[llvm] [InstCombine] Canonicalize gep scaling factor (PR #180216)
Nikita Popov via llvm-commits
llvm-commits at lists.llvm.org
Mon Mar 2 07:16:50 PST 2026
https://github.com/nikic updated https://github.com/llvm/llvm-project/pull/180216
>From bb73c88e073aae01934d850fb7d3f915506c5a53 Mon Sep 17 00:00:00 2001
From: Nikita Popov <npopov at redhat.com>
Date: Fri, 6 Feb 2026 16:32:30 +0100
Subject: [PATCH] [InstCombine] Canonicalize gep scaling factor
---
.../InstCombine/InstructionCombining.cpp | 16 +
llvm/test/Analysis/BasicAA/featuretest.ll | 15 +-
.../CodeGen/AArch64/sme-aarch64-svcount-O3.ll | 3 +-
.../amdgpu-uniform-intrinsic-combine.ll | 30 +-
.../Hexagon/autohvx/vector-align-tbaa.ll | 27 +-
.../InstCombine/2006-12-15-Range-Test.ll | 6 +-
.../AArch64/sve-intrinsic-gatherscatter.ll | 12 +-
.../AMDGPU/canonicalize-add-to-gep.ll | 4 +-
.../AMDGPU/memcpy-from-constant.ll | 6 +-
.../test/Transforms/InstCombine/align-addr.ll | 7 +-
.../Transforms/InstCombine/and-or-icmps.ll | 6 +-
llvm/test/Transforms/InstCombine/array.ll | 106 ++--
.../InstCombine/assume-loop-align.ll | 6 +-
.../InstCombine/canonicalize-gep-constglob.ll | 12 +-
.../InstCombine/canonicalize-gep-mul.ll | 3 +-
llvm/test/Transforms/InstCombine/cast.ll | 3 +-
llvm/test/Transforms/InstCombine/cast_phi.ll | 12 +-
llvm/test/Transforms/InstCombine/cast_ptr.ll | 29 +-
.../constant-fold-address-space-pointer.ll | 3 +-
.../Transforms/InstCombine/dependent-ivs.ll | 3 +-
llvm/test/Transforms/InstCombine/fmul.ll | 3 +-
.../Transforms/InstCombine/gep-addrspace.ll | 6 +-
llvm/test/Transforms/InstCombine/gep-alias.ll | 3 +-
.../gep-canonicalize-constant-indices.ll | 60 +-
.../InstCombine/gep-combine-loop-invariant.ll | 22 +-
.../Transforms/InstCombine/gep-custom-dl.ll | 12 +-
.../gep-fold-chained-const-select.ll | 5 +-
.../InstCombine/gep-merge-constant-indices.ll | 12 +-
llvm/test/Transforms/InstCombine/gep-sext.ll | 12 +-
.../InstCombine/gep-srem-to-and-deref.ll | 5 +-
.../InstCombine/gep-vector-indices.ll | 24 +-
.../test/Transforms/InstCombine/gep-vector.ll | 48 +-
.../Transforms/InstCombine/gepofconstgepi8.ll | 97 ++-
llvm/test/Transforms/InstCombine/gepphigep.ll | 28 +-
.../Transforms/InstCombine/getelementptr.ll | 171 ++++--
.../Transforms/InstCombine/icmp-custom-dl.ll | 10 +-
llvm/test/Transforms/InstCombine/icmp-gep.ll | 70 ++-
.../test/Transforms/InstCombine/icmp-trunc.ll | 8 +-
.../InstCombine/indexed-gep-compares.ll | 9 +-
.../InstCombine/known-phi-recurse.ll | 3 +-
.../InstCombine/load-bitcast-select.ll | 6 +-
llvm/test/Transforms/InstCombine/load-cmp.ll | 262 +++++---
llvm/test/Transforms/InstCombine/load.ll | 12 +-
.../InstCombine/loadstore-alignment.ll | 18 +-
.../InstCombine/loadstore-metadata.ll | 6 +-
.../masked_intrinsics-inseltpoison.ll | 10 +-
.../InstCombine/masked_intrinsics.ll | 10 +-
.../Transforms/InstCombine/mem-gep-zidx.ll | 20 +-
.../InstCombine/memcpy-addrspace.ll | 48 +-
.../InstCombine/memcpy-from-global.ll | 23 +-
.../multi-size-address-space-pointer.ll | 6 +-
llvm/test/Transforms/InstCombine/narrow.ll | 6 +-
.../test/Transforms/InstCombine/opaque-ptr.ll | 43 +-
llvm/test/Transforms/InstCombine/phi.ll | 3 +-
.../InstCombine/pr38984-inseltpoison.ll | 13 +-
llvm/test/Transforms/InstCombine/pr38984.ll | 13 +-
llvm/test/Transforms/InstCombine/pr58901.ll | 9 +-
.../InstCombine/ptrtoint-nullgep.ll | 38 +-
.../remove-loop-phi-multiply-by-zero.ll | 30 +-
.../Transforms/InstCombine/scalable-select.ll | 6 +-
