT: at_with_int_conversion.exit22.peel:
; CHECK-NEXT: [[START_I:%.*]] = load ptr, ptr [[A:%.*]], align 8
-; CHECK-NEXT: [[GEP_END_I:%.*]] = getelementptr [[VEC:%.*]], ptr [[A]], i64 0, i32 1
+; CHECK-NEXT: [[GEP_END_I:%.*]] = getelementptr i8, ptr [[A]], i64 8
; CHECK-NEXT: [[END_I:%.*]] = load ptr, ptr [[GEP_END_I]], align 8
; CHECK-NEXT: [[START_INT_I:%.*]] = ptrtoint ptr [[START_I]] to i64
; CHECK-NEXT: [[END_INT_I:%.*]] = ptrtoint ptr [[END_I]] to i64
; CHECK-NEXT: [[SUB_I:%.*]] = sub i64 [[END_INT_I]], [[START_INT_I]]
-; CHECK-NEXT: [[GEP_END_I13:%.*]] = getelementptr [[VEC]], ptr [[C:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[GEP_END_I13:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 8
; CHECK-NEXT: [[SMAX:%.*]] = tail call i64 @llvm.smax.i64(i64 [[N:%.*]], i64 0)
-; CHECK-NEXT: [[GEP_END_I2:%.*]] = getelementptr [[VEC]], ptr [[B:%.*]], i64 0, i32 1
+; CHECK-NEXT: [[GEP_END_I2:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 8
; CHECK-NEXT: [[LV_I_PEEL:%.*]] = load i64, ptr [[START_I]], align 8
; CHECK-NEXT: [[START_I1_PEEL:%.*]] = load ptr, ptr [[B]], align 8
; CHECK-NEXT: [[END_I3_PEEL:%.*]] = load ptr, ptr [[GEP_END_I2]], align 8
@@ -171,15 +171,15 @@ define i64 @sum_3_at_with_int_conversion(ptr %A, ptr %B, ptr %C, i64 %N) {
; CHECK-NEXT: [[VEC_PHI28:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP17:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = or disjoint i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i64, ptr [[START_I]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i64, ptr [[TMP6]], i64 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP6]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP6]], align 8
; CHECK-NEXT: [[WIDE_LOAD29:%.*]] = load <2 x i64>, ptr [[TMP7]], align 8
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i64, ptr [[START_I1_PEEL]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i64, ptr [[TMP8]], i64 2
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 16
; CHECK-NEXT: [[WIDE_LOAD30:%.*]] = load <2 x i64>, ptr [[TMP8]], align 8
; CHECK-NEXT: [[WIDE_LOAD31:%.*]] = load <2 x i64>, ptr [[TMP9]], align 8
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i64, ptr [[START_I12_PEEL]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i64, ptr [[TMP10]], i64 2
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP10]], i64 16
; CHECK-NEXT: [[WIDE_LOAD32:%.*]] = load <2 x i64>, ptr [[TMP10]], align 8
; CHECK-NEXT: [[WIDE_LOAD33:%.*]] = load <2 x i64>, ptr [[TMP11]], align 8
; CHECK-NEXT: [[TMP12:%.*]] = add <2 x i64> [[WIDE_LOAD]], [[VEC_PHI]]
@@ -260,7 +260,7 @@ exit:
define i64 @at_with_int_conversion(ptr %ptr, i64 %idx) {
; CHECK-LABEL: @at_with_int_conversion(
; CHECK-NEXT: [[START:%.*]] = load ptr, ptr [[PTR:%.*]], align 8
-; CHECK-NEXT: [[GEP_END:%.*]] = getelementptr [[VEC:%.*]], ptr [[PTR]], i64 0, i32 1
+; CHECK-NEXT: [[GEP_END:%.*]] = getelementptr i8, ptr [[PTR]], i64 8
; CHECK-NEXT: [[END:%.*]] = load ptr, ptr [[GEP_END]], align 8
; CHECK-NEXT: [[START_INT:%.*]] = ptrtoint ptr [[START]] to i64
; CHECK-NEXT: [[END_INT:%.*]] = ptrtoint ptr [[END]] to i64
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll
index d60b2327967674..7b9f3a2e13a649 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll
@@ -40,14 +40,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[TMP12:%.*]] = trunc <8 x i32> [[TMP11]] to <8 x i16>
; CHECK-NEXT: [[PREDPHI:%.*]] = select <8 x i1> [[TMP1]], <8 x i16> [[TMP12]], <8 x i16> [[TMP8]]
; CHECK-NEXT: store <8 x i16> [[PREDPHI]], ptr [[DCT]], align 2, !alias.scope !0, !noalias !3
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 8
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 16
; CHECK-NEXT: [[WIDE_LOAD_1:%.*]] = load <8 x i16>, ptr [[TMP13]], align 2, !alias.scope !0, !noalias !3
; CHECK-NEXT: [[TMP14:%.*]] = sext <8 x i16> [[WIDE_LOAD_1]] to <8 x i32>
; CHECK-NEXT: [[TMP15:%.*]] = icmp sgt <8 x i16> [[WIDE_LOAD_1]], zeroinitializer
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 8
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 16
; CHECK-NEXT: [[WIDE_LOAD28_1:%.*]] = load <8 x i16>, ptr [[TMP16]], align 2, !alias.scope !6
; CHECK-NEXT: [[TMP17:%.*]] = zext <8 x i16> [[WIDE_LOAD28_1]] to <8 x i32>
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 8
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 16
; CHECK-NEXT: [[WIDE_LOAD29_1:%.*]] = load <8 x i16>, ptr [[TMP18]], align 2, !alias.scope !7
; CHECK-NEXT: [[TMP19:%.*]] = zext <8 x i16> [[WIDE_LOAD29_1]] to <8 x i32>
; CHECK-NEXT: [[TMP20:%.*]] = sub nsw <8 x i32> [[TMP17]], [[TMP14]]
@@ -95,14 +95,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK: if.end:
; CHECK-NEXT: [[STOREMERGE:%.*]] = phi i16 [ [[CONV28]], [[IF_ELSE]] ], [ [[CONV12]], [[IF_THEN]] ]
; CHECK-NEXT: store i16 [[STOREMERGE]], ptr [[DCT]], align 2
-; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 1
+; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 2
; CHECK-NEXT: [[TMP36:%.*]] = load i16, ptr [[ARRAYIDX_1]], align 2
; CHECK-NEXT: [[CONV_1:%.*]] = sext i16 [[TMP36]] to i32
; CHECK-NEXT: [[CMP1_1:%.*]] = icmp sgt i16 [[TMP36]], 0
-; CHECK-NEXT: [[ARRAYIDX4_1:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 1
+; CHECK-NEXT: [[ARRAYIDX4_1:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 2
; CHECK-NEXT: [[TMP37:%.*]] = load i16, ptr [[ARRAYIDX4_1]], align 2
; CHECK-NEXT: [[CONV5_1:%.*]] = zext i16 [[TMP37]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_1:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 1
+; CHECK-NEXT: [[ARRAYIDX10_1:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 2
; CHECK-NEXT: [[TMP38:%.*]] = load i16, ptr [[ARRAYIDX10_1]], align 2
; CHECK-NEXT: [[CONV11_1:%.*]] = zext i16 [[TMP38]] to i32
; CHECK-NEXT: br i1 [[CMP1_1]], label [[IF_THEN_1:%.*]], label [[IF_ELSE_1:%.*]]
@@ -123,14 +123,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_1:%.*]] = phi i16 [ [[CONV28_1]], [[IF_ELSE_1]] ], [ [[CONV12_1]], [[IF_THEN_1]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_1]], ptr [[ARRAYIDX_1]], align 2
; CHECK-NEXT: [[OR_131:%.*]] = or i16 [[STOREMERGE]], [[STOREMERGE_1]]
-; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 2
+; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 4
; CHECK-NEXT: [[TMP40:%.*]] = load i16, ptr [[ARRAYIDX_2]], align 2
; CHECK-NEXT: [[CONV_2:%.*]] = sext i16 [[TMP40]] to i32
; CHECK-NEXT: [[CMP1_2:%.*]] = icmp sgt i16 [[TMP40]], 0
-; CHECK-NEXT: [[ARRAYIDX4_2:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 2
+; CHECK-NEXT: [[ARRAYIDX4_2:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 4
; CHECK-NEXT: [[TMP41:%.*]] = load i16, ptr [[ARRAYIDX4_2]], align 2
; CHECK-NEXT: [[CONV5_2:%.*]] = zext i16 [[TMP41]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_2:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 2
+; CHECK-NEXT: [[ARRAYIDX10_2:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 4
; CHECK-NEXT: [[TMP42:%.*]] = load i16, ptr [[ARRAYIDX10_2]], align 2
; CHECK-NEXT: [[CONV11_2:%.*]] = zext i16 [[TMP42]] to i32
; CHECK-NEXT: br i1 [[CMP1_2]], label [[IF_THEN_2:%.*]], label [[IF_ELSE_2:%.*]]
@@ -151,14 +151,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_2:%.*]] = phi i16 [ [[CONV28_2]], [[IF_ELSE_2]] ], [ [[CONV12_2]], [[IF_THEN_2]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_2]], ptr [[ARRAYIDX_2]], align 2
; CHECK-NEXT: [[OR_232:%.*]] = or i16 [[OR_131]], [[STOREMERGE_2]]
-; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 3
+; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 6
; CHECK-NEXT: [[TMP44:%.*]] = load i16, ptr [[ARRAYIDX_3]], align 2
; CHECK-NEXT: [[CONV_3:%.*]] = sext i16 [[TMP44]] to i32
; CHECK-NEXT: [[CMP1_3:%.*]] = icmp sgt i16 [[TMP44]], 0
-; CHECK-NEXT: [[ARRAYIDX4_3:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 3
+; CHECK-NEXT: [[ARRAYIDX4_3:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 6
; CHECK-NEXT: [[TMP45:%.*]] = load i16, ptr [[ARRAYIDX4_3]], align 2
; CHECK-NEXT: [[CONV5_3:%.*]] = zext i16 [[TMP45]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_3:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 3
+; CHECK-NEXT: [[ARRAYIDX10_3:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 6
; CHECK-NEXT: [[TMP46:%.*]] = load i16, ptr [[ARRAYIDX10_3]], align 2
; CHECK-NEXT: [[CONV11_3:%.*]] = zext i16 [[TMP46]] to i32
; CHECK-NEXT: br i1 [[CMP1_3]], label [[IF_THEN_3:%.*]], label [[IF_ELSE_3:%.*]]
@@ -179,14 +179,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_3:%.*]] = phi i16 [ [[CONV28_3]], [[IF_ELSE_3]] ], [ [[CONV12_3]], [[IF_THEN_3]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_3]], ptr [[ARRAYIDX_3]], align 2
; CHECK-NEXT: [[OR_333:%.*]] = or i16 [[OR_232]], [[STOREMERGE_3]]
-; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 4
+; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 8
; CHECK-NEXT: [[TMP48:%.*]] = load i16, ptr [[ARRAYIDX_4]], align 2
; CHECK-NEXT: [[CONV_4:%.*]] = sext i16 [[TMP48]] to i32
; CHECK-NEXT: [[CMP1_4:%.*]] = icmp sgt i16 [[TMP48]], 0
-; CHECK-NEXT: [[ARRAYIDX4_4:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 4
+; CHECK-NEXT: [[ARRAYIDX4_4:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 8
; CHECK-NEXT: [[TMP49:%.*]] = load i16, ptr [[ARRAYIDX4_4]], align 2
; CHECK-NEXT: [[CONV5_4:%.*]] = zext i16 [[TMP49]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_4:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 4
+; CHECK-NEXT: [[ARRAYIDX10_4:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 8
; CHECK-NEXT: [[TMP50:%.*]] = load i16, ptr [[ARRAYIDX10_4]], align 2
; CHECK-NEXT: [[CONV11_4:%.*]] = zext i16 [[TMP50]] to i32
; CHECK-NEXT: br i1 [[CMP1_4]], label [[IF_THEN_4:%.*]], label [[IF_ELSE_4:%.*]]
@@ -207,14 +207,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_4:%.*]] = phi i16 [ [[CONV28_4]], [[IF_ELSE_4]] ], [ [[CONV12_4]], [[IF_THEN_4]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_4]], ptr [[ARRAYIDX_4]], align 2
; CHECK-NEXT: [[OR_434:%.*]] = or i16 [[OR_333]], [[STOREMERGE_4]]
-; CHECK-NEXT: [[ARRAYIDX_5:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 5
+; CHECK-NEXT: [[ARRAYIDX_5:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 10
; CHECK-NEXT: [[TMP52:%.*]] = load i16, ptr [[ARRAYIDX_5]], align 2
; CHECK-NEXT: [[CONV_5:%.*]] = sext i16 [[TMP52]] to i32
; CHECK-NEXT: [[CMP1_5:%.*]] = icmp sgt i16 [[TMP52]], 0
-; CHECK-NEXT: [[ARRAYIDX4_5:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 5
+; CHECK-NEXT: [[ARRAYIDX4_5:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 10
; CHECK-NEXT: [[TMP53:%.*]] = load i16, ptr [[ARRAYIDX4_5]], align 2
; CHECK-NEXT: [[CONV5_5:%.*]] = zext i16 [[TMP53]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_5:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 5
+; CHECK-NEXT: [[ARRAYIDX10_5:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 10
; CHECK-NEXT: [[TMP54:%.*]] = load i16, ptr [[ARRAYIDX10_5]], align 2
; CHECK-NEXT: [[CONV11_5:%.*]] = zext i16 [[TMP54]] to i32
; CHECK-NEXT: br i1 [[CMP1_5]], label [[IF_THEN_5:%.*]], label [[IF_ELSE_5:%.*]]
@@ -235,14 +235,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_5:%.*]] = phi i16 [ [[CONV28_5]], [[IF_ELSE_5]] ], [ [[CONV12_5]], [[IF_THEN_5]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_5]], ptr [[ARRAYIDX_5]], align 2
; CHECK-NEXT: [[OR_535:%.*]] = or i16 [[OR_434]], [[STOREMERGE_5]]
-; CHECK-NEXT: [[ARRAYIDX_6:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 6
+; CHECK-NEXT: [[ARRAYIDX_6:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 12
; CHECK-NEXT: [[TMP56:%.*]] = load i16, ptr [[ARRAYIDX_6]], align 2
; CHECK-NEXT: [[CONV_6:%.*]] = sext i16 [[TMP56]] to i32
; CHECK-NEXT: [[CMP1_6:%.*]] = icmp sgt i16 [[TMP56]], 0
-; CHECK-NEXT: [[ARRAYIDX4_6:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 6
+; CHECK-NEXT: [[ARRAYIDX4_6:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 12
; CHECK-NEXT: [[TMP57:%.*]] = load i16, ptr [[ARRAYIDX4_6]], align 2
; CHECK-NEXT: [[CONV5_6:%.*]] = zext i16 [[TMP57]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_6:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 6
+; CHECK-NEXT: [[ARRAYIDX10_6:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 12
; CHECK-NEXT: [[TMP58:%.*]] = load i16, ptr [[ARRAYIDX10_6]], align 2
; CHECK-NEXT: [[CONV11_6:%.*]] = zext i16 [[TMP58]] to i32
; CHECK-NEXT: br i1 [[CMP1_6]], label [[IF_THEN_6:%.*]], label [[IF_ELSE_6:%.*]]
@@ -263,14 +263,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_6:%.*]] = phi i16 [ [[CONV28_6]], [[IF_ELSE_6]] ], [ [[CONV12_6]], [[IF_THEN_6]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_6]], ptr [[ARRAYIDX_6]], align 2
; CHECK-NEXT: [[OR_636:%.*]] = or i16 [[OR_535]], [[STOREMERGE_6]]
-; CHECK-NEXT: [[ARRAYIDX_7:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 7
+; CHECK-NEXT: [[ARRAYIDX_7:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 14
; CHECK-NEXT: [[TMP60:%.*]] = load i16, ptr [[ARRAYIDX_7]], align 2
; CHECK-NEXT: [[CONV_7:%.*]] = sext i16 [[TMP60]] to i32
; CHECK-NEXT: [[CMP1_7:%.*]] = icmp sgt i16 [[TMP60]], 0
-; CHECK-NEXT: [[ARRAYIDX4_7:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 7
+; CHECK-NEXT: [[ARRAYIDX4_7:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 14
; CHECK-NEXT: [[TMP61:%.*]] = load i16, ptr [[ARRAYIDX4_7]], align 2
; CHECK-NEXT: [[CONV5_7:%.*]] = zext i16 [[TMP61]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_7:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 7
+; CHECK-NEXT: [[ARRAYIDX10_7:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 14
; CHECK-NEXT: [[TMP62:%.*]] = load i16, ptr [[ARRAYIDX10_7]], align 2
; CHECK-NEXT: [[CONV11_7:%.*]] = zext i16 [[TMP62]] to i32
; CHECK-NEXT: br i1 [[CMP1_7]], label [[IF_THEN_7:%.*]], label [[IF_ELSE_7:%.*]]
@@ -291,14 +291,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_7:%.*]] = phi i16 [ [[CONV28_7]], [[IF_ELSE_7]] ], [ [[CONV12_7]], [[IF_THEN_7]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_7]], ptr [[ARRAYIDX_7]], align 2
; CHECK-NEXT: [[OR_737:%.*]] = or i16 [[OR_636]], [[STOREMERGE_7]]
-; CHECK-NEXT: [[ARRAYIDX_8:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 8
+; CHECK-NEXT: [[ARRAYIDX_8:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 16
; CHECK-NEXT: [[TMP64:%.*]] = load i16, ptr [[ARRAYIDX_8]], align 2
; CHECK-NEXT: [[CONV_8:%.*]] = sext i16 [[TMP64]] to i32
; CHECK-NEXT: [[CMP1_8:%.*]] = icmp sgt i16 [[TMP64]], 0
-; CHECK-NEXT: [[ARRAYIDX4_8:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 8
+; CHECK-NEXT: [[ARRAYIDX4_8:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 16
; CHECK-NEXT: [[TMP65:%.*]] = load i16, ptr [[ARRAYIDX4_8]], align 2
; CHECK-NEXT: [[CONV5_8:%.*]] = zext i16 [[TMP65]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_8:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 8
+; CHECK-NEXT: [[ARRAYIDX10_8:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 16
; CHECK-NEXT: [[TMP66:%.*]] = load i16, ptr [[ARRAYIDX10_8]], align 2
; CHECK-NEXT: [[CONV11_8:%.*]] = zext i16 [[TMP66]] to i32
; CHECK-NEXT: br i1 [[CMP1_8]], label [[IF_THEN_8:%.*]], label [[IF_ELSE_8:%.*]]
@@ -319,14 +319,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_8:%.*]] = phi i16 [ [[CONV28_8]], [[IF_ELSE_8]] ], [ [[CONV12_8]], [[IF_THEN_8]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_8]], ptr [[ARRAYIDX_8]], align 2
; CHECK-NEXT: [[OR_838:%.*]] = or i16 [[OR_737]], [[STOREMERGE_8]]
-; CHECK-NEXT: [[ARRAYIDX_9:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 9
+; CHECK-NEXT: [[ARRAYIDX_9:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 18
; CHECK-NEXT: [[TMP68:%.*]] = load i16, ptr [[ARRAYIDX_9]], align 2
; CHECK-NEXT: [[CONV_9:%.*]] = sext i16 [[TMP68]] to i32
; CHECK-NEXT: [[CMP1_9:%.*]] = icmp sgt i16 [[TMP68]], 0
-; CHECK-NEXT: [[ARRAYIDX4_9:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 9
+; CHECK-NEXT: [[ARRAYIDX4_9:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 18
; CHECK-NEXT: [[TMP69:%.*]] = load i16, ptr [[ARRAYIDX4_9]], align 2
; CHECK-NEXT: [[CONV5_9:%.*]] = zext i16 [[TMP69]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_9:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 9
+; CHECK-NEXT: [[ARRAYIDX10_9:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 18
; CHECK-NEXT: [[TMP70:%.*]] = load i16, ptr [[ARRAYIDX10_9]], align 2
; CHECK-NEXT: [[CONV11_9:%.*]] = zext i16 [[TMP70]] to i32
; CHECK-NEXT: br i1 [[CMP1_9]], label [[IF_THEN_9:%.*]], label [[IF_ELSE_9:%.*]]
@@ -347,14 +347,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_9:%.*]] = phi i16 [ [[CONV28_9]], [[IF_ELSE_9]] ], [ [[CONV12_9]], [[IF_THEN_9]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_9]], ptr [[ARRAYIDX_9]], align 2
; CHECK-NEXT: [[OR_939:%.*]] = or i16 [[OR_838]], [[STOREMERGE_9]]
-; CHECK-NEXT: [[ARRAYIDX_10:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 10
+; CHECK-NEXT: [[ARRAYIDX_10:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 20
; CHECK-NEXT: [[TMP72:%.*]] = load i16, ptr [[ARRAYIDX_10]], align 2
; CHECK-NEXT: [[CONV_10:%.*]] = sext i16 [[TMP72]] to i32
; CHECK-NEXT: [[CMP1_10:%.*]] = icmp sgt i16 [[TMP72]], 0
-; CHECK-NEXT: [[ARRAYIDX4_10:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 10
+; CHECK-NEXT: [[ARRAYIDX4_10:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 20
; CHECK-NEXT: [[TMP73:%.*]] = load i16, ptr [[ARRAYIDX4_10]], align 2
; CHECK-NEXT: [[CONV5_10:%.*]] = zext i16 [[TMP73]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_10:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 10
+; CHECK-NEXT: [[ARRAYIDX10_10:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 20
; CHECK-NEXT: [[TMP74:%.*]] = load i16, ptr [[ARRAYIDX10_10]], align 2
; CHECK-NEXT: [[CONV11_10:%.*]] = zext i16 [[TMP74]] to i32
; CHECK-NEXT: br i1 [[CMP1_10]], label [[IF_THEN_10:%.*]], label [[IF_ELSE_10:%.*]]
@@ -375,14 +375,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_10:%.*]] = phi i16 [ [[CONV28_10]], [[IF_ELSE_10]] ], [ [[CONV12_10]], [[IF_THEN_10]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_10]], ptr [[ARRAYIDX_10]], align 2
; CHECK-NEXT: [[OR_1040:%.*]] = or i16 [[OR_939]], [[STOREMERGE_10]]
-; CHECK-NEXT: [[ARRAYIDX_11:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 11
+; CHECK-NEXT: [[ARRAYIDX_11:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 22
; CHECK-NEXT: [[TMP76:%.*]] = load i16, ptr [[ARRAYIDX_11]], align 2
; CHECK-NEXT: [[CONV_11:%.*]] = sext i16 [[TMP76]] to i32
; CHECK-NEXT: [[CMP1_11:%.*]] = icmp sgt i16 [[TMP76]], 0
-; CHECK-NEXT: [[ARRAYIDX4_11:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 11
+; CHECK-NEXT: [[ARRAYIDX4_11:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 22
; CHECK-NEXT: [[TMP77:%.*]] = load i16, ptr [[ARRAYIDX4_11]], align 2
; CHECK-NEXT: [[CONV5_11:%.*]] = zext i16 [[TMP77]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_11:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 11
+; CHECK-NEXT: [[ARRAYIDX10_11:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 22
; CHECK-NEXT: [[TMP78:%.*]] = load i16, ptr [[ARRAYIDX10_11]], align 2
; CHECK-NEXT: [[CONV11_11:%.*]] = zext i16 [[TMP78]] to i32
; CHECK-NEXT: br i1 [[CMP1_11]], label [[IF_THEN_11:%.*]], label [[IF_ELSE_11:%.*]]
@@ -403,14 +403,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_11:%.*]] = phi i16 [ [[CONV28_11]], [[IF_ELSE_11]] ], [ [[CONV12_11]], [[IF_THEN_11]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_11]], ptr [[ARRAYIDX_11]], align 2
; CHECK-NEXT: [[OR_1141:%.*]] = or i16 [[OR_1040]], [[STOREMERGE_11]]
-; CHECK-NEXT: [[ARRAYIDX_12:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 12
+; CHECK-NEXT: [[ARRAYIDX_12:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 24
; CHECK-NEXT: [[TMP80:%.*]] = load i16, ptr [[ARRAYIDX_12]], align 2
; CHECK-NEXT: [[CONV_12:%.*]] = sext i16 [[TMP80]] to i32
; CHECK-NEXT: [[CMP1_12:%.*]] = icmp sgt i16 [[TMP80]], 0
-; CHECK-NEXT: [[ARRAYIDX4_12:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 12
+; CHECK-NEXT: [[ARRAYIDX4_12:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 24
; CHECK-NEXT: [[TMP81:%.*]] = load i16, ptr [[ARRAYIDX4_12]], align 2
; CHECK-NEXT: [[CONV5_12:%.*]] = zext i16 [[TMP81]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_12:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 12
+; CHECK-NEXT: [[ARRAYIDX10_12:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 24
; CHECK-NEXT: [[TMP82:%.*]] = load i16, ptr [[ARRAYIDX10_12]], align 2
; CHECK-NEXT: [[CONV11_12:%.*]] = zext i16 [[TMP82]] to i32
; CHECK-NEXT: br i1 [[CMP1_12]], label [[IF_THEN_12:%.*]], label [[IF_ELSE_12:%.*]]
@@ -431,14 +431,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_12:%.*]] = phi i16 [ [[CONV28_12]], [[IF_ELSE_12]] ], [ [[CONV12_12]], [[IF_THEN_12]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_12]], ptr [[ARRAYIDX_12]], align 2
; CHECK-NEXT: [[OR_1242:%.*]] = or i16 [[OR_1141]], [[STOREMERGE_12]]
-; CHECK-NEXT: [[ARRAYIDX_13:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 13
+; CHECK-NEXT: [[ARRAYIDX_13:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 26
; CHECK-NEXT: [[TMP84:%.*]] = load i16, ptr [[ARRAYIDX_13]], align 2
; CHECK-NEXT: [[CONV_13:%.*]] = sext i16 [[TMP84]] to i32
; CHECK-NEXT: [[CMP1_13:%.*]] = icmp sgt i16 [[TMP84]], 0
-; CHECK-NEXT: [[ARRAYIDX4_13:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 13
+; CHECK-NEXT: [[ARRAYIDX4_13:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 26
; CHECK-NEXT: [[TMP85:%.*]] = load i16, ptr [[ARRAYIDX4_13]], align 2
; CHECK-NEXT: [[CONV5_13:%.*]] = zext i16 [[TMP85]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_13:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 13
+; CHECK-NEXT: [[ARRAYIDX10_13:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 26
; CHECK-NEXT: [[TMP86:%.*]] = load i16, ptr [[ARRAYIDX10_13]], align 2
; CHECK-NEXT: [[CONV11_13:%.*]] = zext i16 [[TMP86]] to i32
; CHECK-NEXT: br i1 [[CMP1_13]], label [[IF_THEN_13:%.*]], label [[IF_ELSE_13:%.*]]
@@ -459,14 +459,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_13:%.*]] = phi i16 [ [[CONV28_13]], [[IF_ELSE_13]] ], [ [[CONV12_13]], [[IF_THEN_13]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_13]], ptr [[ARRAYIDX_13]], align 2
; CHECK-NEXT: [[OR_1343:%.*]] = or i16 [[OR_1242]], [[STOREMERGE_13]]
-; CHECK-NEXT: [[ARRAYIDX_14:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 14
+; CHECK-NEXT: [[ARRAYIDX_14:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 28
; CHECK-NEXT: [[TMP88:%.*]] = load i16, ptr [[ARRAYIDX_14]], align 2
; CHECK-NEXT: [[CONV_14:%.*]] = sext i16 [[TMP88]] to i32
; CHECK-NEXT: [[CMP1_14:%.*]] = icmp sgt i16 [[TMP88]], 0
-; CHECK-NEXT: [[ARRAYIDX4_14:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 14
+; CHECK-NEXT: [[ARRAYIDX4_14:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 28
; CHECK-NEXT: [[TMP89:%.*]] = load i16, ptr [[ARRAYIDX4_14]], align 2
; CHECK-NEXT: [[CONV5_14:%.*]] = zext i16 [[TMP89]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_14:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 14
+; CHECK-NEXT: [[ARRAYIDX10_14:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 28
; CHECK-NEXT: [[TMP90:%.*]] = load i16, ptr [[ARRAYIDX10_14]], align 2
; CHECK-NEXT: [[CONV11_14:%.*]] = zext i16 [[TMP90]] to i32
; CHECK-NEXT: br i1 [[CMP1_14]], label [[IF_THEN_14:%.*]], label [[IF_ELSE_14:%.*]]
@@ -487,14 +487,14 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: [[STOREMERGE_14:%.*]] = phi i16 [ [[CONV28_14]], [[IF_ELSE_14]] ], [ [[CONV12_14]], [[IF_THEN_14]] ]
; CHECK-NEXT: store i16 [[STOREMERGE_14]], ptr [[ARRAYIDX_14]], align 2
; CHECK-NEXT: [[OR_1444:%.*]] = or i16 [[OR_1343]], [[STOREMERGE_14]]
-; CHECK-NEXT: [[ARRAYIDX_15:%.*]] = getelementptr inbounds i16, ptr [[DCT]], i64 15
+; CHECK-NEXT: [[ARRAYIDX_15:%.*]] = getelementptr inbounds i8, ptr [[DCT]], i64 30
; CHECK-NEXT: [[TMP92:%.*]] = load i16, ptr [[ARRAYIDX_15]], align 2
; CHECK-NEXT: [[CONV_15:%.*]] = sext i16 [[TMP92]] to i32
; CHECK-NEXT: [[CMP1_15:%.*]] = icmp sgt i16 [[TMP92]], 0
-; CHECK-NEXT: [[ARRAYIDX4_15:%.*]] = getelementptr inbounds i16, ptr [[BIAS]], i64 15
+; CHECK-NEXT: [[ARRAYIDX4_15:%.*]] = getelementptr inbounds i8, ptr [[BIAS]], i64 30
; CHECK-NEXT: [[TMP93:%.*]] = load i16, ptr [[ARRAYIDX4_15]], align 2
; CHECK-NEXT: [[CONV5_15:%.*]] = zext i16 [[TMP93]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_15:%.*]] = getelementptr inbounds i16, ptr [[MF]], i64 15
+; CHECK-NEXT: [[ARRAYIDX10_15:%.*]] = getelementptr inbounds i8, ptr [[MF]], i64 30
; CHECK-NEXT: [[TMP94:%.*]] = load i16, ptr [[ARRAYIDX10_15]], align 2
; CHECK-NEXT: [[CONV11_15:%.*]] = zext i16 [[TMP94]] to i32
; CHECK-NEXT: br i1 [[CMP1_15]], label [[IF_THEN_15:%.*]], label [[IF_ELSE_15:%.*]]
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/sinking-vs-if-conversion.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/sinking-vs-if-conversion.ll
index 06d7b36509a52f..885529926fc4b9 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/sinking-vs-if-conversion.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/sinking-vs-if-conversion.ll
@@ -12,13 +12,13 @@ define void @test_find_min(ptr noundef nonnull align 8 dereferenceable(24) %this
; CHECK-LABEL: define void @test_find_min(
; CHECK-SAME: ptr nocapture noundef nonnull align 8 dereferenceable(24) [[THIS:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[NUM_:%.*]] = getelementptr inbounds [[STRUCT_TREE:%.*]], ptr [[THIS]], i64 0, i32 2
+; CHECK-NEXT: [[NUM_:%.*]] = getelementptr inbounds i8, ptr [[THIS]], i64 16
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[NUM_]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[THIS]], align 8
; CHECK-NEXT: [[CMP8:%.*]] = icmp sgt i32 [[TMP0]], 0
; CHECK-NEXT: br i1 [[CMP8]], label [[FOR_BODY_LR_PH:%.*]], label [[FOR_END:%.*]]
; CHECK: for.body.lr.ph:
-; CHECK-NEXT: [[ARRAY_:%.*]] = getelementptr inbounds [[STRUCT_TREE]], ptr [[THIS]], i64 0, i32 1
+; CHECK-NEXT: [[ARRAY_:%.*]] = getelementptr inbounds i8, ptr [[THIS]], i64 8
; CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[ARRAY_]], align 8
; CHECK-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext nneg i32 [[TMP0]] to i64
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
@@ -30,9 +30,9 @@ define void @test_find_min(ptr noundef nonnull align 8 dereferenceable(24) %this
; CHECK-NEXT: [[CMP3:%.*]] = icmp eq ptr [[MIN_010]], null
; CHECK-NEXT: br i1 [[CMP3]], label [[COND_END7]], label [[COND_FALSE:%.*]]
; CHECK: cond.false:
-; CHECK-NEXT: [[KEY2:%.*]] = getelementptr inbounds [[STRUCT_NODE:%.*]], ptr [[MIN_010]], i64 0, i32 1
+; CHECK-NEXT: [[KEY2:%.*]] = getelementptr inbounds i8, ptr [[MIN_010]], i64 4