.../test/Transforms/InstCombine/select-gep.ll | 94 +--
llvm/test/Transforms/InstCombine/shift.ll | 24 +-
.../InstCombine/sink_instruction.ll | 9 +-
llvm/test/Transforms/InstCombine/store.ll | 12 +-
llvm/test/Transforms/InstCombine/strcmp-3.ll | 6 +-
llvm/test/Transforms/InstCombine/sub-gep.ll | 30 +-
.../vec_demanded_elts-inseltpoison.ll | 21 +-
.../InstCombine/vec_demanded_elts.ll | 21 +-
.../vec_phi_extract-inseltpoison.ll | 9 +-
.../Transforms/InstCombine/vec_phi_extract.ll | 9 +-
.../InstCombine/vector_gep1-inseltpoison.ll | 2 +-
.../Transforms/InstCombine/vector_gep1.ll | 2 +-
.../Transforms/InstCombine/vector_gep2.ll | 6 +-
.../Transforms/InstCombine/vectorgep-crash.ll | 7 +-
llvm/test/Transforms/InstCombine/wcslen-1.ll | 46 +-
llvm/test/Transforms/InstCombine/wcslen-3.ll | 20 +-
llvm/test/Transforms/InstCombine/wcslen-5.ll | 24 +-
.../Transforms/LoopSimplify/merge-exits.ll | 12 +-
.../AArch64/runtime-unroll-generic.ll | 64 +-
.../LoopUnroll/WebAssembly/basic-unrolling.ll | 60 +-
.../LoopUnroll/runtime-multiexit-heuristic.ll | 63 +-
.../LoopUnroll/runtime-unroll-remainder.ll | 42 +-
.../AArch64/aarch64-predication.ll | 5 +-
.../AArch64/sve-cond-inv-loads.ll | 21 +-
.../LoopVectorize/AArch64/sve-inductions.ll | 6 +-
.../AArch64/sve-interleaved-accesses.ll | 174 ++++--
.../sve-tail-folding-overflow-checks.ll | 12 +-
.../AArch64/sve-vector-reverse.ll | 36 +-
.../LoopVectorize/AArch64/sve-vfabi.ll | 41 +-
.../LoopVectorize/AArch64/sve-widen-phi.ll | 26 +-
.../AArch64/sve2-histcnt-epilogue.ll | 16 +-
.../sve2-histcnt-no-scalar-interleave.ll | 6 +-
.../AArch64/sve2-histcnt-too-many-deps.ll | 30 +-
.../LoopVectorize/AArch64/sve2-histcnt.ll | 105 ++--
.../AArch64/uniform-args-call-variants.ll | 132 ++--
.../AArch64/vector-reverse-mask4.ll | 12 +-
.../LoopVectorize/AMDGPU/packed-math.ll | 9 +-
.../ARM/mve-reduction-predselect.ll | 60 +-
.../ARM/mve-reductions-interleave.ll | 18 +-
.../LoopVectorize/ARM/mve-reductions.ll | 142 +++--
.../ARM/tail-fold-multiple-icmps.ll | 6 +-
.../LoopVectorize/PowerPC/vsx-tsvc-s173.ll | 9 +-
.../PowerPC/widened-massv-call.ll | 3 +-
.../PowerPC/widened-massv-vfabi-attr.ll | 3 +-
.../LoopVectorize/SystemZ/addressing.ll | 11 +-
.../X86/consecutive-ptr-uniforms.ll | 9 +-
.../LoopVectorize/X86/interleaving.ll | 35 +-
.../LoopVectorize/X86/metadata-enable.ll | 65 +-
.../LoopVectorize/X86/parallel-loops.ll | 78 ++-
.../LoopVectorize/X86/small-size.ll | 59 +-
...86-interleaved-store-accesses-with-gaps.ll | 102 ++--
.../Transforms/LoopVectorize/bsd_regex.ll | 12 +-
.../LoopVectorize/extract-last-veclane.ll | 24 +-
.../LoopVectorize/float-induction.ll | 273 ++++++---
.../LoopVectorize/forked-pointers.ll | 36 +-
.../Transforms/LoopVectorize/histograms.ll | 15 +-
.../LoopVectorize/if-conversion-nest.ll | 38 +-
.../Transforms/LoopVectorize/if-conversion.ll | 24 +-
.../Transforms/LoopVectorize/induction.ll | 489 +++++++++------
.../LoopVectorize/interleaved-accesses-2.ll | 6 +-
.../interleaved-accesses-pred-stores.ll | 54 +-
.../LoopVectorize/interleaved-accesses.ll | 265 +++++---
.../invariant-store-vectorization-2.ll | 26 +-
.../invariant-store-vectorization.ll | 57 +-
.../Transforms/LoopVectorize/loop-scalars.ll | 54 +-
.../Transforms/LoopVectorize/non-const-n.ll | 9 +-
.../test/Transforms/LoopVectorize/phi-cost.ll | 56 +-
.../LoopVectorize/reduction-inloop-cond.ll | 252 +++++---
.../LoopVectorize/reduction-inloop-min-max.ll | 86 +--