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[KEY2]], align 4
-; CHECK-NEXT: [[KEY:%.*]] = getelementptr inbounds [[STRUCT_NODE]], ptr [[TMP3]], i64 0, i32 1
+; CHECK-NEXT: [[KEY:%.*]] = getelementptr inbounds i8, ptr [[TMP3]], i64 4
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[KEY]], align 4
; CHECK-NEXT: [[CMP4:%.*]] = icmp slt i32 [[TMP5]], [[TMP4]]
; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP4]], ptr [[TMP3]], ptr [[MIN_010]]
diff --git a/llvm/test/Transforms/PhaseOrdering/ARM/arm_mult_q15.ll b/llvm/test/Transforms/PhaseOrdering/ARM/arm_mult_q15.ll
index 7cbce461a49240..73bcee5fb74f0c 100644
--- a/llvm/test/Transforms/PhaseOrdering/ARM/arm_mult_q15.ll
+++ b/llvm/test/Transforms/PhaseOrdering/ARM/arm_mult_q15.ll
@@ -60,21 +60,21 @@ define void @arm_mult_q15(ptr %pSrcA, ptr %pSrcB, ptr noalias %pDst, i32 %blockS
; CHECK-NEXT: [[PSRCA_ADDR_05:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[PSRCA_ADDR_05_PH]], [[WHILE_BODY_PREHEADER16]] ]
; CHECK-NEXT: [[PDST_ADDR_04:%.*]] = phi ptr [ [[INCDEC_PTR4:%.*]], [[WHILE_BODY]] ], [ [[PDST_ADDR_04_PH]], [[WHILE_BODY_PREHEADER16]] ]
; CHECK-NEXT: [[PSRCB_ADDR_03:%.*]] = phi ptr [ [[INCDEC_PTR1:%.*]], [[WHILE_BODY]] ], [ [[PSRCB_ADDR_03_PH]], [[WHILE_BODY_PREHEADER16]] ]
-; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i16, ptr [[PSRCA_ADDR_05]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRCA_ADDR_05]], i32 2
; CHECK-NEXT: [[TMP13:%.*]] = load i16, ptr [[PSRCA_ADDR_05]], align 2
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP13]] to i32
-; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i16, ptr [[PSRCB_ADDR_03]], i32 1
+; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i8, ptr [[PSRCB_ADDR_03]], i32 2
; CHECK-NEXT: [[TMP14:%.*]] = load i16, ptr [[PSRCB_ADDR_03]], align 2
; CHECK-NEXT: [[CONV2:%.*]] = sext i16 [[TMP14]] to i32
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV2]], [[CONV]]
; CHECK-NEXT: [[SHR:%.*]] = ashr i32 [[MUL]], 15
-; CHECK-NEXT: [[TMP15:%.*]] = tail call i32 @llvm.smin.i32(i32 [[SHR]], i32 32767)
-; CHECK-NEXT: [[CONV3:%.*]] = trunc i32 [[TMP15]] to i16
-; CHECK-NEXT: [[INCDEC_PTR4]] = getelementptr inbounds i16, ptr [[PDST_ADDR_04]], i32 1
+; CHECK-NEXT: [[SPEC_SELECT_I:%.*]] = tail call i32 @llvm.smin.i32(i32 [[SHR]], i32 32767)
+; CHECK-NEXT: [[CONV3:%.*]] = trunc i32 [[SPEC_SELECT_I]] to i16
+; CHECK-NEXT: [[INCDEC_PTR4]] = getelementptr inbounds i8, ptr [[PDST_ADDR_04]], i32 2
; CHECK-NEXT: store i16 [[CONV3]], ptr [[PDST_ADDR_04]], align 2
; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_06]], -1
; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0
-; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP_NOT]], label [[WHILE_END]], label [[WHILE_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: while.end:
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll b/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll
index 8f37e79539b8c7..fd6bb61d0369d0 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll
@@ -180,17 +180,17 @@ define void @test_runtime_trip_count(i32 %N) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [58 x double], ptr @b, i64 0, i64 [[INDEX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr [[TMP0]], i64 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP0]], align 16
; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <2 x double>, ptr [[TMP1]], align 16
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [58 x double], ptr @c, i64 0, i64 [[INDEX]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds double, ptr [[TMP2]], i64 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 16
; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <2 x double>, ptr [[TMP2]], align 16
; CHECK-NEXT: [[WIDE_LOAD6:%.*]] = load <2 x double>, ptr [[TMP3]], align 16
; CHECK-NEXT: [[TMP4:%.*]] = fadd <2 x double> [[WIDE_LOAD]], [[WIDE_LOAD5]]
; CHECK-NEXT: [[TMP5:%.*]] = fadd <2 x double> [[WIDE_LOAD4]], [[WIDE_LOAD6]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [58 x double], ptr @a, i64 0, i64 [[INDEX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds double, ptr [[TMP6]], i64 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP6]], i64 16
; CHECK-NEXT: store <2 x double> [[TMP4]], ptr [[TMP6]], align 16
; CHECK-NEXT: store <2 x double> [[TMP5]], ptr [[TMP7]], align 16
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/hoist-load-of-baseptr.ll b/llvm/test/Transforms/PhaseOrdering/X86/hoist-load-of-baseptr.ll
index eb8fe102745213..f0a18463895806 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/hoist-load-of-baseptr.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/hoist-load-of-baseptr.ll
@@ -56,7 +56,7 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O2: vector.body:
; O2-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ], [ 0, [[FOR_BODY4_PREHEADER]] ]
; O2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 [[INDEX]]
-; O2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; O2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; O2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
; O2-NEXT: [[WIDE_LOAD8:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !tbaa [[TBAA0]]
; O2-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], <i32 1, i32 1, i32 1, i32 1>
@@ -76,7 +76,7 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O2: for.cond.cleanup3:
; O2-NEXT: [[INC7]] = add nuw nsw i64 [[I_06]], 1
; O2-NEXT: [[EXITCOND7_NOT:%.*]] = icmp eq i64 [[INC7]], 100
-; O2-NEXT: br i1 [[EXITCOND7_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP7:![0-9]+]]
+; O2-NEXT: br i1 [[EXITCOND7_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP8:![0-9]+]]
; O2: for.body4:
; O2-NEXT: [[J_05:%.*]] = phi i64 [ [[INC5:%.*]], [[FOR_BODY4]] ], [ [[J_05_PH]], [[FOR_BODY4_PREHEADER9]] ]
; O2-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 [[J_05]]
@@ -85,7 +85,7 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O2-NEXT: store i32 [[INC]], ptr [[ADD_PTR_I]], align 4, !tbaa [[TBAA0]]
; O2-NEXT: [[INC5]] = add nuw i64 [[J_05]], 1
; O2-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC5]], [[NUMELEMS]]
-; O2-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP3]], label [[FOR_BODY4]], !llvm.loop [[LOOP8:![0-9]+]]
+; O2-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP3]], label [[FOR_BODY4]], !llvm.loop [[LOOP9:![0-9]+]]
;
; O3-LABEL: define {{[^@]+}}@_Z7computeRSt6vectorIiSaIiEEy
; O3-SAME: (ptr nocapture noundef nonnull readonly align 8 dereferenceable(24) [[DATA:%.*]], i64 noundef [[NUMELEMS:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
@@ -104,7 +104,7 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O3: vector.body:
; O3-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ], [ 0, [[FOR_COND1_PREHEADER_US]] ]
; O3-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 [[INDEX]]
-; O3-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; O3-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; O3-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
; O3-NEXT: [[WIDE_LOAD9:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !tbaa [[TBAA0]]
; O3-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], <i32 1, i32 1, i32 1, i32 1>
@@ -127,7 +127,7 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O3-NEXT: store i32 [[INC_US]], ptr [[ADD_PTR_I_US]], align 4, !tbaa [[TBAA0]]
; O3-NEXT: [[INC5_US]] = add nuw i64 [[J_05_US]], 1
; O3-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC5_US]], [[NUMELEMS]]
-; O3-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND1_FOR_COND_CLEANUP3_CRIT_EDGE_US]], label [[FOR_BODY4_US]], !llvm.loop [[LOOP7:![0-9]+]]
+; O3-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND1_FOR_COND_CLEANUP3_CRIT_EDGE_US]], label [[FOR_BODY4_US]], !llvm.loop [[LOOP8:![0-9]+]]
; O3: for.cond1.for.cond.cleanup3_crit_edge.us:
; O3-NEXT: [[INC7_US]] = add nuw nsw i64 [[I_06_US]], 1
; O3-NEXT: [[EXITCOND8_NOT:%.*]] = icmp eq i64 [[INC7_US]], 100
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll b/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
index c50f164046301c..05e39efc7745ae 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
@@ -43,7 +43,7 @@ define void @loop_or(ptr noalias %pIn, ptr noalias %pOut, i32 %s) {
; CHECK-NEXT: [[TMP6:%.*]] = or disjoint <4 x i32> [[TMP4]], <i32 -16777216, i32 -16777216, i32 -16777216, i32 -16777216>
; CHECK-NEXT: [[TMP7:%.*]] = or disjoint <4 x i32> [[TMP5]], <i32 -16777216, i32 -16777216, i32 -16777216, i32 -16777216>
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[POUT:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP8]], i64 4
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP8]], i64 16
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr [[TMP8]], align 4
; CHECK-NEXT: store <4 x i32> [[TMP7]], ptr [[TMP9]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll b/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll
index b52dd6743649a3..40c42ffdfd1079 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pr48844-br-to-switch-vectorization.ll
@@ -23,7 +23,7 @@ define dso_local void @test(ptr %start, ptr %end) #0 {
; CHECK-NEXT: store i32 42, ptr [[PTR2]], align 4
; CHECK-NEXT: br label [[LATCH]]
; CHECK: latch:
-; CHECK-NEXT: [[PTR_NEXT]] = getelementptr inbounds i32, ptr [[PTR2]], i64 1
+; CHECK-NEXT: [[PTR_NEXT]] = getelementptr inbounds i8, ptr [[PTR2]], i64 4
; CHECK-NEXT: [[I11_NOT:%.*]] = icmp eq ptr [[PTR_NEXT]], [[END]]
; CHECK-NEXT: br i1 [[I11_NOT]], label [[EXIT]], label [[BB12]]
; CHECK: exit:
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pr50555.ll b/llvm/test/Transforms/PhaseOrdering/X86/pr50555.ll
index 18d1c642685506..d9101272ba3d0f 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pr50555.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pr50555.ll
@@ -11,7 +11,7 @@ define void @trunc_through_one_add(ptr noalias %0, ptr noalias readonly %1) {
; SSE-NEXT: [[TMP7:%.*]] = lshr <8 x i16> [[TMP6]], <i16 2, i16 2, i16 2, i16 2, i16 2, i16 2, i16 2, i16 2>
; SSE-NEXT: store <8 x i16> [[TMP7]], ptr [[TMP0:%.*]], align 2
; SSE-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 8
-; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i16, ptr [[TMP0]], i64 8
+; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; SSE-NEXT: [[TMP10:%.*]] = load <8 x i8>, ptr [[TMP8]], align 1
; SSE-NEXT: [[TMP11:%.*]] = zext <8 x i8> [[TMP10]] to <8 x i16>
; SSE-NEXT: [[TMP12:%.*]] = lshr <8 x i16> [[TMP11]], <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
@@ -187,7 +187,7 @@ define void @trunc_through_two_adds(ptr noalias %0, ptr noalias readonly %1, ptr
; SSE-NEXT: store <8 x i16> [[TMP11]], ptr [[TMP0:%.*]], align 2
; SSE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 8
; SSE-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i64 8
-; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i16, ptr [[TMP0]], i64 8
+; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; SSE-NEXT: [[TMP15:%.*]] = load <8 x i8>, ptr [[TMP12]], align 1
; SSE-NEXT: [[TMP16:%.*]] = zext <8 x i8> [[TMP15]] to <8 x i16>
; SSE-NEXT: [[TMP17:%.*]] = load <8 x i8>, ptr [[TMP13]], align 1
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll b/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll
index 24cc18625af821..c5deb716d80302 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll
@@ -40,7 +40,7 @@ define void @licm(ptr align 8 dereferenceable(8) %_M_start.i, i64 %numElem) {
; O23: vector.body:
; O23-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; O23-NEXT: [[TMP1:%.*]] = getelementptr inbounds double, ptr [[TMP0]], i64 [[INDEX]]
-; O23-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[TMP1]], i64 2
+; O23-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; O23-NEXT: store <2 x double> <double 2.000000e+00, double 2.000000e+00>, ptr [[TMP1]], align 8, !tbaa [[TBAA8:![0-9]+]]
; O23-NEXT: store <2 x double> <double 2.000000e+00, double 2.000000e+00>, ptr [[TMP2]], align 8, !tbaa [[TBAA8]]
; O23-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll b/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll
index 57951d7754269a..e645a309b47930 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll
@@ -16,14 +16,14 @@ define dso_local void @_Z13vecIncFromPtrP12FloatVecPair(ptr %FVP) {
; O1-LABEL: define {{[^@]+}}@_Z13vecIncFromPtrP12FloatVecPair
; O1-SAME: (ptr nocapture readonly [[FVP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; O1-NEXT: entry:
-; O1-NEXT: [[VSRC23_I:%.*]] = getelementptr inbounds [[CLASS_FLOATVECPAIR:%.*]], ptr [[FVP]], i64 0, i32 1
+; O1-NEXT: [[VSRC23_I:%.*]] = getelementptr inbounds i8, ptr [[FVP]], i64 16
; O1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[VSRC23_I]], align 8, !tbaa [[TBAA0:![0-9]+]]
-; O1-NEXT: [[SIZE4_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0:%.*]], ptr [[TMP0]], i64 undef, i32 1
+; O1-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0:%.*]], ptr [[TMP0]], i64 undef
+; O1-NEXT: [[SIZE4_I:%.*]] = getelementptr inbounds i8, ptr [[ARRAYIDX_I_I]], i64 8
; O1-NEXT: [[TMP1:%.*]] = load i32, ptr [[SIZE4_I]], align 8, !tbaa [[TBAA6:![0-9]+]]
; O1-NEXT: [[CMP56_NOT_I:%.*]] = icmp eq i32 [[TMP1]], 0
; O1-NEXT: br i1 [[CMP56_NOT_I]], label [[_ZN12FLOATVECPAIR6VECINCEV_EXIT:%.*]], label [[FOR_BODY7_LR_PH_I:%.*]]
; O1: for.body7.lr.ph.i:
-; O1-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0]], ptr [[TMP0]], i64 undef
; O1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX_I_I]], align 8, !tbaa [[TBAA8:![0-9]+]]
; O1-NEXT: [[ARRAYIDX_I3_I:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 undef
; O1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[FVP]], align 8, !tbaa [[TBAA0]]
@@ -46,14 +46,14 @@ define dso_local void @_Z13vecIncFromPtrP12FloatVecPair(ptr %FVP) {
; O23-LABEL: define {{[^@]+}}@_Z13vecIncFromPtrP12FloatVecPair
; O23-SAME: (ptr nocapture readonly [[FVP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; O23-NEXT: entry:
-; O23-NEXT: [[VSRC23_I:%.*]] = getelementptr inbounds [[CLASS_FLOATVECPAIR:%.*]], ptr [[FVP]], i64 0, i32 1
+; O23-NEXT: [[VSRC23_I:%.*]] = getelementptr inbounds i8, ptr [[FVP]], i64 16
; O23-NEXT: [[TMP0:%.*]] = load ptr, ptr [[VSRC23_I]], align 8, !tbaa [[TBAA0:![0-9]+]]
-; O23-NEXT: [[SIZE4_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0:%.*]], ptr [[TMP0]], i64 undef, i32 1
+; O23-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0:%.*]], ptr [[TMP0]], i64 undef
+; O23-NEXT: [[SIZE4_I:%.*]] = getelementptr inbounds i8, ptr [[ARRAYIDX_I_I]], i64 8
; O23-NEXT: [[TMP1:%.*]] = load i32, ptr [[SIZE4_I]], align 8, !tbaa [[TBAA6:![0-9]+]]
; O23-NEXT: [[CMP56_NOT_I:%.*]] = icmp eq i32 [[TMP1]], 0
; O23-NEXT: br i1 [[CMP56_NOT_I]], label [[_ZN12FLOATVECPAIR6VECINCEV_EXIT:%.*]], label [[FOR_BODY7_LR_PH_I:%.*]]
; O23: for.body7.lr.ph.i:
-; O23-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0]], ptr [[TMP0]], i64 undef
; O23-NEXT: [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX_I_I]], align 8, !tbaa [[TBAA8:![0-9]+]]
; O23-NEXT: [[ARRAYIDX_I3_I:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 undef
; O23-NEXT: [[TMP3:%.*]] = load ptr, ptr [[FVP]], align 8, !tbaa [[TBAA0]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll b/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
index 5053fd64f59cac..87d3900765490f 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
@@ -37,9 +37,9 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds double, ptr [[TMP5]], i64 4
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds double, ptr [[TMP5]], i64 8
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds double, ptr [[TMP5]], i64 12
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i64 32
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i64 64
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i64 96
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP5]], align 8, !tbaa [[TBAA3:![0-9]+]]
; CHECK-NEXT: [[WIDE_LOAD6:%.*]] = load <4 x double>, ptr [[TMP6]], align 8, !tbaa [[TBAA3]]
; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x double>, ptr [[TMP7]], align 8, !tbaa [[TBAA3]]
@@ -49,9 +49,9 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: [[TMP11:%.*]] = fmul fast <4 x double> [[WIDE_LOAD7]], [[TMP3]]
; CHECK-NEXT: [[TMP12:%.*]] = fmul fast <4 x double> [[WIDE_LOAD8]], [[TMP4]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds double, ptr [[TMP13]], i64 4
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds double, ptr [[TMP13]], i64 8
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds double, ptr [[TMP13]], i64 12
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP13]], i64 32
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP13]], i64 64
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP13]], i64 96
; CHECK-NEXT: store <4 x double> [[TMP9]], ptr [[TMP13]], align 8, !tbaa [[TBAA3]]
; CHECK-NEXT: store <4 x double> [[TMP10]], ptr [[TMP14]], align 8, !tbaa [[TBAA3]]
; CHECK-NEXT: store <4 x double> [[TMP11]], ptr [[TMP15]], align 8, !tbaa [[TBAA3]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/vec-shift.ll b/llvm/test/Transforms/PhaseOrdering/X86/vec-shift.ll
index ba7f8009349ade..7ee9812b3e74cc 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/vec-shift.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/vec-shift.ll
@@ -37,17 +37,17 @@ define void @bar(ptr noundef %0) {
; SSE-NEXT: [[TMP3:%.*]] = shl <2 x i64> [[TMP2]], <i64 44, i64 44>
; SSE-NEXT: [[TMP4:%.*]] = sub nuw nsw <2 x i64> <i64 -17592186044416, i64 -17592186044416>, [[TMP3]]
; SSE-NEXT: store <2 x i64> [[TMP4]], ptr [[TMP0]], align 8
-; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 2
+; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; SSE-NEXT: [[TMP6:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
; SSE-NEXT: [[TMP7:%.*]] = shl <2 x i64> [[TMP6]], <i64 44, i64 44>
; SSE-NEXT: [[TMP8:%.*]] = sub nuw nsw <2 x i64> <i64 -17592186044416, i64 -17592186044416>, [[TMP7]]
; SSE-NEXT: store <2 x i64> [[TMP8]], ptr [[TMP5]], align 8
-; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 4
+; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 32
; SSE-NEXT: [[TMP10:%.*]] = load <2 x i64>, ptr [[TMP9]], align 8
; SSE-NEXT: [[TMP11:%.*]] = shl <2 x i64> [[TMP10]], <i64 44, i64 44>
; SSE-NEXT: [[TMP12:%.*]] = sub nuw nsw <2 x i64> <i64 -17592186044416, i64 -17592186044416>, [[TMP11]]
; SSE-NEXT: store <2 x i64> [[TMP12]], ptr [[TMP9]], align 8
-; SSE-NEXT: [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 6
+; SSE-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 48
; SSE-NEXT: [[TMP14:%.*]] = load <2 x i64>, ptr [[TMP13]], align 8
; SSE-NEXT: [[TMP15:%.*]] = shl <2 x i64> [[TMP14]], <i64 44, i64 44>
; SSE-NEXT: [[TMP16:%.*]] = sub nuw nsw <2 x i64> <i64 -17592186044416, i64 -17592186044416>, [[TMP15]]
@@ -59,7 +59,7 @@ define void @bar(ptr noundef %0) {
; AVX-NEXT: [[TMP3:%.*]] = shl <4 x i64> [[TMP2]], <i64 44, i64 44, i64 44, i64 44>
; AVX-NEXT: [[TMP4:%.*]] = sub nuw nsw <4 x i64> <i64 -17592186044416, i64 -17592186044416, i64 -17592186044416, i64 -17592186044416>, [[TMP3]]
; AVX-NEXT: store <4 x i64> [[TMP4]], ptr [[TMP0]], align 8
-; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 4
+; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 32
; AVX-NEXT: [[TMP6:%.*]] = load <4 x i64>, ptr [[TMP5]], align 8
; AVX-NEXT: [[TMP7:%.*]] = shl <4 x i64> [[TMP6]], <i64 44, i64 44, i64 44, i64 44>
; AVX-NEXT: [[TMP8:%.*]] = sub nuw nsw <4 x i64> <i64 -17592186044416, i64 -17592186044416, i64 -17592186044416, i64 -17592186044416>, [[TMP7]]
diff --git a/llvm/test/Transforms/PhaseOrdering/basic.ll b/llvm/test/Transforms/PhaseOrdering/basic.ll
index 43d2e522e96e37..343afedff1f852 100644
--- a/llvm/test/Transforms/PhaseOrdering/basic.ll
+++ b/llvm/test/Transforms/PhaseOrdering/basic.ll
@@ -32,10 +32,10 @@ define void @test1() nounwind ssp {
define i32 @test2(i32 %a, ptr %p) nounwind uwtable ssp {
; CHECK-LABEL: @test2(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[DIV:%.*]] = lshr i32 [[A:%.*]], 2
-; CHECK-NEXT: store i32 [[DIV]], ptr [[P:%.*]], align 4
-; CHECK-NEXT: [[ADD:%.*]] = shl nuw nsw i32 [[DIV]], 1
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 1
+; CHECK-NEXT: [[DIV1:%.*]] = lshr i32 [[A:%.*]], 2
+; CHECK-NEXT: store i32 [[DIV1]], ptr [[P:%.*]], align 4
+; CHECK-NEXT: [[ADD:%.*]] = shl nuw nsw i32 [[DIV1]], 1
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 4
; CHECK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: ret i32 0
;
diff --git a/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll b/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll
index dbe29a071092e4..bd509509c321f8 100644
--- a/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll
+++ b/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll
@@ -38,7 +38,7 @@ define void @test_fill_with_foreach([2 x i64] %elems.coerce) {
; CHECK: for.body:
; CHECK-NEXT: [[__BEGIN1_SROA_0_03:%.*]] = phi ptr [ [[INCDEC_PTR_I:%.*]], [[FOR_BODY]] ], [ [[TMP0]], [[FOR_COND_PREHEADER_SPLIT]] ]
; CHECK-NEXT: tail call void @use(ptr noundef nonnull align 4 dereferenceable(4) [[__BEGIN1_SROA_0_03]])
-; CHECK-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i32, ptr [[__BEGIN1_SROA_0_03]], i64 1
+; CHECK-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds i8, ptr [[__BEGIN1_SROA_0_03]], i64 4
; CHECK-NEXT: [[CMP_I_NOT:%.*]] = icmp eq ptr [[INCDEC_PTR_I]], [[ADD_PTR_I]]
; CHECK-NEXT: br i1 [[CMP_I_NOT]], label [[COMMON_RET]], label [[FOR_BODY]]
;
@@ -133,7 +133,7 @@ define void @foo(ptr noundef nonnull align 8 dereferenceable(24) noalias %vec) #
; CHECK-LABEL: define void @foo
; CHECK-SAME: (ptr noalias nocapture noundef nonnull readonly align 8 dereferenceable(24) [[VEC:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[_M_FINISH_I_I:%.*]] = getelementptr inbounds [[VECTOR_IMPL_DATA:%.*]], ptr [[VEC]], i64 0, i32 1
+; CHECK-NEXT: [[_M_FINISH_I_I:%.*]] = getelementptr inbounds i8, ptr [[VEC]], i64 8
; CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[_M_FINISH_I_I]], align 8, !tbaa [[TBAA0:![0-9]+]]
; CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[VEC]], align 8, !tbaa [[TBAA5:![0-9]+]]
; CHECK-NEXT: [[SUB_PTR_LHS_CAST_I_I:%.*]] = ptrtoint ptr [[TMP0]] to i64
@@ -272,7 +272,7 @@ define void @loop_with_signed_induction(ptr noundef nonnull align 8 dereferencea
; CHECK-LABEL: define void @loop_with_signed_induction
; CHECK-SAME: (ptr nocapture noundef nonnull readonly align 8 dereferenceable(24) [[VEC:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[_M_FINISH_I_I:%.*]] = getelementptr inbounds [[VECTOR_IMPL_DATA:%.*]], ptr [[VEC]], i64 0, i32 1
+; CHECK-NEXT: [[_M_FINISH_I_I:%.*]] = getelementptr inbounds i8, ptr [[VEC]], i64 8
; CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[_M_FINISH_I_I]], align 8, !tbaa [[TBAA0]]
; CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[VEC]], align 8, !tbaa [[TBAA5]]
; CHECK-NEXT: [[SUB_PTR_LHS_CAST_I_I:%.*]] = ptrtoint ptr [[TMP0]] to i64
diff --git a/llvm/test/Transforms/PhaseOrdering/pr39282.ll b/llvm/test/Transforms/PhaseOrdering/pr39282.ll
index 776c5c482bab5e..2bbc01b677bb69 100644
--- a/llvm/test/Transforms/PhaseOrdering/pr39282.ll
+++ b/llvm/test/Transforms/PhaseOrdering/pr39282.ll
@@ -21,8 +21,8 @@ define void @pr39282(ptr %addr1, ptr %addr2) {
; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]])
; CHECK-NEXT: [[X_I:%.*]] = load i32, ptr [[ADDR1:%.*]], align 4, !alias.scope !3, !noalias !0
; CHECK-NEXT: store i32 [[X_I]], ptr [[ADDR2:%.*]], align 4, !alias.scope !0, !noalias !3
-; CHECK-NEXT: [[ADDR1I_1:%.*]] = getelementptr inbounds i32, ptr [[ADDR1]], i64 1
-; CHECK-NEXT: [[ADDR2I_1:%.*]] = getelementptr inbounds i32, ptr [[ADDR2]], i64 1
+; CHECK-NEXT: [[ADDR1I_1:%.*]] = getelementptr inbounds i8, ptr [[ADDR1]], i64 4
+; CHECK-NEXT: [[ADDR2I_1:%.*]] = getelementptr inbounds i8, ptr [[ADDR2]], i64 4
; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])
; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META7:![0-9]+]])
; CHECK-NEXT: [[X_I_1:%.*]] = load i32, ptr [[ADDR1I_1]], align 4, !alias.scope !7, !noalias !5
diff --git a/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll b/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll
index f186aceca4993d..f5807c2105abe1 100644
--- a/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll
+++ b/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll
@@ -9,7 +9,7 @@ define i1 @PR33605(i32 %a, i32 %b, ptr %c) {
; CHECK-LABEL: @PR33605(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[OR:%.*]] = or i32 [[B:%.*]], [[A:%.*]]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i64 4
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ne i32 [[OR]], [[TMP0]]
; CHECK-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
diff --git a/llvm/test/Transforms/PhaseOrdering/switch_with_geps.ll b/llvm/test/Transforms/PhaseOrdering/switch_with_geps.ll
index 8f497e92e4ea1e..aba003b13b4593 100644
--- a/llvm/test/Transforms/PhaseOrdering/switch_with_geps.ll
+++ b/llvm/test/Transforms/PhaseOrdering/switch_with_geps.ll
@@ -10,34 +10,11 @@ target datalayout = "n64"
%"OpKind::Two" = type { [1 x i32], i32, i16, i16 }
%"OpKind::Three" = type { [1 x i32], i32, i16, i16, i16, [1 x i16] }
-; FIXME: The switch should be optimized away.
define i32 @test(ptr %ptr) {
; CHECK-LABEL: define i32 @test(
; CHECK-SAME: ptr nocapture readonly [[PTR:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: start:
-; CHECK-NEXT: [[T:%.*]] = load i32, ptr [[PTR]], align 4
-; CHECK-NEXT: switch i32 [[T]], label [[DEFAULT:%.*]] [
-; CHECK-NEXT: i32 0, label [[BB4:%.*]]
-; CHECK-NEXT: i32 1, label [[BB5:%.*]]
-; CHECK-NEXT: i32 2, label [[BB6:%.*]]
-; CHECK-NEXT: i32 3, label [[BB7:%.*]]
-; CHECK-NEXT: ]
-; CHECK: default:
-; CHECK-NEXT: unreachable
-; CHECK: bb4:
-; CHECK-NEXT: [[GEP0:%.*]] = getelementptr inbounds %"OpKind::Zero", ptr [[PTR]], i64 0, i32 1
-; CHECK-NEXT: br label [[EXIT:%.*]]
-; CHECK: bb5:
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds %"OpKind::One", ptr [[PTR]], i64 0, i32 1
-; CHECK-NEXT: br label [[EXIT]]
-; CHECK: bb6:
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds %"OpKind::Two", ptr [[PTR]], i64 0, i32 1
-; CHECK-NEXT: br label [[EXIT]]
-; CHECK: bb7:
-; CHECK-NEXT: [[GEP3:%.*]] = getelementptr inbounds %"OpKind::Three", ptr [[PTR]], i64 0, i32 1
-; CHECK-NEXT: br label [[EXIT]]
-; CHECK: exit:
-; CHECK-NEXT: [[PHI:%.*]] = phi ptr [ [[GEP3]], [[BB7]] ], [ [[GEP2]], [[BB6]] ], [ [[GEP1]], [[BB5]] ], [ [[GEP0]], [[BB4]] ]
+; CHECK-NEXT: [[PHI:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 4
; CHECK-NEXT: [[RET:%.*]] = load i32, ptr [[PHI]], align 4
; CHECK-NEXT: ret i32 [[RET]]
;
@@ -77,38 +54,13 @@ exit:
%X = type { i64, i64, i64, i64, i64, i64 }
-; FIXME: The switch should be optimized away.