.../LoopVectorize/reduction-inloop-pred.ll | 270 ++++++---
.../LoopVectorize/reduction-inloop-uf4.ll | 105 ++--
.../LoopVectorize/reduction-predselect.ll | 204 ++++---
.../Transforms/LoopVectorize/reduction.ll | 160 +++--
.../Transforms/LoopVectorize/runtime-check.ll | 117 ++--
.../LoopVectorize/scalable-inductions.ll | 48 +-
.../LoopVectorize/trunc-reductions.ll | 18 +-
.../uniform-args-call-variants.ll | 54 +-
.../Transforms/LoopVectorize/vector-geps.ll | 18 +-
.../LoopVersioningLICM/loopversioningLICM1.ll | 9 +-
.../multiply-fused-loops.ll | 36 +-
.../AArch64/block_scaling_decompr_8bit.ll | 6 +-
.../constraint-elimination-placement.ll | 3 +-
.../AArch64/extra-unroll-simplifications.ll | 38 +-
.../AArch64/hoist-load-from-vector-loop.ll | 19 +-
.../AArch64/hoist-runtime-checks.ll | 15 +-
...ting-sinking-required-for-vectorization.ll | 24 +-
.../AArch64/indvars-vectorization.ll | 96 +--
.../PhaseOrdering/AArch64/interleave_vec.ll | 249 ++++----
.../AArch64/interleavevectorization.ll | 42 +-
.../PhaseOrdering/AArch64/loopflatten.ll | 3 +-
.../AArch64/matrix-extract-insert.ll | 72 ++-
...ple-unreachable-exits-for-vectorization.ll | 27 +-
.../AArch64/predicated-reduction.ll | 12 +-
.../PhaseOrdering/AArch64/quant_4x4.ll | 459 +-------------
.../PhaseOrdering/AArch64/reduce_submuladd.ll | 168 +++---
.../AArch64/sinking-vs-if-conversion.ll | 10 +-
.../PhaseOrdering/SystemZ/sub-xor.ll | 181 +++---
.../X86/SROA-after-final-loop-unrolling-2.ll | 53 +-
...-simplifycfg-two-entry-phi-node-folding.ll | 3 +-
.../PhaseOrdering/X86/excessive-unrolling.ll | 14 +-
.../X86/hoist-load-of-baseptr.ll | 15 +-
.../PhaseOrdering/X86/merge-functions2.ll | 5 +-
.../PhaseOrdering/X86/merge-functions3.ll | 5 +-
.../PhaseOrdering/X86/pixel-splat.ll | 6 +-
.../Transforms/PhaseOrdering/X86/pr88239.ll | 7 +-
.../X86/preserve-access-group.ll | 45 +-
.../PhaseOrdering/X86/simplifycfg-late.ll | 5 +-
.../PhaseOrdering/X86/speculation-vs-tbaa.ll | 9 +-
.../PhaseOrdering/X86/spurious-peeling.ll | 28 +-
.../PhaseOrdering/X86/vdiv-nounroll.ll | 7 +-
.../test/Transforms/PhaseOrdering/X86/vdiv.ll | 90 +--
.../PhaseOrdering/gvn-replacement-vs-hoist.ll | 12 +-
.../PhaseOrdering/loop-access-checks.ll | 14 +-
.../PhaseOrdering/loop-vectorize-bfi.ll | 3 +-
.../PhaseOrdering/lto-argpromotion-ipsccp.ll | 3 +-
.../test/Transforms/PhaseOrdering/lto-licm.ll | 3 +-
.../PhaseOrdering/scev-custom-dl.ll | 566 ++++++++++++++++--
.../PhaseOrdering/simplifycfg-options.ll | 5 +-
.../AMDGPU/uniform-unswitch.ll | 3 +-
.../Hexagon/switch-to-lookup-table.ll | 3 +-
180 files changed, 5242 insertions(+), 3214 deletions(-)
diff --git a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
index e5c750fc6b130..c20739aac11d1 100644
--- a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
@@ -3577,6 +3577,22 @@ Instruction *InstCombinerImpl::visitGetElementPtrInst(GetElementPtrInst &GEP) {
}
}
}
+
+ if (!GEPEltType->isIntegerTy(8) && Indices.size() == 1) {
+ TypeSize Scale = DL.getTypeAllocSize(GEPEltType);
+ if (!Scale.isScalable()) {
+ Value *NewIndex = Builder.CreateMul(
+ Indices[0],
+ ConstantInt::getSigned(Indices[0]->getType(), Scale.getFixedValue(),
+ /*ImplicitTrunc=*/true),
+ "",
+ /*NUW=*/GEP.hasNoUnsignedWrap(),
+ /*NSW=*/GEP.hasNoUnsignedSignedWrap());
+ return GetElementPtrInst::Create(Type::getInt8Ty(GEP.getContext()), PtrOp,
+ NewIndex, GEP.getNoWrapFlags());
+ }
+ }
+
// We do not handle pointer-vector geps here.