define void @test2(ptr %self, i64 %v, i64 %ix) {
; CHECK-LABEL: define void @test2(
; CHECK-SAME: ptr nocapture writeonly [[SELF:%.*]], i64 [[V:%.*]], i64 [[IX:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] {
; CHECK-NEXT: start:
-; CHECK-NEXT: switch i64 [[IX]], label [[DEFAULT:%.*]] [
-; CHECK-NEXT: i64 1, label [[BB3:%.*]]
-; CHECK-NEXT: i64 2, label [[BB4:%.*]]
-; CHECK-NEXT: i64 3, label [[BB5:%.*]]
-; CHECK-NEXT: i64 4, label [[BB6:%.*]]
-; CHECK-NEXT: i64 5, label [[BB7:%.*]]
-; CHECK-NEXT: ]
-; CHECK: default:
-; CHECK-NEXT: unreachable
-; CHECK: bb3:
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds [[X:%.*]], ptr [[SELF]], i64 0, i32 1
-; CHECK-NEXT: br label [[BB8:%.*]]
-; CHECK: bb4:
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds [[X]], ptr [[SELF]], i64 0, i32 2
-; CHECK-NEXT: br label [[BB8]]
-; CHECK: bb5:
-; CHECK-NEXT: [[GEP3:%.*]] = getelementptr inbounds [[X]], ptr [[SELF]], i64 0, i32 3
-; CHECK-NEXT: br label [[BB8]]
-; CHECK: bb6:
-; CHECK-NEXT: [[GEP4:%.*]] = getelementptr inbounds [[X]], ptr [[SELF]], i64 0, i32 4
-; CHECK-NEXT: br label [[BB8]]
-; CHECK: bb7:
-; CHECK-NEXT: [[GEP5:%.*]] = getelementptr inbounds [[X]], ptr [[SELF]], i64 0, i32 5
-; CHECK-NEXT: br label [[BB8]]
-; CHECK: bb8:
-; CHECK-NEXT: [[PTR:%.*]] = phi ptr [ [[GEP5]], [[BB7]] ], [ [[GEP4]], [[BB6]] ], [ [[GEP3]], [[BB5]] ], [ [[GEP2]], [[BB4]] ], [ [[GEP1]], [[BB3]] ]
-; CHECK-NEXT: store i64 [[V]], ptr [[PTR]], align 8
+; CHECK-NEXT: [[SWITCH_TABLEIDX:%.*]] = shl i64 [[IX]], 3
+; CHECK-NEXT: [[GEP5:%.*]] = getelementptr inbounds i8, ptr [[SELF]], i64 [[SWITCH_TABLEIDX]]
+; CHECK-NEXT: store i64 [[V]], ptr [[GEP5]], align 8
; CHECK-NEXT: ret void
;
start:
diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/gather-cost.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/gather-cost.ll
index 3a2a482a1ae7fb..2ea47216925037 100644
--- a/llvm/test/Transforms/SLPVectorizer/AArch64/gather-cost.ll
+++ b/llvm/test/Transforms/SLPVectorizer/AArch64/gather-cost.ll
@@ -16,8 +16,8 @@ target triple = "aarch64--linux-gnu"
define internal i32 @gather_multiple_use(i32 %a, i32 %b, i32 %c, i32 %d) {
; CHECK-LABEL: @gather_multiple_use(
-; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> poison, i32 [[C:%.*]], i64 0
-; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> [[TMP1]], i32 [[A:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> poison, i32 [[A:%.*]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> [[TMP1]], i32 [[C:%.*]], i64 1
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x i32> [[TMP2]], i32 [[B:%.*]], i64 2
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> [[TMP3]], i32 [[D:%.*]], i64 3
; CHECK-NEXT: [[TMP5:%.*]] = lshr <4 x i32> [[TMP4]], <i32 15, i32 15, i32 15, i32 15>
@@ -57,10 +57,10 @@ define internal i32 @gather_multiple_use(i32 %a, i32 %b, i32 %c, i32 %d) {
@data = global [6 x [258 x i8]] zeroinitializer, align 1
define void @gather_load(ptr noalias %ptr) {
; CHECK-LABEL: @gather_load(
-; CHECK-NEXT: [[ARRAYIDX182:%.*]] = getelementptr inbounds i16, ptr [[PTR:%.*]], i64 1
-; CHECK-NEXT: [[ARRAYIDX183:%.*]] = getelementptr inbounds i16, ptr [[PTR]], i64 2
-; CHECK-NEXT: [[ARRAYIDX184:%.*]] = getelementptr inbounds i16, ptr [[PTR]], i64 3
-; CHECK-NEXT: [[ARRAYIDX185:%.*]] = getelementptr inbounds i16, ptr [[PTR]], i64 4
+; CHECK-NEXT: [[ARRAYIDX182:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 2
+; CHECK-NEXT: [[ARRAYIDX183:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 4
+; CHECK-NEXT: [[ARRAYIDX184:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 6
+; CHECK-NEXT: [[ARRAYIDX185:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 8
; CHECK-NEXT: [[L0:%.*]] = load i8, ptr getelementptr inbounds ([6 x [258 x i8]], ptr @data, i64 0, i64 1, i64 0), align 1
; CHECK-NEXT: [[CONV150:%.*]] = zext i8 [[L0]] to i16
; CHECK-NEXT: [[ADD152:%.*]] = add nuw nsw i16 [[CONV150]], 10
diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/gather-reduce.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/gather-reduce.ll
index d13597b43f9f72..51c6673a143537 100644
--- a/llvm/test/Transforms/SLPVectorizer/AArch64/gather-reduce.ll
+++ b/llvm/test/Transforms/SLPVectorizer/AArch64/gather-reduce.ll
@@ -36,7 +36,7 @@ define i32 @gather_reduce_8x16_i32(ptr nocapture readonly %a, ptr nocapture read
; GENERIC-NEXT: [[I_0103:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; GENERIC-NEXT: [[SUM_0102:%.*]] = phi i32 [ [[ADD66]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; GENERIC-NEXT: [[A_ADDR_0101:%.*]] = phi ptr [ [[INCDEC_PTR58:%.*]], [[FOR_BODY]] ], [ [[A:%.*]], [[FOR_BODY_PREHEADER]] ]
-; GENERIC-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i16, ptr [[A_ADDR_0101]], i64 8
+; GENERIC-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i8, ptr [[A_ADDR_0101]], i64 16
; GENERIC-NEXT: [[TMP0:%.*]] = load <8 x i16>, ptr [[A_ADDR_0101]], align 2
; GENERIC-NEXT: [[TMP1:%.*]] = zext <8 x i16> [[TMP0]] to <8 x i32>
; GENERIC-NEXT: [[TMP2:%.*]] = load <8 x i16>, ptr [[B:%.*]], align 2
@@ -109,7 +109,7 @@ define i32 @gather_reduce_8x16_i32(ptr nocapture readonly %a, ptr nocapture read
; KRYO-NEXT: [[I_0103:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; KRYO-NEXT: [[SUM_0102:%.*]] = phi i32 [ [[ADD66]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; KRYO-NEXT: [[A_ADDR_0101:%.*]] = phi ptr [ [[INCDEC_PTR58:%.*]], [[FOR_BODY]] ], [ [[A:%.*]], [[FOR_BODY_PREHEADER]] ]
-; KRYO-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i16, ptr [[A_ADDR_0101]], i64 8
+; KRYO-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i8, ptr [[A_ADDR_0101]], i64 16
; KRYO-NEXT: [[TMP0:%.*]] = load <8 x i16>, ptr [[A_ADDR_0101]], align 2
; KRYO-NEXT: [[TMP1:%.*]] = zext <8 x i16> [[TMP0]] to <8 x i32>
; KRYO-NEXT: [[TMP2:%.*]] = load <8 x i16>, ptr [[B:%.*]], align 2
@@ -293,7 +293,7 @@ define i32 @gather_reduce_8x16_i64(ptr nocapture readonly %a, ptr nocapture read
; GENERIC-NEXT: [[I_0103:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; GENERIC-NEXT: [[SUM_0102:%.*]] = phi i32 [ [[ADD66]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; GENERIC-NEXT: [[A_ADDR_0101:%.*]] = phi ptr [ [[INCDEC_PTR58:%.*]], [[FOR_BODY]] ], [ [[A:%.*]], [[FOR_BODY_PREHEADER]] ]
-; GENERIC-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i16, ptr [[A_ADDR_0101]], i64 8
+; GENERIC-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i8, ptr [[A_ADDR_0101]], i64 16
; GENERIC-NEXT: [[TMP0:%.*]] = load <8 x i16>, ptr [[A_ADDR_0101]], align 2
; GENERIC-NEXT: [[TMP1:%.*]] = zext <8 x i16> [[TMP0]] to <8 x i32>
; GENERIC-NEXT: [[TMP2:%.*]] = load <8 x i16>, ptr [[B:%.*]], align 2
@@ -366,7 +366,7 @@ define i32 @gather_reduce_8x16_i64(ptr nocapture readonly %a, ptr nocapture read
; KRYO-NEXT: [[I_0103:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; KRYO-NEXT: [[SUM_0102:%.*]] = phi i32 [ [[ADD66]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; KRYO-NEXT: [[A_ADDR_0101:%.*]] = phi ptr [ [[INCDEC_PTR58:%.*]], [[FOR_BODY]] ], [ [[A:%.*]], [[FOR_BODY_PREHEADER]] ]
-; KRYO-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i16, ptr [[A_ADDR_0101]], i64 8
+; KRYO-NEXT: [[INCDEC_PTR58]] = getelementptr inbounds i8, ptr [[A_ADDR_0101]], i64 16
; KRYO-NEXT: [[TMP0:%.*]] = load <8 x i16>, ptr [[A_ADDR_0101]], align 2
; KRYO-NEXT: [[TMP1:%.*]] = zext <8 x i16> [[TMP0]] to <8 x i32>
; KRYO-NEXT: [[TMP2:%.*]] = load <8 x i16>, ptr [[B:%.*]], align 2
diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/loadorder.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/loadorder.ll
index efe8bd9e610bb3..f98d0ad0527c29 100644
--- a/llvm/test/Transforms/SLPVectorizer/AArch64/loadorder.ll
+++ b/llvm/test/Transforms/SLPVectorizer/AArch64/loadorder.ll
@@ -133,7 +133,7 @@ define i16 @reduce_blockstrided2(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-LABEL: @reduce_blockstrided2(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[X:%.*]], align 2
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, ptr [[X]], i64 1
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[X]], i64 2
; CHECK-NEXT: [[TMP1:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2
; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[STRIDE:%.*]] to i64
; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, ptr [[X]], i64 [[IDXPROM]]
@@ -165,7 +165,7 @@ define i16 @reduce_blockstrided2(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-NEXT: [[TMP10:%.*]] = load i16, ptr [[ARRAYIDX28]], align 2
; CHECK-NEXT: [[ARRAYIDX32:%.*]] = getelementptr inbounds i16, ptr [[Y]], i64 [[IDXPROM15]]
; CHECK-NEXT: [[TMP11:%.*]] = load i16, ptr [[ARRAYIDX32]], align 2
-; CHECK-NEXT: [[ARRAYIDX33:%.*]] = getelementptr inbounds i16, ptr [[Y]], i64 1
+; CHECK-NEXT: [[ARRAYIDX33:%.*]] = getelementptr inbounds i8, ptr [[Y]], i64 2
; CHECK-NEXT: [[TMP12:%.*]] = load i16, ptr [[ARRAYIDX33]], align 2
; CHECK-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i16, ptr [[Y]], i64 [[IDXPROM4]]
; CHECK-NEXT: [[TMP13:%.*]] = load i16, ptr [[ARRAYIDX36]], align 2
@@ -254,9 +254,9 @@ define i16 @reduce_blockstrided3(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-LABEL: @reduce_blockstrided3(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[L0:%.*]] = load i16, ptr [[X:%.*]], align 2
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, ptr [[X]], i64 1
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[X]], i64 2
; CHECK-NEXT: [[L1:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, ptr [[X]], i64 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[X]], i64 4
; CHECK-NEXT: [[L2:%.*]] = load i16, ptr [[ARRAYIDX2]], align 2
; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[STRIDE:%.*]] to i64
; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, ptr [[X]], i64 [[IDXPROM]]
@@ -270,9 +270,9 @@ define i16 @reduce_blockstrided3(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i16, ptr [[X]], i64 [[IDXPROM9]]
; CHECK-NEXT: [[L6:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
; CHECK-NEXT: [[L8:%.*]] = load i16, ptr [[Y:%.*]], align 2
-; CHECK-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds i16, ptr [[Y]], i64 1
+; CHECK-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds i8, ptr [[Y]], i64 2
; CHECK-NEXT: [[L9:%.*]] = load i16, ptr [[ARRAYIDX15]], align 2
-; CHECK-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds i16, ptr [[Y]], i64 2
+; CHECK-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds i8, ptr [[Y]], i64 4
; CHECK-NEXT: [[L10:%.*]] = load i16, ptr [[ARRAYIDX16]], align 2
; CHECK-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds i16, ptr [[Y]], i64 [[IDXPROM]]
; CHECK-NEXT: [[L12:%.*]] = load i16, ptr [[ARRAYIDX20]], align 2
@@ -678,7 +678,7 @@ define void @store_blockstrided3(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-LABEL: @store_blockstrided3(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[X:%.*]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[X]], i64 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[X]], i64 8
; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD4:%.*]] = add nsw i32 [[STRIDE:%.*]], 1
; CHECK-NEXT: [[IDXPROM5:%.*]] = sext i32 [[ADD4]] to i64
@@ -699,7 +699,7 @@ define void @store_blockstrided3(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-NEXT: [[ARRAYIDX32:%.*]] = getelementptr inbounds i32, ptr [[X]], i64 [[IDXPROM31]]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX32]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = load <2 x i32>, ptr [[Y:%.*]], align 4
-; CHECK-NEXT: [[ARRAYIDX35:%.*]] = getelementptr inbounds i32, ptr [[Y]], i64 2
+; CHECK-NEXT: [[ARRAYIDX35:%.*]] = getelementptr inbounds i8, ptr [[Y]], i64 8
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX35]], align 4
; CHECK-NEXT: [[ARRAYIDX41:%.*]] = getelementptr inbounds i32, ptr [[Y]], i64 [[IDXPROM5]]
; CHECK-NEXT: [[TMP7:%.*]] = load <2 x i32>, ptr [[ARRAYIDX41]], align 4
@@ -709,19 +709,19 @@ define void @store_blockstrided3(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-NEXT: [[ARRAYIDX60:%.*]] = getelementptr inbounds i32, ptr [[Y]], i64 [[IDXPROM23]]
; CHECK-NEXT: [[ARRAYIDX68:%.*]] = getelementptr inbounds i32, ptr [[Y]], i64 [[IDXPROM31]]
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX68]], align 4
-; CHECK-NEXT: [[ARRAYIDX72:%.*]] = getelementptr inbounds i32, ptr [[Z:%.*]], i64 1
+; CHECK-NEXT: [[ARRAYIDX72:%.*]] = getelementptr inbounds i8, ptr [[Z:%.*]], i64 4
; CHECK-NEXT: [[MUL73:%.*]] = mul nsw i32 [[TMP6]], [[TMP1]]
-; CHECK-NEXT: [[ARRAYIDX76:%.*]] = getelementptr inbounds i32, ptr [[Z]], i64 6
+; CHECK-NEXT: [[ARRAYIDX76:%.*]] = getelementptr inbounds i8, ptr [[Z]], i64 24
; CHECK-NEXT: [[TMP10:%.*]] = mul nsw <2 x i32> [[TMP5]], [[TMP0]]
; CHECK-NEXT: [[TMP11:%.*]] = mul nsw <2 x i32> [[TMP7]], [[TMP2]]
; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <2 x i32> [[TMP10]], <2 x i32> [[TMP11]], <4 x i32> <i32 1, i32 0, i32 3, i32 2>
; CHECK-NEXT: [[MUL81:%.*]] = mul nsw i32 [[TMP8]], [[TMP3]]
-; CHECK-NEXT: [[ARRAYIDX82:%.*]] = getelementptr inbounds i32, ptr [[Z]], i64 8
+; CHECK-NEXT: [[ARRAYIDX82:%.*]] = getelementptr inbounds i8, ptr [[Z]], i64 32
; CHECK-NEXT: [[TMP13:%.*]] = load <2 x i32>, ptr [[ARRAYIDX16]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = load <2 x i32>, ptr [[ARRAYIDX52]], align 4
; CHECK-NEXT: [[TMP15:%.*]] = mul nsw <2 x i32> [[TMP14]], [[TMP13]]
; CHECK-NEXT: [[TMP16:%.*]] = shufflevector <2 x i32> [[TMP15]], <2 x i32> poison, <2 x i32> <i32 1, i32 0>
-; CHECK-NEXT: [[ARRAYIDX90:%.*]] = getelementptr inbounds i32, ptr [[Z]], i64 10
+; CHECK-NEXT: [[ARRAYIDX90:%.*]] = getelementptr inbounds i8, ptr [[Z]], i64 40
; CHECK-NEXT: [[TMP17:%.*]] = load <2 x i32>, ptr [[ARRAYIDX24]], align 4
; CHECK-NEXT: [[TMP18:%.*]] = load <2 x i32>, ptr [[ARRAYIDX60]], align 4
; CHECK-NEXT: store i32 [[MUL73]], ptr [[Z]], align 4
@@ -732,7 +732,7 @@ define void @store_blockstrided3(ptr nocapture noundef readonly %x, ptr nocaptur
; CHECK-NEXT: [[TMP20:%.*]] = shufflevector <2 x i32> [[TMP19]], <2 x i32> poison, <2 x i32> <i32 1, i32 0>
; CHECK-NEXT: store <2 x i32> [[TMP20]], ptr [[ARRAYIDX90]], align 4
; CHECK-NEXT: [[MUL91:%.*]] = mul nsw i32 [[TMP9]], [[TMP4]]
-; CHECK-NEXT: [[ARRAYIDX92:%.*]] = getelementptr inbounds i32, ptr [[Z]], i64 9
+; CHECK-NEXT: [[ARRAYIDX92:%.*]] = getelementptr inbounds i8, ptr [[Z]], i64 36
; CHECK-NEXT: store i32 [[MUL91]], ptr [[ARRAYIDX92]], align 4
; CHECK-NEXT: ret void
;
@@ -921,9 +921,9 @@ define void @store_blockstrided4x4(ptr nocapture noundef readonly %p1, i32 nound
; CHECK-NEXT: [[ADD_PTR64:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[IDX_EXT63]]
; CHECK-NEXT: [[ARRAYIDX3_1:%.*]] = getelementptr inbounds i8, ptr [[ADD_PTR]], i64 4
; CHECK-NEXT: [[ARRAYIDX5_1:%.*]] = getelementptr inbounds i8, ptr [[ADD_PTR64]], i64 4
-; CHECK-NEXT: [[DST4:%.*]] = getelementptr inbounds i32, ptr [[DST0:%.*]], i64 4
-; CHECK-NEXT: [[DST8:%.*]] = getelementptr inbounds i32, ptr [[DST0]], i64 8
-; CHECK-NEXT: [[DST12:%.*]] = getelementptr inbounds i32, ptr [[DST0]], i64 12
+; CHECK-NEXT: [[DST4:%.*]] = getelementptr inbounds i8, ptr [[DST0:%.*]], i64 16
+; CHECK-NEXT: [[DST8:%.*]] = getelementptr inbounds i8, ptr [[DST0]], i64 32
+; CHECK-NEXT: [[DST12:%.*]] = getelementptr inbounds i8, ptr [[DST0]], i64 48
; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[P1]], align 1
; CHECK-NEXT: [[TMP1:%.*]] = zext <4 x i8> [[TMP0]] to <4 x i32>
; CHECK-NEXT: [[TMP2:%.*]] = load <4 x i8>, ptr [[ARRAYIDX3]], align 1
diff --git a/llvm/test/Transforms/SLPVectorizer/WebAssembly/no-vectorize-rotate.ll b/llvm/test/Transforms/SLPVectorizer/WebAssembly/no-vectorize-rotate.ll
index 24365a850d91d8..1c408f19a8a324 100644
--- a/llvm/test/Transforms/SLPVectorizer/WebAssembly/no-vectorize-rotate.ll
+++ b/llvm/test/Transforms/SLPVectorizer/WebAssembly/no-vectorize-rotate.ll
@@ -18,7 +18,7 @@ define void @foo(<2 x i64> %x, <4 x i32> %y, ptr %out) #0 {
; CHECK-NEXT: [[E:%.*]] = extractelement <4 x i32> [[Y]], i64 3
; CHECK-NEXT: [[CONV17:%.*]] = zext i32 [[E]] to i64
; CHECK-NEXT: [[F:%.*]] = tail call i64 @llvm.fshl.i64(i64 [[D]], i64 [[D]], i64 [[CONV17]])
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i64, ptr [[OUT]], i32 1
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[OUT]], i32 8
; CHECK-NEXT: store i64 [[F]], ptr [[ARRAYIDX2]], align 8
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/operandorder.ll b/llvm/test/Transforms/SLPVectorizer/X86/operandorder.ll
index b58e5f06d78da1..593aad82ad5d87 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/operandorder.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/operandorder.ll
@@ -233,7 +233,7 @@ define void @vecload_vs_broadcast4(ptr noalias %from, ptr noalias %to, double %v
; CHECK-NEXT: br label [[LP:%.*]]
; CHECK: lp:
; CHECK-NEXT: [[P:%.*]] = phi double [ 1.000000e+00, [[LP]] ], [ 0.000000e+00, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[FROM_1:%.*]] = getelementptr double, ptr [[FROM:%.*]], i32 1
+; CHECK-NEXT: [[FROM_1:%.*]] = getelementptr i8, ptr [[FROM:%.*]], i32 8
; CHECK-NEXT: [[V0_1:%.*]] = load double, ptr [[FROM]], align 4
; CHECK-NEXT: [[V0_2:%.*]] = load double, ptr [[FROM_1]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <2 x double> poison, double [[V0_2]], i64 0
@@ -286,7 +286,7 @@ define void @shuffle_nodes_match2(ptr noalias %from, ptr noalias %to, double %v1
; CHECK-NEXT: br label [[LP:%.*]]
; CHECK: lp:
; CHECK-NEXT: [[P:%.*]] = phi double [ 1.000000e+00, [[LP]] ], [ 0.000000e+00, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[FROM_1:%.*]] = getelementptr double, ptr [[FROM:%.*]], i32 1
+; CHECK-NEXT: [[FROM_1:%.*]] = getelementptr i8, ptr [[FROM:%.*]], i32 8
; CHECK-NEXT: [[V0_1:%.*]] = load double, ptr [[FROM]], align 4
; CHECK-NEXT: [[V0_2:%.*]] = load double, ptr [[FROM_1]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <2 x double> poison, double [[V0_1]], i64 0
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/opt.ll b/llvm/test/Transforms/SLPVectorizer/X86/opt.ll
index 85447902618c8b..95da098b8b1435 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/opt.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/opt.ll
@@ -10,10 +10,10 @@ target triple = "x86_64-apple-macosx10.8.0"
define void @test1(ptr %a, ptr %b, ptr %c) {
; SLP-LABEL: @test1(
; SLP-NEXT: entry:
-; SLP-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
-; SLP-NEXT: [[TMP3:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
-; SLP-NEXT: [[TMP4:%.*]] = fmul <2 x double> [[TMP1]], [[TMP3]]
-; SLP-NEXT: store <2 x double> [[TMP4]], ptr [[C:%.*]], align 8
+; SLP-NEXT: [[TMP0:%.*]] = load <2 x double>, ptr [[A:%.*]], align 8
+; SLP-NEXT: [[TMP1:%.*]] = load <2 x double>, ptr [[B:%.*]], align 8
+; SLP-NEXT: [[TMP2:%.*]] = fmul <2 x double> [[TMP0]], [[TMP1]]
+; SLP-NEXT: store <2 x double> [[TMP2]], ptr [[C:%.*]], align 8
; SLP-NEXT: ret void
;
; NOSLP-LABEL: @test1(
@@ -21,13 +21,13 @@ define void @test1(ptr %a, ptr %b, ptr %c) {
; NOSLP-NEXT: [[I0:%.*]] = load double, ptr [[A:%.*]], align 8
; NOSLP-NEXT: [[I1:%.*]] = load double, ptr [[B:%.*]], align 8
; NOSLP-NEXT: [[MUL:%.*]] = fmul double [[I0]], [[I1]]
-; NOSLP-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds double, ptr [[A]], i64 1
+; NOSLP-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 8
; NOSLP-NEXT: [[I3:%.*]] = load double, ptr [[ARRAYIDX3]], align 8
-; NOSLP-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds double, ptr [[B]], i64 1
+; NOSLP-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 8
; NOSLP-NEXT: [[I4:%.*]] = load double, ptr [[ARRAYIDX4]], align 8
; NOSLP-NEXT: [[MUL5:%.*]] = fmul double [[I3]], [[I4]]
; NOSLP-NEXT: store double [[MUL]], ptr [[C:%.*]], align 8
-; NOSLP-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds double, ptr [[C]], i64 1