if (GEPType->isVectorTy())
return nullptr;
diff --git a/llvm/test/Analysis/BasicAA/featuretest.ll b/llvm/test/Analysis/BasicAA/featuretest.ll
index 04c4725d26c1d..6dab8acb05cbb 100644
--- a/llvm/test/Analysis/BasicAA/featuretest.ll
+++ b/llvm/test/Analysis/BasicAA/featuretest.ll
@@ -20,7 +20,8 @@ define i32 @different_array_test(i64 %A, i64 %B) {
; CHECK-NEXT: [[ARRAY22:%.*]] = alloca [200 x i32], align 4
; CHECK-NEXT: call void @external(ptr nonnull [[ARRAY11]])
; CHECK-NEXT: call void @external(ptr nonnull [[ARRAY22]])
-; CHECK-NEXT: [[POINTER2:%.*]] = getelementptr i32, ptr [[ARRAY22]], i64 [[B:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[B:%.*]], 2
+; CHECK-NEXT: [[POINTER2:%.*]] = getelementptr i8, ptr [[ARRAY22]], i64 [[TMP1]]
; CHECK-NEXT: store i32 7, ptr [[POINTER2]], align 4
; CHECK-NEXT: ret i32 0
;
@@ -92,13 +93,15 @@ define i32 @gep_distance_test(ptr %A) {
; cannot alias, even if there is a variable offset between them...
define i32 @gep_distance_test2(ptr %A, i64 %distance) {
; NO_ASSUME-LABEL: @gep_distance_test2(
-; NO_ASSUME-NEXT: [[B_SPLIT:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 [[DISTANCE:%.*]]
+; NO_ASSUME-NEXT: [[TMP1:%.*]] = shl i64 [[DISTANCE:%.*]], 3
+; NO_ASSUME-NEXT: [[B_SPLIT:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; NO_ASSUME-NEXT: [[B:%.*]] = getelementptr i8, ptr [[B_SPLIT]], i64 4
; NO_ASSUME-NEXT: store i32 7, ptr [[B]], align 4
; NO_ASSUME-NEXT: ret i32 0
;
; USE_ASSUME-LABEL: @gep_distance_test2(
-; USE_ASSUME-NEXT: [[B_SPLIT:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 [[DISTANCE:%.*]]
+; USE_ASSUME-NEXT: [[TMP1:%.*]] = shl i64 [[DISTANCE:%.*]], 3
+; USE_ASSUME-NEXT: [[B_SPLIT:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; USE_ASSUME-NEXT: [[B:%.*]] = getelementptr i8, ptr [[B_SPLIT]], i64 4
; 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) ]
@@ -161,11 +164,13 @@ define i16 @zext_sext_confusion(ptr %row2col, i5 %j) nounwind{
; CHECK-LABEL: @zext_sext_confusion(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[SUM5_CAST:%.*]] = zext i5 [[J:%.*]] to i64
-; CHECK-NEXT: [[P1:%.*]] = getelementptr i16, ptr [[ROW2COL:%.*]], i64 [[SUM5_CAST]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw nsw i64 [[SUM5_CAST]], 1
+; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[ROW2COL:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[ROW2COL_LOAD_1_2:%.*]] = load i16, ptr [[P1]], align 1
; CHECK-NEXT: [[SUM13_CAST31:%.*]] = sext i5 [[J]] to i6
; CHECK-NEXT: [[SUM13_CAST:%.*]] = zext i6 [[SUM13_CAST31]] to i64
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i16, ptr [[ROW2COL]], i64 [[SUM13_CAST]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[SUM13_CAST]], 1
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[ROW2COL]], i64 [[TMP1]]
; CHECK-NEXT: [[ROW2COL_LOAD_1_6:%.*]] = load i16, ptr [[P2]], align 1
; CHECK-NEXT: [[DOTRET:%.*]] = sub i16 [[ROW2COL_LOAD_1_6]], [[ROW2COL_LOAD_1_2]]
; CHECK-NEXT: ret i16 [[DOTRET]]
diff --git a/llvm/test/CodeGen/AArch64/sme-aarch64-svcount-O3.ll b/llvm/test/CodeGen/AArch64/sme-aarch64-svcount-O3.ll
index 079f2e7eb481b..ae225afad469c 100644
--- a/llvm/test/CodeGen/AArch64/sme-aarch64-svcount-O3.ll
+++ b/llvm/test/CodeGen/AArch64/sme-aarch64-svcount-O3.ll
@@ -11,7 +11,8 @@ define target("aarch64.svcount") @test_alloca_store_reload(target("aarch64.svcou
; CHECK-NEXT: br i1 [[I1_PEEL]], label [[LOOP_EXIT:%.*]], label [[LOOP_BODY:%.*]]
; CHECK: loop.body:
; CHECK-NEXT: [[IND:%.*]] = phi i64 [ [[IND_NEXT:%.*]], [[LOOP_BODY]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[IPTR_GEP:%.*]] = getelementptr i64, ptr [[IPTR]], i64 [[IND]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[IND]], 3
+; CHECK-NEXT: [[IPTR_GEP:%.*]] = getelementptr i8, ptr [[IPTR]], i64 [[TMP0]]
; CHECK-NEXT: store i64 [[IND]], ptr [[IPTR_GEP]], align 8
; CHECK-NEXT: store target("aarch64.svcount") [[VAL1:%.*]], ptr [[PPTR]], align 2
; CHECK-NEXT: [[IND_NEXT]] = add i64 [[IND]], 1
diff --git a/llvm/test/CodeGen/AMDGPU/amdgpu-uniform-intrinsic-combine.ll b/llvm/test/CodeGen/AMDGPU/amdgpu-uniform-intrinsic-combine.ll
index 5391611742efd..263c5c93dcbee 100644
--- a/llvm/test/CodeGen/AMDGPU/amdgpu-uniform-intrinsic-combine.ll
+++ b/llvm/test/CodeGen/AMDGPU/amdgpu-uniform-intrinsic-combine.ll
@@ -51,8 +51,9 @@ define amdgpu_kernel void @permlane64_nonuniform(ptr addrspace(1) %out) {
; CURRENT-CHECK-SAME: ptr addrspace(1) writeonly captures(none) [[OUT:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] {
; CURRENT-CHECK-NEXT: [[TID:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; CURRENT-CHECK-NEXT: [[V:%.*]] = tail call i32 @llvm.amdgcn.permlane64.i32(i32 [[TID]])
-; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TID]] to i64
-; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i32, ptr addrspace(1) [[OUT]], i64 [[TMP1]]
+; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[TID]], 2
+; CURRENT-CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i8, ptr addrspace(1) [[OUT]], i64 [[TMP2]]
; CURRENT-CHECK-NEXT: store i32 [[V]], ptr addrspace(1) [[OUT_PTR]], align 4
; CURRENT-CHECK-NEXT: ret void
;
@@ -85,8 +86,9 @@ define amdgpu_kernel void @permlane64_nonuniform_expression(ptr addrspace(1) %ou
; CURRENT-CHECK-NEXT: [[TID:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; CURRENT-CHECK-NEXT: [[TID2:%.*]] = add nuw nsw i32 [[TID]], 1
; CURRENT-CHECK-NEXT: [[V:%.*]] = tail call i32 @llvm.amdgcn.permlane64.i32(i32 [[TID2]])
-; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TID]] to i64
-; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i32, ptr addrspace(1) [[OUT]], i64 [[TMP1]]
+; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[TID]], 2
+; CURRENT-CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i8, ptr addrspace(1) [[OUT]], i64 [[TMP2]]
; CURRENT-CHECK-NEXT: store i32 [[V]], ptr addrspace(1) [[OUT_PTR]], align 4
; CURRENT-CHECK-NEXT: ret void
;
@@ -164,8 +166,9 @@ define amdgpu_kernel void @readlane_nonuniform_workitem(ptr addrspace(1) %out) {
; CURRENT-CHECK-NEXT: [[TIDX:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; CURRENT-CHECK-NEXT: [[TIDY:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.y()
; CURRENT-CHECK-NEXT: [[V:%.*]] = tail call i32 @llvm.amdgcn.readlane.i32(i32 [[TIDX]], i32 [[TIDY]])
-; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TIDX]] to i64
-; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i32, ptr addrspace(1) [[OUT]], i64 [[TMP1]]
+; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[TIDX]], 2
+; CURRENT-CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i8, ptr addrspace(1) [[OUT]], i64 [[TMP2]]
; CURRENT-CHECK-NEXT: store i32 [[V]], ptr addrspace(1) [[OUT_PTR]], align 4
; CURRENT-CHECK-NEXT: ret void
;
@@ -203,8 +206,9 @@ define amdgpu_kernel void @readlane_nonuniform_expression(ptr addrspace(1) %out)
; CURRENT-CHECK-NEXT: [[TIDX2:%.*]] = add nuw nsw i32 [[TIDX]], 1
; CURRENT-CHECK-NEXT: [[TIDY2:%.*]] = add nuw nsw i32 [[TIDY]], 2
; CURRENT-CHECK-NEXT: [[V:%.*]] = tail call i32 @llvm.amdgcn.readlane.i32(i32 [[TIDX2]], i32 [[TIDY2]])
-; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TIDX]] to i64
-; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i32, ptr addrspace(1) [[OUT]], i64 [[TMP1]]
+; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[TIDX]], 2
+; CURRENT-CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i8, ptr addrspace(1) [[OUT]], i64 [[TMP2]]
; CURRENT-CHECK-NEXT: store i32 [[V]], ptr addrspace(1) [[OUT_PTR]], align 4
; CURRENT-CHECK-NEXT: ret void
;
@@ -291,8 +295,9 @@ define amdgpu_kernel void @readfirstlane_with_workitem_id(ptr addrspace(1) %out)
; CURRENT-CHECK-SAME: ptr addrspace(1) writeonly captures(none) [[OUT:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CURRENT-CHECK-NEXT: [[TID:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; CURRENT-CHECK-NEXT: [[V:%.*]] = tail call i32 @llvm.amdgcn.readfirstlane.i32(i32 [[TID]])
-; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TID]] to i64
-; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i32, ptr addrspace(1) [[OUT]], i64 [[TMP1]]
+; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[TID]], 2
+; CURRENT-CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i8, ptr addrspace(1) [[OUT]], i64 [[TMP2]]
; CURRENT-CHECK-NEXT: store i32 [[V]], ptr addrspace(1) [[OUT_PTR]], align 4
; CURRENT-CHECK-NEXT: ret void
;
@@ -325,8 +330,9 @@ define amdgpu_kernel void @readfirstlane_expression(ptr addrspace(1) %out) {
; CURRENT-CHECK-NEXT: [[TID:%.*]] = tail call i32 @llvm.amdgcn.workitem.id.x()
; CURRENT-CHECK-NEXT: [[TID2:%.*]] = add nuw nsw i32 [[TID]], 1
; CURRENT-CHECK-NEXT: [[V:%.*]] = tail call i32 @llvm.amdgcn.readfirstlane.i32(i32 [[TID2]])
-; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TID2]] to i64
-; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i32, ptr addrspace(1) [[OUT]], i64 [[TMP1]]
+; CURRENT-CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[TID2]], 2
+; CURRENT-CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CURRENT-CHECK-NEXT: [[OUT_PTR:%.*]] = getelementptr i8, ptr addrspace(1) [[OUT]], i64 [[TMP2]]
; CURRENT-CHECK-NEXT: store i32 [[V]], ptr addrspace(1) [[OUT_PTR]], align 4
; CURRENT-CHECK-NEXT: ret void
;
diff --git a/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll b/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll
index 15402c9992b18..c6fd87ce70f5d 100644
--- a/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll
+++ b/llvm/test/CodeGen/Hexagon/autohvx/vector-align-tbaa.ll
@@ -12,10 +12,12 @@ target triple = "hexagon"
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: [[V0:%.*]] = shl i32 [[A1:%.*]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A0:%.*]], i32 [[V0]]
; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[CST12:%.*]] = load <64 x i16>, ptr [[V1]], align 2, !tbaa [[TBAA0:![0-9]+]]
-; CHECK-NEXT: [[ITP:%.*]] = getelementptr i16, ptr [[A0]], i32 [[A1]]
+; CHECK-NEXT: [[V4:%.*]] = shl i32 [[A1]], 1
+; CHECK-NEXT: [[ITP:%.*]] = getelementptr i8, ptr [[A0]], i32 [[V4]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[ITP]], i32 256
; CHECK-NEXT: [[CST13:%.*]] = load <64 x i16>, ptr [[GEP3]], align 2, !tbaa [[TBAA0]]
; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST12]], [[CST13]]
@@ -37,10 +39,12 @@ b0:
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: [[V0:%.*]] = shl i32 [[A1:%.*]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A0:%.*]], i32 [[V0]]
; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[CST12:%.*]] = load <64 x i16>, ptr [[V1]], align 2, !tbaa [[TBAA0]]
-; CHECK-NEXT: [[ITP:%.*]] = getelementptr i16, ptr [[A0]], i32 [[A1]]
+; CHECK-NEXT: [[V4:%.*]] = shl i32 [[A1]], 1
+; CHECK-NEXT: [[ITP:%.*]] = getelementptr i8, ptr [[A0]], i32 [[V4]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[ITP]], i32 256
; CHECK-NEXT: [[CST13:%.*]] = load <64 x i16>, ptr [[GEP3]], align 2
; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST12]], [[CST13]]
@@ -62,10 +66,12 @@ b0:
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: [[V0:%.*]] = shl i32 [[A1:%.*]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A0:%.*]], i32 [[V0]]
; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[CST12:%.*]] = load <64 x i16>, ptr [[V1]], align 2, !tbaa [[TBAA0]]
-; CHECK-NEXT: [[ITP:%.*]] = getelementptr i16, ptr [[A0]], i32 [[A1]]
+; CHECK-NEXT: [[V4:%.*]] = shl i32 [[A1]], 1
+; CHECK-NEXT: [[ITP:%.*]] = getelementptr i8, ptr [[A0]], i32 [[V4]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[ITP]], i32 256
; CHECK-NEXT: [[CST13:%.*]] = load <64 x i16>, ptr [[GEP3]], align 2, !tbaa [[TBAA3:![0-9]+]]
; CHECK-NEXT: [[V8:%.*]] = add <64 x i16> [[CST12]], [[CST13]]
@@ -87,7 +93,8 @@ b0:
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: [[V0:%.*]] = shl i32 [[A1:%.*]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A0:%.*]], i32 [[V0]]
; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
@@ -140,7 +147,8 @@ b0:
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: [[V0:%.*]] = shl i32 [[A1:%.*]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A0:%.*]], i32 [[V0]]
; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
@@ -193,7 +201,8 @@ b0:
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: [[V0:%.*]] = shl i32 [[A1:%.*]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A0:%.*]], i32 [[V0]]
; CHECK-NEXT: [[V1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 128
; CHECK-NEXT: [[PTI:%.*]] = ptrtoint ptr [[V1]] to i32
; CHECK-NEXT: [[AND:%.*]] = and i32 [[PTI]], -128
diff --git a/llvm/test/Transforms/InstCombine/2006-12-15-Range-Test.ll b/llvm/test/Transforms/InstCombine/2006-12-15-Range-Test.ll
index 42b30eba1740f..694311ff8c132 100644
--- a/llvm/test/Transforms/InstCombine/2006-12-15-Range-Test.ll
+++ b/llvm/test/Transforms/InstCombine/2006-12-15-Range-Test.ll
@@ -16,7 +16,8 @@ define i1 @print_pgm_cond_true(i32 %tmp12.reload, ptr %tmp16.out) {
; CHECK-NEXT: store i32 [[TMP16]], ptr [[TMP16_OUT]], align 4
; CHECK-NEXT: ret i1 false
; CHECK: cond_true:
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i32, ptr @r, i32 [[TMP12_RELOAD:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[TMP12_RELOAD:%.*]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr @r, i32 [[TMP1]]
; CHECK-NEXT: [[TMP16]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[TMP16]], -32
; CHECK-NEXT: [[BOTHCOND:%.*]] = icmp ult i32 [[TMP0]], -63
@@ -53,7 +54,8 @@ define i1 @print_pgm_cond_true_logical(i32 %tmp12.reload, ptr %tmp16.out) {
; CHECK-NEXT: store i32 [[TMP16]], ptr [[TMP16_OUT]], align 4
; CHECK-NEXT: ret i1 false
; CHECK: cond_true:
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i32, ptr @r, i32 [[TMP12_RELOAD:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[TMP12_RELOAD:%.*]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr @r, i32 [[TMP1]]
; CHECK-NEXT: [[TMP16]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[TMP16]], -32
; CHECK-NEXT: [[BOTHCOND:%.*]] = icmp ult i32 [[TMP0]], -63
diff --git a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-gatherscatter.ll b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-gatherscatter.ll
index 9f3fdcd8a767c..60f150050adfb 100644
--- a/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-gatherscatter.ll
+++ b/llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-gatherscatter.ll
@@ -9,7 +9,8 @@ target triple = "aarch64-unknown-linux-gnu"
define <vscale x 2 x double> @test_ld1_gather_index_nxv2f64_stride1(<vscale x 2 x i1> %pred, ptr %x, i64 %base) #0 {
; CHECK-LABEL: @test_ld1_gather_index_nxv2f64_stride1(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[X:%.*]], i64 [[BASE:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[BASE:%.*]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[LD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 1 [[TMP1]], <vscale x 2 x i1> [[PRED:%.*]], <vscale x 2 x double> zeroinitializer)
; CHECK-NEXT: ret <vscale x 2 x double> [[LD]]
;
@@ -31,7 +32,8 @@ define <vscale x 2 x double> @test_ld1_gather_index_nxv2f64_stride2_negtest(<vsc
define <vscale x 2 x double> @test_ld1_gather_index_nxv2f64_stride1_align8(<vscale x 2 x i1> %pred, ptr align 8 %x, i64 %base) #0 {
; CHECK-LABEL: @test_ld1_gather_index_nxv2f64_stride1_align8(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[X:%.*]], i64 [[BASE:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[BASE:%.*]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[LD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP1]], <vscale x 2 x i1> [[PRED:%.*]], <vscale x 2 x double> zeroinitializer)
; CHECK-NEXT: ret <vscale x 2 x double> [[LD]]
;
@@ -46,7 +48,8 @@ define <vscale x 2 x double> @test_ld1_gather_index_nxv2f64_stride1_align8(<vsca
define void @test_st1_scatter_index_nxv2f64_stride1(<vscale x 2 x i1> %pred, ptr %x, i64 %base, <vscale x 2 x double> %val) #0 {