+; NOSLP-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i8, ptr [[C]], i64 8
; NOSLP-NEXT: store double [[MUL5]], ptr [[ARRAYIDX5]], align 8
; NOSLP-NEXT: ret void
;
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/pr46983.ll b/llvm/test/Transforms/SLPVectorizer/X86/pr46983.ll
index 48d8de2a961868..c38f2748a9763d 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/pr46983.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/pr46983.ll
@@ -7,10 +7,10 @@
define void @store_i32(ptr nocapture %0, i32 %1, i32 %2) {
; CHECK-LABEL: @store_i32(
-; CHECK-NEXT: [[TMP5:%.*]] = load <4 x i32>, ptr [[TMP0:%.*]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; CHECK-NEXT: [[TMP6:%.*]] = insertelement <4 x i32> poison, i32 [[TMP1:%.*]], i64 0
-; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <4 x i32> [[TMP6]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP7:%.*]] = mul <4 x i32> [[TMP5]], [[SHUFFLE]]
+; CHECK-NEXT: [[TMP4:%.*]] = load <4 x i32>, ptr [[TMP0:%.*]], align 4, !tbaa [[TBAA0:![0-9]+]]
+; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x i32> poison, i32 [[TMP1:%.*]], i64 0
+; CHECK-NEXT: [[TMP6:%.*]] = shufflevector <4 x i32> [[TMP5]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP7:%.*]] = mul <4 x i32> [[TMP4]], [[TMP6]]
; CHECK-NEXT: [[TMP8:%.*]] = lshr <4 x i32> [[TMP7]], <i32 15, i32 15, i32 15, i32 15>
; CHECK-NEXT: [[TMP9:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP8]], <4 x i32> <i32 255, i32 255, i32 255, i32 255>)
; CHECK-NEXT: store <4 x i32> [[TMP9]], ptr [[TMP0]], align 4, !tbaa [[TBAA0]]
@@ -48,11 +48,11 @@ define void @store_i32(ptr nocapture %0, i32 %1, i32 %2) {
define void @store_i8(ptr nocapture %0, i32 %1, i32 %2) {
; CHECK-LABEL: @store_i8(
-; CHECK-NEXT: [[TMP5:%.*]] = load <4 x i8>, ptr [[TMP0:%.*]], align 1, !tbaa [[TBAA4:![0-9]+]]
-; CHECK-NEXT: [[TMP6:%.*]] = zext <4 x i8> [[TMP5]] to <4 x i32>
-; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP1:%.*]], i64 0
-; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <4 x i32> [[TMP7]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP8:%.*]] = mul <4 x i32> [[SHUFFLE]], [[TMP6]]
+; CHECK-NEXT: [[TMP4:%.*]] = load <4 x i8>, ptr [[TMP0:%.*]], align 1, !tbaa [[TBAA4:![0-9]+]]
+; CHECK-NEXT: [[TMP5:%.*]] = zext <4 x i8> [[TMP4]] to <4 x i32>
+; CHECK-NEXT: [[TMP6:%.*]] = insertelement <4 x i32> poison, i32 [[TMP1:%.*]], i64 0
+; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <4 x i32> [[TMP6]], <4 x i32> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP8:%.*]] = mul <4 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT: [[TMP9:%.*]] = lshr <4 x i32> [[TMP8]], <i32 15, i32 15, i32 15, i32 15>
; CHECK-NEXT: [[TMP10:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP9]], <4 x i32> <i32 255, i32 255, i32 255, i32 255>)
; CHECK-NEXT: [[TMP11:%.*]] = trunc <4 x i32> [[TMP10]] to <4 x i8>
@@ -108,7 +108,7 @@ define void @store_i64(ptr nocapture %0, i32 %1, i32 %2) {
; SSE-NEXT: [[TMP10:%.*]] = and i64 [[TMP7]], 4294967295
; SSE-NEXT: [[TMP11:%.*]] = select i1 [[TMP9]], i64 [[TMP10]], i64 255
; SSE-NEXT: store i64 [[TMP11]], ptr [[TMP0]], align 8, !tbaa [[TBAA5]]
-; SSE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 1
+; SSE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; SSE-NEXT: [[TMP13:%.*]] = load i64, ptr [[TMP12]], align 8, !tbaa [[TBAA5]]
; SSE-NEXT: [[TMP14:%.*]] = mul i64 [[TMP13]], [[TMP4]]
; SSE-NEXT: [[TMP15:%.*]] = lshr i64 [[TMP14]], 15
@@ -117,7 +117,7 @@ define void @store_i64(ptr nocapture %0, i32 %1, i32 %2) {
; SSE-NEXT: [[TMP18:%.*]] = and i64 [[TMP15]], 4294967295
; SSE-NEXT: [[TMP19:%.*]] = select i1 [[TMP17]], i64 [[TMP18]], i64 255
; SSE-NEXT: store i64 [[TMP19]], ptr [[TMP12]], align 8, !tbaa [[TBAA5]]
-; SSE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 2
+; SSE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; SSE-NEXT: [[TMP21:%.*]] = load i64, ptr [[TMP20]], align 8, !tbaa [[TBAA5]]
; SSE-NEXT: [[TMP22:%.*]] = mul i64 [[TMP21]], [[TMP4]]
; SSE-NEXT: [[TMP23:%.*]] = lshr i64 [[TMP22]], 15
@@ -126,7 +126,7 @@ define void @store_i64(ptr nocapture %0, i32 %1, i32 %2) {
; SSE-NEXT: [[TMP26:%.*]] = and i64 [[TMP23]], 4294967295
; SSE-NEXT: [[TMP27:%.*]] = select i1 [[TMP25]], i64 [[TMP26]], i64 255
; SSE-NEXT: store i64 [[TMP27]], ptr [[TMP20]], align 8, !tbaa [[TBAA5]]
-; SSE-NEXT: [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[TMP0]], i64 3
+; SSE-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 24
; SSE-NEXT: [[TMP29:%.*]] = load i64, ptr [[TMP28]], align 8, !tbaa [[TBAA5]]
; SSE-NEXT: [[TMP30:%.*]] = mul i64 [[TMP29]], [[TMP4]]
; SSE-NEXT: [[TMP31:%.*]] = lshr i64 [[TMP30]], 15
@@ -139,10 +139,10 @@ define void @store_i64(ptr nocapture %0, i32 %1, i32 %2) {
;
; AVX-LABEL: @store_i64(
; AVX-NEXT: [[TMP4:%.*]] = zext i32 [[TMP1:%.*]] to i64
-; AVX-NEXT: [[TMP6:%.*]] = load <4 x i64>, ptr [[TMP0:%.*]], align 8, !tbaa [[TBAA5:![0-9]+]]
-; AVX-NEXT: [[TMP7:%.*]] = insertelement <4 x i64> poison, i64 [[TMP4]], i64 0
-; AVX-NEXT: [[SHUFFLE:%.*]] = shufflevector <4 x i64> [[TMP7]], <4 x i64> poison, <4 x i32> zeroinitializer
-; AVX-NEXT: [[TMP8:%.*]] = mul <4 x i64> [[TMP6]], [[SHUFFLE]]
+; AVX-NEXT: [[TMP5:%.*]] = load <4 x i64>, ptr [[TMP0:%.*]], align 8, !tbaa [[TBAA5:![0-9]+]]
+; AVX-NEXT: [[TMP6:%.*]] = insertelement <4 x i64> poison, i64 [[TMP4]], i64 0
+; AVX-NEXT: [[TMP7:%.*]] = shufflevector <4 x i64> [[TMP6]], <4 x i64> poison, <4 x i32> zeroinitializer
+; AVX-NEXT: [[TMP8:%.*]] = mul <4 x i64> [[TMP5]], [[TMP7]]
; AVX-NEXT: [[TMP9:%.*]] = lshr <4 x i64> [[TMP8]], <i64 15, i64 15, i64 15, i64 15>
; AVX-NEXT: [[TMP10:%.*]] = trunc <4 x i64> [[TMP9]] to <4 x i32>
; AVX-NEXT: [[TMP11:%.*]] = icmp ult <4 x i32> [[TMP10]], <i32 255, i32 255, i32 255, i32 255>
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/pr47629-inseltpoison.ll b/llvm/test/Transforms/SLPVectorizer/X86/pr47629-inseltpoison.ll
index 8e17088b275697..5b33c6e889363e 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/pr47629-inseltpoison.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/pr47629-inseltpoison.ll
@@ -7,11 +7,11 @@
define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonly %1) {
; SSE-LABEL: @gather_load(
-; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; SSE-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; SSE-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; SSE-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -23,11 +23,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; SSE-NEXT: ret void
;
; AVX-LABEL: @gather_load(
-; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -39,11 +39,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; AVX-NEXT: ret void
;
; AVX2-LABEL: @gather_load(
-; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -55,11 +55,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; AVX2-NEXT: ret void
;
; AVX512F-LABEL: @gather_load(
-; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX512F-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX512F-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX512F-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX512F-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX512F-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -71,11 +71,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; AVX512F-NEXT: ret void
;
; AVX512VL-LABEL: @gather_load(
-; AVX512VL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX512VL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX512VL-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX512VL-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX512VL-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX512VL-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX512VL-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX512VL-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX512VL-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX512VL-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX512VL-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -107,35 +107,35 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
define void @gather_load_2(ptr noalias nocapture %0, ptr noalias nocapture readonly %1) {
; SSE-LABEL: @gather_load_2(
-; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; SSE-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP5:%.*]] = add nsw i32 [[TMP4]], 1
-; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP0:%.*]], i64 1
+; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP0:%.*]], i64 4
; SSE-NEXT: store i32 [[TMP5]], ptr [[TMP0]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; SSE-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP9:%.*]] = add nsw i32 [[TMP8]], 2
-; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 2
+; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; SSE-NEXT: store i32 [[TMP9]], ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; SSE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; SSE-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP13:%.*]] = add nsw i32 [[TMP12]], 3
-; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 3
+; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 12
; SSE-NEXT: store i32 [[TMP13]], ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; SSE-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; SSE-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP17:%.*]] = add nsw i32 [[TMP16]], 4
; SSE-NEXT: store i32 [[TMP17]], ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: ret void
;
; AVX-LABEL: @gather_load_2(
-; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; AVX-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; AVX-NEXT: [[TMP12:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP6]], i64 1
@@ -146,13 +146,13 @@ define void @gather_load_2(ptr noalias nocapture %0, ptr noalias nocapture reado
; AVX-NEXT: ret void
;
; AVX2-LABEL: @gather_load_2(
-; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; AVX2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX2-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; AVX2-NEXT: [[TMP12:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP6]], i64 1
@@ -163,13 +163,13 @@ define void @gather_load_2(ptr noalias nocapture %0, ptr noalias nocapture reado
; AVX2-NEXT: ret void
;
; AVX512F-LABEL: @gather_load_2(
-; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX512F-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX512F-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX512F-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; AVX512F-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX512F-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
; AVX512F-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; AVX512F-NEXT: [[TMP12:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP6]], i64 1
@@ -215,39 +215,39 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
; SSE-LABEL: @gather_load_3(
; SSE-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP4:%.*]] = add i32 [[TMP3]], 1
-; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP0:%.*]], i64 1
+; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP0:%.*]], i64 4
; SSE-NEXT: store i32 [[TMP4]], ptr [[TMP0]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; SSE-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP8:%.*]] = add i32 [[TMP7]], 2
-; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 2
+; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; SSE-NEXT: store i32 [[TMP8]], ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; SSE-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP12:%.*]] = add i32 [[TMP11]], 3
-; SSE-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 3
+; SSE-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 12
; SSE-NEXT: store i32 [[TMP12]], ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 15
+; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 60
; SSE-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP16:%.*]] = add i32 [[TMP15]], 4
-; SSE-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 4
+; SSE-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; SSE-NEXT: store i32 [[TMP16]], ptr [[TMP13]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 18
+; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 72
; SSE-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP20:%.*]] = add i32 [[TMP19]], 1
-; SSE-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 5
+; SSE-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 20
; SSE-NEXT: store i32 [[TMP20]], ptr [[TMP17]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 9
+; SSE-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 36
; SSE-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP24:%.*]] = add i32 [[TMP23]], 2
-; SSE-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 6
+; SSE-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 24
; SSE-NEXT: store i32 [[TMP24]], ptr [[TMP21]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 6
+; SSE-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 24
; SSE-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP28:%.*]] = add i32 [[TMP27]], 3
-; SSE-NEXT: [[TMP29:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 7
+; SSE-NEXT: [[TMP29:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 28
; SSE-NEXT: store i32 [[TMP28]], ptr [[TMP25]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 21
+; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 84
; SSE-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP32:%.*]] = add i32 [[TMP31]], 4
; SSE-NEXT: store i32 [[TMP32]], ptr [[TMP29]], align 4, !tbaa [[TBAA0]]
@@ -255,19 +255,19 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
;
; AVX-LABEL: @gather_load_3(
; AVX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 15
+; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 60
; AVX-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 18
+; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 72
; AVX-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 9
+; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 36
; AVX-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 6
+; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 24
; AVX-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 21
+; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 84
; AVX-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP18:%.*]] = insertelement <8 x i32> poison, i32 [[TMP3]], i64 0
; AVX-NEXT: [[TMP19:%.*]] = insertelement <8 x i32> [[TMP18]], i32 [[TMP5]], i64 1
@@ -283,19 +283,19 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
;
; AVX2-LABEL: @gather_load_3(
; AVX2-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX2-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX2-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 15
+; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 60
; AVX2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 18
+; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 72
; AVX2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 9
+; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 36
; AVX2-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 6
+; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 24
; AVX2-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 21
+; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 84
; AVX2-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP18:%.*]] = insertelement <8 x i32> poison, i32 [[TMP3]], i64 0
; AVX2-NEXT: [[TMP19:%.*]] = insertelement <8 x i32> [[TMP18]], i32 [[TMP5]], i64 1
@@ -370,20 +370,20 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture readonly %t1) {
; SSE-LABEL: @gather_load_4(
-; SSE-NEXT: [[T5:%.*]] = getelementptr inbounds i32, ptr [[T0:%.*]], i64 1
-; SSE-NEXT: [[T6:%.*]] = getelementptr inbounds i32, ptr [[T1:%.*]], i64 11
-; SSE-NEXT: [[T9:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 2
-; SSE-NEXT: [[T10:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 4
-; SSE-NEXT: [[T13:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 3
-; SSE-NEXT: [[T14:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 15
-; SSE-NEXT: [[T17:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 4
-; SSE-NEXT: [[T18:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 18
-; SSE-NEXT: [[T21:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 5
-; SSE-NEXT: [[T22:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 9
-; SSE-NEXT: [[T25:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 6
-; SSE-NEXT: [[T26:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 6
-; SSE-NEXT: [[T29:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 7
-; SSE-NEXT: [[T30:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 21
+; SSE-NEXT: [[T5:%.*]] = getelementptr inbounds i8, ptr [[T0:%.*]], i64 4
+; SSE-NEXT: [[T6:%.*]] = getelementptr inbounds i8, ptr [[T1:%.*]], i64 44
+; SSE-NEXT: [[T9:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 8
+; SSE-NEXT: [[T10:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 16
+; SSE-NEXT: [[T13:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 12
+; SSE-NEXT: [[T14:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 60
+; SSE-NEXT: [[T17:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 16
+; SSE-NEXT: [[T18:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 72
+; SSE-NEXT: [[T21:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 20
+; SSE-NEXT: [[T22:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 36
+; SSE-NEXT: [[T25:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 24
+; SSE-NEXT: [[T26:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 24
+; SSE-NEXT: [[T29:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 28
+; SSE-NEXT: [[T30:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 84
; SSE-NEXT: [[T3:%.*]] = load i32, ptr [[T1]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[T7:%.*]] = load i32, ptr [[T6]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[T11:%.*]] = load i32, ptr [[T10]], align 4, !tbaa [[TBAA0]]
@@ -411,13 +411,13 @@ define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture read
; SSE-NEXT: ret void
;
; AVX-LABEL: @gather_load_4(
-; AVX-NEXT: [[T6:%.*]] = getelementptr inbounds i32, ptr [[T1:%.*]], i64 11
-; AVX-NEXT: [[T10:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 4
-; AVX-NEXT: [[T14:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 15
-; AVX-NEXT: [[T18:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 18
-; AVX-NEXT: [[T22:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 9
-; AVX-NEXT: [[T26:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 6
-; AVX-NEXT: [[T30:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 21
+; AVX-NEXT: [[T6:%.*]] = getelementptr inbounds i8, ptr [[T1:%.*]], i64 44
+; AVX-NEXT: [[T10:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 16
+; AVX-NEXT: [[T14:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 60
+; AVX-NEXT: [[T18:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 72
+; AVX-NEXT: [[T22:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 36
+; AVX-NEXT: [[T26:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 24
+; AVX-NEXT: [[T30:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 84
; AVX-NEXT: [[T3:%.*]] = load i32, ptr [[T1]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[T7:%.*]] = load i32, ptr [[T6]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[T11:%.*]] = load i32, ptr [[T10]], align 4, !tbaa [[TBAA0]]
@@ -439,13 +439,13 @@ define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture read
; AVX-NEXT: ret void
;
; AVX2-LABEL: @gather_load_4(
-; AVX2-NEXT: [[T6:%.*]] = getelementptr inbounds i32, ptr [[T1:%.*]], i64 11
-; AVX2-NEXT: [[T10:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 4
-; AVX2-NEXT: [[T14:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 15
-; AVX2-NEXT: [[T18:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 18
-; AVX2-NEXT: [[T22:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 9
-; AVX2-NEXT: [[T26:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 6
-; AVX2-NEXT: [[T30:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 21
+; AVX2-NEXT: [[T6:%.*]] = getelementptr inbounds i8, ptr [[T1:%.*]], i64 44
+; AVX2-NEXT: [[T10:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 16
+; AVX2-NEXT: [[T14:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 60
+; AVX2-NEXT: [[T18:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 72
+; AVX2-NEXT: [[T22:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 36
+; AVX2-NEXT: [[T26:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 24
+; AVX2-NEXT: [[T30:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 84
; AVX2-NEXT: [[T3:%.*]] = load i32, ptr [[T1]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[T7:%.*]] = load i32, ptr [[T6]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[T11:%.*]] = load i32, ptr [[T10]], align 4, !tbaa [[TBAA0]]
@@ -533,21 +533,21 @@ define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture read
define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture readonly %1) {
; SSE-LABEL: @gather_load_div(
; SSE-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 4
+; SSE-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; SSE-NEXT: [[TMP5:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 10
+; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; SSE-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 13
+; SSE-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 52
; SSE-NEXT: [[TMP9:%.*]] = load float, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 3
+; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; SSE-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 11
+; SSE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; SSE-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 14
+; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 56
; SSE-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 44
+; SSE-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 176
; SSE-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP0:%.*]], i64 4
+; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP0:%.*]], i64 16
; SSE-NEXT: [[TMP19:%.*]] = insertelement <4 x float> poison, float [[TMP3]], i64 0
; SSE-NEXT: [[TMP20:%.*]] = insertelement <4 x float> [[TMP19]], float [[TMP7]], i64 1
; SSE-NEXT: [[TMP21:%.*]] = insertelement <4 x float> [[TMP20]], float [[TMP11]], i64 2
@@ -558,21 +558,21 @@ define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture rea
; SSE-NEXT: [[TMP26:%.*]] = insertelement <4 x float> [[TMP25]], float [[TMP17]], i64 3
; SSE-NEXT: [[TMP27:%.*]] = fdiv <4 x float> [[TMP22]], [[TMP26]]
; SSE-NEXT: store <4 x float> [[TMP27]], ptr [[TMP0]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 17
+; SSE-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 68
; SSE-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP28]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 33
+; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 132
; SSE-NEXT: [[TMP31:%.*]] = load float, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 8
+; SSE-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 32
; SSE-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP34:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 30
+; SSE-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 120
; SSE-NEXT: [[TMP35:%.*]] = load float, ptr [[TMP34]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 5
+; SSE-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; SSE-NEXT: [[TMP37:%.*]] = load float, ptr [[TMP36]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP38:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 27
+; SSE-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 108
; SSE-NEXT: [[TMP39:%.*]] = load float, ptr [[TMP38]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 20
+; SSE-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 80
; SSE-NEXT: [[TMP41:%.*]] = load float, ptr [[TMP40]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP42:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 23
+; SSE-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 92
; SSE-NEXT: [[TMP43:%.*]] = load float, ptr [[TMP42]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP44:%.*]] = insertelement <4 x float> poison, float [[TMP29]], i64 0
; SSE-NEXT: [[TMP45:%.*]] = insertelement <4 x float> [[TMP44]], float [[TMP33]], i64 1
@@ -588,35 +588,35 @@ define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture rea
;
; AVX-LABEL: @gather_load_div(
; AVX-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 4
+; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX-NEXT: [[TMP5:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 10
+; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 13
+; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 52
; AVX-NEXT: [[TMP9:%.*]] = load float, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 3
+; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 11
+; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 14
+; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 56
; AVX-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 44
+; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 176
; AVX-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 17
+; AVX-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 68
; AVX-NEXT: [[TMP19:%.*]] = load float, ptr [[TMP18]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 33
+; AVX-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 132
; AVX-NEXT: [[TMP21:%.*]] = load float, ptr [[TMP20]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 8
+; AVX-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 32
; AVX-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP24:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 30
+; AVX-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 120
; AVX-NEXT: [[TMP25:%.*]] = load float, ptr [[TMP24]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 5
+; AVX-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX-NEXT: [[TMP27:%.*]] = load float, ptr [[TMP26]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 27
+; AVX-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 108
; AVX-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP28]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP30:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 20
+; AVX-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 80
; AVX-NEXT: [[TMP31:%.*]] = load float, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 23
+; AVX-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 92
; AVX-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP34:%.*]] = insertelement <8 x float> poison, float [[TMP3]], i64 0
; AVX-NEXT: [[TMP35:%.*]] = insertelement <8 x float> [[TMP34]], float [[TMP7]], i64 1
@@ -640,35 +640,35 @@ define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture rea
;
; AVX2-LABEL: @gather_load_div(
; AVX2-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 4
+; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX2-NEXT: [[TMP5:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 10
+; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX2-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 13
+; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 52
; AVX2-NEXT: [[TMP9:%.*]] = load float, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 3
+; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX2-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 11
+; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX2-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 14
+; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 56
; AVX2-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 44
+; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 176
; AVX2-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 17
+; AVX2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 68
; AVX2-NEXT: [[TMP19:%.*]] = load float, ptr [[TMP18]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 33
+; AVX2-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 132
; AVX2-NEXT: [[TMP21:%.*]] = load float, ptr [[TMP20]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 8
+; AVX2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 32
; AVX2-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP24:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 30
+; AVX2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 120
; AVX2-NEXT: [[TMP25:%.*]] = load float, ptr [[TMP24]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 5
+; AVX2-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX2-NEXT: [[TMP27:%.*]] = load float, ptr [[TMP26]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 27
+; AVX2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 108
; AVX2-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP28]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP30:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 20
+; AVX2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 80
; AVX2-NEXT: [[TMP31:%.*]] = load float, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 23
+; AVX2-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 92
; AVX2-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP34:%.*]] = insertelement <8 x float> poison, float [[TMP3]], i64 0
; AVX2-NEXT: [[TMP35:%.*]] = insertelement <8 x float> [[TMP34]], float [[TMP7]], i64 1
diff --git a/llvm/test/Transforms/SLPVectorizer/X86/pr47629.ll b/llvm/test/Transforms/SLPVectorizer/X86/pr47629.ll
index 500856a0d66a92..09d6c77557efaa 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/pr47629.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/pr47629.ll
@@ -7,11 +7,11 @@
define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonly %1) {
; SSE-LABEL: @gather_load(
-; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; SSE-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; SSE-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; SSE-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -23,11 +23,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; SSE-NEXT: ret void
;
; AVX-LABEL: @gather_load(
-; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -39,11 +39,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; AVX-NEXT: ret void
;
; AVX2-LABEL: @gather_load(
-; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -55,11 +55,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; AVX2-NEXT: ret void
;
; AVX512F-LABEL: @gather_load(
-; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX512F-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX512F-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX512F-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX512F-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX512F-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -71,11 +71,11 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
; AVX512F-NEXT: ret void
;
; AVX512VL-LABEL: @gather_load(
-; AVX512VL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX512VL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX512VL-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP1]], align 4, !tbaa [[TBAA0:![0-9]+]]
-; AVX512VL-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX512VL-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX512VL-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX512VL-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX512VL-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX512VL-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; AVX512VL-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; AVX512VL-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
@@ -107,35 +107,35 @@ define void @gather_load(ptr noalias nocapture %0, ptr noalias nocapture readonl
define void @gather_load_2(ptr noalias nocapture %0, ptr noalias nocapture readonly %1) {
; SSE-LABEL: @gather_load_2(
-; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; SSE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; SSE-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP5:%.*]] = add nsw i32 [[TMP4]], 1
-; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP0:%.*]], i64 1
+; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP0:%.*]], i64 4
; SSE-NEXT: store i32 [[TMP5]], ptr [[TMP0]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; SSE-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP9:%.*]] = add nsw i32 [[TMP8]], 2
-; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 2
+; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; SSE-NEXT: store i32 [[TMP9]], ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; SSE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; SSE-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP13:%.*]] = add nsw i32 [[TMP12]], 3
-; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 3
+; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 12
; SSE-NEXT: store i32 [[TMP13]], ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; SSE-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; SSE-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP17:%.*]] = add nsw i32 [[TMP16]], 4
; SSE-NEXT: store i32 [[TMP17]], ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: ret void
;
; AVX-LABEL: @gather_load_2(
-; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; AVX-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; AVX-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; AVX-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; AVX-NEXT: [[TMP12:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP6]], i64 1
@@ -146,13 +146,13 @@ define void @gather_load_2(ptr noalias nocapture %0, ptr noalias nocapture reado
; AVX-NEXT: ret void
;
; AVX2-LABEL: @gather_load_2(
-; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX2-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; AVX2-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; AVX2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; AVX2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX2-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; AVX2-NEXT: [[TMP12:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP6]], i64 1
@@ -163,13 +163,13 @@ define void @gather_load_2(ptr noalias nocapture %0, ptr noalias nocapture reado
; AVX2-NEXT: ret void
;
; AVX512F-LABEL: @gather_load_2(
-; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[TMP1:%.*]], i64 1
+; AVX512F-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1:%.*]], i64 4
; AVX512F-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 10
+; AVX512F-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX512F-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 3
+; AVX512F-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX512F-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4, !tbaa [[TBAA0]]
-; AVX512F-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 5
+; AVX512F-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX512F-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
; AVX512F-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; AVX512F-NEXT: [[TMP12:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP6]], i64 1
@@ -215,39 +215,39 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
; SSE-LABEL: @gather_load_3(
; SSE-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP4:%.*]] = add i32 [[TMP3]], 1
-; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP0:%.*]], i64 1
+; SSE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP0:%.*]], i64 4
; SSE-NEXT: store i32 [[TMP4]], ptr [[TMP0]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; SSE-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP8:%.*]] = add i32 [[TMP7]], 2
-; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 2
+; SSE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 8
; SSE-NEXT: store i32 [[TMP8]], ptr [[TMP5]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; SSE-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP12:%.*]] = add i32 [[TMP11]], 3
-; SSE-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 3
+; SSE-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 12
; SSE-NEXT: store i32 [[TMP12]], ptr [[TMP9]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 15
+; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 60
; SSE-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP16:%.*]] = add i32 [[TMP15]], 4
-; SSE-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 4
+; SSE-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 16
; SSE-NEXT: store i32 [[TMP16]], ptr [[TMP13]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 18
+; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 72
; SSE-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP20:%.*]] = add i32 [[TMP19]], 1
-; SSE-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 5
+; SSE-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 20
; SSE-NEXT: store i32 [[TMP20]], ptr [[TMP17]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 9
+; SSE-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 36
; SSE-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP24:%.*]] = add i32 [[TMP23]], 2
-; SSE-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 6
+; SSE-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 24
; SSE-NEXT: store i32 [[TMP24]], ptr [[TMP21]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 6
+; SSE-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 24
; SSE-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP28:%.*]] = add i32 [[TMP27]], 3
-; SSE-NEXT: [[TMP29:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 7
+; SSE-NEXT: [[TMP29:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 28
; SSE-NEXT: store i32 [[TMP28]], ptr [[TMP25]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 21
+; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 84
; SSE-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP32:%.*]] = add i32 [[TMP31]], 4
; SSE-NEXT: store i32 [[TMP32]], ptr [[TMP29]], align 4, !tbaa [[TBAA0]]
@@ -255,19 +255,19 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
;
; AVX-LABEL: @gather_load_3(
; AVX-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 15
+; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 60
; AVX-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 18
+; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 72
; AVX-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 9
+; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 36
; AVX-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 6
+; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 24
; AVX-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 21
+; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 84
; AVX-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP18:%.*]] = insertelement <8 x i32> poison, i32 [[TMP3]], i64 0
; AVX-NEXT: [[TMP19:%.*]] = insertelement <8 x i32> [[TMP18]], i32 [[TMP5]], i64 1
@@ -283,19 +283,19 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
;
; AVX2-LABEL: @gather_load_3(
; AVX2-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 11
+; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX2-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 4
+; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX2-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 15
+; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 60
; AVX2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 18
+; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 72
; AVX2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 9
+; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 36
; AVX2-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 6
+; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 24
; AVX2-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 21
+; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 84
; AVX2-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP18:%.*]] = insertelement <8 x i32> poison, i32 [[TMP3]], i64 0
; AVX2-NEXT: [[TMP19:%.*]] = insertelement <8 x i32> [[TMP18]], i32 [[TMP5]], i64 1
@@ -370,20 +370,20 @@ define void @gather_load_3(ptr noalias nocapture %0, ptr noalias nocapture reado
define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture readonly %t1) {
; SSE-LABEL: @gather_load_4(
-; SSE-NEXT: [[T5:%.*]] = getelementptr inbounds i32, ptr [[T0:%.*]], i64 1
-; SSE-NEXT: [[T6:%.*]] = getelementptr inbounds i32, ptr [[T1:%.*]], i64 11
-; SSE-NEXT: [[T9:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 2
-; SSE-NEXT: [[T10:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 4
-; SSE-NEXT: [[T13:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 3
-; SSE-NEXT: [[T14:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 15
-; SSE-NEXT: [[T17:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 4
-; SSE-NEXT: [[T18:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 18
-; SSE-NEXT: [[T21:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 5
-; SSE-NEXT: [[T22:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 9
-; SSE-NEXT: [[T25:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 6
-; SSE-NEXT: [[T26:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 6
-; SSE-NEXT: [[T29:%.*]] = getelementptr inbounds i32, ptr [[T0]], i64 7
-; SSE-NEXT: [[T30:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 21
+; SSE-NEXT: [[T5:%.*]] = getelementptr inbounds i8, ptr [[T0:%.*]], i64 4
+; SSE-NEXT: [[T6:%.*]] = getelementptr inbounds i8, ptr [[T1:%.*]], i64 44
+; SSE-NEXT: [[T9:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 8
+; SSE-NEXT: [[T10:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 16
+; SSE-NEXT: [[T13:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 12
+; SSE-NEXT: [[T14:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 60
+; SSE-NEXT: [[T17:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 16
+; SSE-NEXT: [[T18:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 72
+; SSE-NEXT: [[T21:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 20
+; SSE-NEXT: [[T22:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 36
+; SSE-NEXT: [[T25:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 24
+; SSE-NEXT: [[T26:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 24
+; SSE-NEXT: [[T29:%.*]] = getelementptr inbounds i8, ptr [[T0]], i64 28
+; SSE-NEXT: [[T30:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 84
; SSE-NEXT: [[T3:%.*]] = load i32, ptr [[T1]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[T7:%.*]] = load i32, ptr [[T6]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[T11:%.*]] = load i32, ptr [[T10]], align 4, !tbaa [[TBAA0]]
@@ -411,13 +411,13 @@ define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture read
; SSE-NEXT: ret void
;
; AVX-LABEL: @gather_load_4(
-; AVX-NEXT: [[T6:%.*]] = getelementptr inbounds i32, ptr [[T1:%.*]], i64 11
-; AVX-NEXT: [[T10:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 4
-; AVX-NEXT: [[T14:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 15
-; AVX-NEXT: [[T18:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 18
-; AVX-NEXT: [[T22:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 9
-; AVX-NEXT: [[T26:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 6
-; AVX-NEXT: [[T30:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 21
+; AVX-NEXT: [[T6:%.*]] = getelementptr inbounds i8, ptr [[T1:%.*]], i64 44
+; AVX-NEXT: [[T10:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 16
+; AVX-NEXT: [[T14:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 60
+; AVX-NEXT: [[T18:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 72
+; AVX-NEXT: [[T22:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 36
+; AVX-NEXT: [[T26:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 24
+; AVX-NEXT: [[T30:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 84
; AVX-NEXT: [[T3:%.*]] = load i32, ptr [[T1]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[T7:%.*]] = load i32, ptr [[T6]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[T11:%.*]] = load i32, ptr [[T10]], align 4, !tbaa [[TBAA0]]
@@ -439,13 +439,13 @@ define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture read
; AVX-NEXT: ret void
;
; AVX2-LABEL: @gather_load_4(
-; AVX2-NEXT: [[T6:%.*]] = getelementptr inbounds i32, ptr [[T1:%.*]], i64 11
-; AVX2-NEXT: [[T10:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 4
-; AVX2-NEXT: [[T14:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 15
-; AVX2-NEXT: [[T18:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 18
-; AVX2-NEXT: [[T22:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 9
-; AVX2-NEXT: [[T26:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 6
-; AVX2-NEXT: [[T30:%.*]] = getelementptr inbounds i32, ptr [[T1]], i64 21
+; AVX2-NEXT: [[T6:%.*]] = getelementptr inbounds i8, ptr [[T1:%.*]], i64 44
+; AVX2-NEXT: [[T10:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 16
+; AVX2-NEXT: [[T14:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 60
+; AVX2-NEXT: [[T18:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 72
+; AVX2-NEXT: [[T22:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 36
+; AVX2-NEXT: [[T26:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 24
+; AVX2-NEXT: [[T30:%.*]] = getelementptr inbounds i8, ptr [[T1]], i64 84
; AVX2-NEXT: [[T3:%.*]] = load i32, ptr [[T1]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[T7:%.*]] = load i32, ptr [[T6]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[T11:%.*]] = load i32, ptr [[T10]], align 4, !tbaa [[TBAA0]]
@@ -533,21 +533,21 @@ define void @gather_load_4(ptr noalias nocapture %t0, ptr noalias nocapture read
define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture readonly %1) {
; SSE-LABEL: @gather_load_div(
; SSE-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 4
+; SSE-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; SSE-NEXT: [[TMP5:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 10
+; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; SSE-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 13
+; SSE-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 52
; SSE-NEXT: [[TMP9:%.*]] = load float, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 3
+; SSE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; SSE-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 11
+; SSE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; SSE-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 14
+; SSE-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 56
; SSE-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 44
+; SSE-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 176
; SSE-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP0:%.*]], i64 4
+; SSE-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP0:%.*]], i64 16
; SSE-NEXT: [[TMP19:%.*]] = insertelement <4 x float> poison, float [[TMP3]], i64 0
; SSE-NEXT: [[TMP20:%.*]] = insertelement <4 x float> [[TMP19]], float [[TMP7]], i64 1
; SSE-NEXT: [[TMP21:%.*]] = insertelement <4 x float> [[TMP20]], float [[TMP11]], i64 2
@@ -558,21 +558,21 @@ define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture rea
; SSE-NEXT: [[TMP26:%.*]] = insertelement <4 x float> [[TMP25]], float [[TMP17]], i64 3
; SSE-NEXT: [[TMP27:%.*]] = fdiv <4 x float> [[TMP22]], [[TMP26]]
; SSE-NEXT: store <4 x float> [[TMP27]], ptr [[TMP0]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 17
+; SSE-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 68
; SSE-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP28]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 33
+; SSE-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 132
; SSE-NEXT: [[TMP31:%.*]] = load float, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 8
+; SSE-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 32
; SSE-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP34:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 30
+; SSE-NEXT: [[TMP34:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 120
; SSE-NEXT: [[TMP35:%.*]] = load float, ptr [[TMP34]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP36:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 5
+; SSE-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; SSE-NEXT: [[TMP37:%.*]] = load float, ptr [[TMP36]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP38:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 27
+; SSE-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 108
; SSE-NEXT: [[TMP39:%.*]] = load float, ptr [[TMP38]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 20
+; SSE-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 80
; SSE-NEXT: [[TMP41:%.*]] = load float, ptr [[TMP40]], align 4, !tbaa [[TBAA0]]
-; SSE-NEXT: [[TMP42:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 23
+; SSE-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 92
; SSE-NEXT: [[TMP43:%.*]] = load float, ptr [[TMP42]], align 4, !tbaa [[TBAA0]]
; SSE-NEXT: [[TMP44:%.*]] = insertelement <4 x float> poison, float [[TMP29]], i64 0
; SSE-NEXT: [[TMP45:%.*]] = insertelement <4 x float> [[TMP44]], float [[TMP33]], i64 1
@@ -588,35 +588,35 @@ define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture rea
;
; AVX-LABEL: @gather_load_div(
; AVX-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 4
+; AVX-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX-NEXT: [[TMP5:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 10
+; AVX-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 13
+; AVX-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 52
; AVX-NEXT: [[TMP9:%.*]] = load float, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 3
+; AVX-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 11
+; AVX-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 14
+; AVX-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 56
; AVX-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 44
+; AVX-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 176
; AVX-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 17
+; AVX-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 68
; AVX-NEXT: [[TMP19:%.*]] = load float, ptr [[TMP18]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 33
+; AVX-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 132
; AVX-NEXT: [[TMP21:%.*]] = load float, ptr [[TMP20]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 8
+; AVX-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 32
; AVX-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP24:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 30
+; AVX-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 120
; AVX-NEXT: [[TMP25:%.*]] = load float, ptr [[TMP24]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 5
+; AVX-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX-NEXT: [[TMP27:%.*]] = load float, ptr [[TMP26]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 27
+; AVX-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 108
; AVX-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP28]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP30:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 20
+; AVX-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 80
; AVX-NEXT: [[TMP31:%.*]] = load float, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
-; AVX-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 23
+; AVX-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 92
; AVX-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4, !tbaa [[TBAA0]]
; AVX-NEXT: [[TMP34:%.*]] = insertelement <8 x float> poison, float [[TMP3]], i64 0
; AVX-NEXT: [[TMP35:%.*]] = insertelement <8 x float> [[TMP34]], float [[TMP7]], i64 1
@@ -640,35 +640,35 @@ define void @gather_load_div(ptr noalias nocapture %0, ptr noalias nocapture rea
;
; AVX2-LABEL: @gather_load_div(
; AVX2-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP1:%.*]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 4
+; AVX2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16
; AVX2-NEXT: [[TMP5:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 10
+; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 40
; AVX2-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP6]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 13
+; AVX2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 52
; AVX2-NEXT: [[TMP9:%.*]] = load float, ptr [[TMP8]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 3
+; AVX2-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 12
; AVX2-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP10]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 11
+; AVX2-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 44
; AVX2-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 14
+; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 56
; AVX2-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP14]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 44
+; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 176
; AVX2-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP16]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 17
+; AVX2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 68
; AVX2-NEXT: [[TMP19:%.*]] = load float, ptr [[TMP18]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 33
+; AVX2-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 132
; AVX2-NEXT: [[TMP21:%.*]] = load float, ptr [[TMP20]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 8
+; AVX2-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 32
; AVX2-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP24:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 30
+; AVX2-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 120
; AVX2-NEXT: [[TMP25:%.*]] = load float, ptr [[TMP24]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP26:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 5
+; AVX2-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 20
; AVX2-NEXT: [[TMP27:%.*]] = load float, ptr [[TMP26]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 27
+; AVX2-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 108
; AVX2-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP28]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP30:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 20
+; AVX2-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 80
; AVX2-NEXT: [[TMP31:%.*]] = load float, ptr [[TMP30]], align 4, !tbaa [[TBAA0]]
-; AVX2-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 23
+; AVX2-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 92
; AVX2-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4, !tbaa [[TBAA0]]
; AVX2-NEXT: [[TMP34:%.*]] = insertelement <8 x float> poison, float [[TMP3]], i64 0
; AVX2-NEXT: [[TMP35:%.*]] = insertelement <8 x float> [[TMP34]], float [[TMP7]], i64 1
diff --git a/llvm/test/Transforms/SampleProfile/pseudo-probe-instcombine.ll b/llvm/test/Transforms/SampleProfile/pseudo-probe-instcombine.ll
index f015bae700754b..d1c15306a72d08 100644
--- a/llvm/test/Transforms/SampleProfile/pseudo-probe-instcombine.ll
+++ b/llvm/test/Transforms/SampleProfile/pseudo-probe-instcombine.ll
@@ -18,9 +18,9 @@
define dso_local void @merge(ptr nocapture readonly %params) local_unnamed_addr align 2 {
; CHECK-LABEL: @merge(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[SAVEPAIRLISTS3:%.*]] = getelementptr inbounds [[STRUCT_NONBONDED:%.*]], ptr [[PARAMS:%.*]], i64 0, i32 11
+; CHECK-NEXT: [[SAVEPAIRLISTS3:%.*]] = getelementptr inbounds i8, ptr [[PARAMS:%.*]], i64 144
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[SAVEPAIRLISTS3]], align 8
-; CHECK-NEXT: [[USEPAIRLISTS4:%.*]] = getelementptr inbounds [[STRUCT_NONBONDED]], ptr [[PARAMS]], i64 0, i32 12
+; CHECK-NEXT: [[USEPAIRLISTS4:%.*]] = getelementptr inbounds i8, ptr [[PARAMS]], i64 148
; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[USEPAIRLISTS4]], align 4
; CHECK-NEXT: [[TOBOOL54_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
; CHECK-NEXT: br i1 [[TOBOOL54_NOT]], label [[LOR_LHS_FALSE55:%.*]], label [[IF_END109:%.*]]
@@ -37,7 +37,7 @@ define dso_local void @merge(ptr nocapture readonly %params) local_unnamed_addr
; CHECK: if.then117:
; CHECK-NEXT: ret void
; CHECK: if.then138:
-; CHECK-NEXT: [[DOTIN:%.*]] = getelementptr inbounds [[STRUCT_NONBONDED]], ptr [[PARAMS]], i64 0, i32 16
+; CHECK-NEXT: [[DOTIN:%.*]] = getelementptr inbounds i8, ptr [[PARAMS]], i64 172
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTIN]], align 4
; CHECK-NEXT: [[TOBOOL139_NOT:%.*]] = icmp eq i32 [[TMP2]], 0
; CHECK-NEXT: br i1 [[TOBOOL139_NOT]], label [[IF_ELSE147:%.*]], label [[IF_THEN140:%.*]]
@@ -89,9 +89,9 @@ if.else147: ; preds = %if.then138
;; Check the last store is deleted.
define i32 @load(ptr nocapture %a, ptr nocapture %b) {
; CHECK-LABEL: @load(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 4
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 8
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 1
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 4
; CHECK-NEXT: store i32 [[TMP2]], ptr [[TMP3]], align 8
; CHECK-NEXT: call void @llvm.pseudoprobe(i64 5116412291814990879, i64 1, i32 0, i64 -1)
; CHECK-NEXT: ret i32 [[TMP2]]
diff --git a/llvm/test/Transforms/Util/strip-gc-relocates.ll b/llvm/test/Transforms/Util/strip-gc-relocates.ll
index 552e3b49dd493b..c6013bc54fbe23 100644
--- a/llvm/test/Transforms/Util/strip-gc-relocates.ll
+++ b/llvm/test/Transforms/Util/strip-gc-relocates.ll
@@ -28,7 +28,7 @@ entry:
define void @test2(ptr addrspace(1) %base) gc "statepoint-example" {
; CHECK-LABEL: @test2(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[PTR_GEP1:%.*]] = getelementptr i32, ptr addrspace(1) [[BASE:%.*]], i64 30
+; CHECK-NEXT: [[PTR_GEP1:%.*]] = getelementptr i8, ptr addrspace(1) [[BASE:%.*]], i64 120
; CHECK-NEXT: [[STATEPOINT_TOKEN:%.*]] = call token (i64, i32, ptr, i32, i32, ...) @llvm.experimental.gc.statepoint.p0(i64 2882400000, i32 0, ptr nonnull elementtype(void ()) @do_safepoint, i32 0, i32 0, i32 0, i32 0) [ "gc-live"() ]
; CHECK-NEXT: call void @use_obj32(ptr addrspace(1) [[PTR_GEP1]])
; CHECK-NEXT: [[STATEPOINT_TOKEN1:%.*]] = call token (i64, i32, ptr, i32, i32, ...) @llvm.experimental.gc.statepoint.p0(i64 2882400000, i32 0, ptr nonnull elementtype(void ()) @do_safepoint, i32 0, i32 0, i32 0, i32 0) [ "gc-live"() ]
@@ -91,7 +91,7 @@ define void @test4(i1 %cond) gc "statepoint-example" {
; CHECK: merge:
; CHECK-NEXT: [[BASEPHI_BASE:%.*]] = phi ptr addrspace(1) [ [[BASE1]], [[HERE]] ], [ [[BASE2]], [[THERE]] ], !is_base_value !0
; CHECK-NEXT: [[STATEPOINT_TOKEN:%.*]] = call token (i64, i32, ptr, i32, i32, ...) @llvm.experimental.gc.statepoint.p0(i64 2882400000, i32 0, ptr nonnull elementtype(void ()) @do_safepoint, i32 0, i32 0, i32 0, i32 0) [ "gc-live"() ]
-; CHECK-NEXT: [[PTR_GEP_REMAT:%.*]] = getelementptr i32, ptr addrspace(1) [[BASEPHI_BASE]], i64 15
+; CHECK-NEXT: [[PTR_GEP_REMAT:%.*]] = getelementptr i8, ptr addrspace(1) [[BASEPHI_BASE]], i64 60
; CHECK-NEXT: call void @use_obj32(ptr addrspace(1) [[PTR_GEP_REMAT]])
; CHECK-NEXT: ret void
;
More information about the flang-commits
mailing list