; CHECK-LABEL: @test_st1_scatter_index_nxv2f64_stride1(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[X:%.*]], i64 [[BASE:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[BASE:%.*]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP2]]
; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[VAL:%.*]], ptr align 1 [[TMP1]], <vscale x 2 x i1> [[PRED:%.*]])
; CHECK-NEXT: ret void
;
@@ -68,7 +71,8 @@ define void @test_st1_scatter_index_nxv2f64_stride2_negtest(<vscale x 2 x i1> %p
define void @test_st1_scatter_index_nxv2f64_stride1_align8(<vscale x 2 x i1> %pred, ptr align 8 %x, i64 %base, <vscale x 2 x double> %val) #0 {
; CHECK-LABEL: @test_st1_scatter_index_nxv2f64_stride1_align8(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[X:%.*]], i64 [[BASE:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[BASE:%.*]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP2]]
; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[VAL:%.*]], ptr align 8 [[TMP1]], <vscale x 2 x i1> [[PRED:%.*]])
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/AMDGPU/canonicalize-add-to-gep.ll b/llvm/test/Transforms/InstCombine/AMDGPU/canonicalize-add-to-gep.ll
index b1016f3a741fc..74ce1d61555e2 100644
--- a/llvm/test/Transforms/InstCombine/AMDGPU/canonicalize-add-to-gep.ll
+++ b/llvm/test/Transforms/InstCombine/AMDGPU/canonicalize-add-to-gep.ll
@@ -7,9 +7,9 @@ define amdgpu_ps <2 x float> @turn_add_into_gep(ptr addrspace(1) inreg %sbase) {
; CHECK-LABEL: define amdgpu_ps <2 x float> @turn_add_into_gep(
; CHECK-SAME: ptr addrspace(1) inreg [[SBASE:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[V:%.*]] = tail call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
-; CHECK-NEXT: [[MUL:%.*]] = shl nuw nsw i32 [[V]], 1
+; CHECK-NEXT: [[MUL:%.*]] = shl nuw nsw i32 [[V]], 3
; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[MUL]] to i64
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw float, ptr addrspace(1) [[SBASE]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(1) [[SBASE]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(1) [[TMP2]], i64 128
; CHECK-NEXT: [[LOAD:%.*]] = load <2 x float>, ptr addrspace(1) [[GEP]], align 8
; CHECK-NEXT: ret <2 x float> [[LOAD]]
diff --git a/llvm/test/Transforms/InstCombine/AMDGPU/memcpy-from-constant.ll b/llvm/test/Transforms/InstCombine/AMDGPU/memcpy-from-constant.ll
index 968138e45b33a..064e133a62128 100644
--- a/llvm/test/Transforms/InstCombine/AMDGPU/memcpy-from-constant.ll
+++ b/llvm/test/Transforms/InstCombine/AMDGPU/memcpy-from-constant.ll
@@ -36,7 +36,8 @@ define i8 @memcpy_constant_arg_ptr_to_alloca_load_metadata(ptr addrspace(4) noal
define i64 @memcpy_constant_arg_ptr_to_alloca_load_alignment(ptr addrspace(4) noalias readonly align 4 dereferenceable(256) %arg, i32 %idx) {
; CHECK-LABEL: @memcpy_constant_arg_ptr_to_alloca_load_alignment(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[IDX:%.*]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i64, ptr addrspace(4) [[ARG:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 3
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr addrspace(4) [[ARG:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[LOAD:%.*]] = load i64, ptr addrspace(4) [[GEP]], align 16
; CHECK-NEXT: ret i64 [[LOAD]]
;
@@ -51,7 +52,8 @@ define i64 @memcpy_constant_arg_ptr_to_alloca_load_atomic(ptr addrspace(4) noali
; CHECK-LABEL: @memcpy_constant_arg_ptr_to_alloca_load_atomic(
; CHECK-NEXT: [[ALLOCA:%.*]] = alloca [32 x i64], align 8, addrspace(5)
; CHECK-NEXT: call void @llvm.memcpy.p5.p4.i64(ptr addrspace(5) noundef align 8 dereferenceable(256) [[ALLOCA]], ptr addrspace(4) noundef align 8 dereferenceable(256) [[ARG:%.*]], i64 256, i1 false)
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i64, ptr addrspace(5) [[ALLOCA]], i32 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr addrspace(5) [[ALLOCA]], i32 [[TMP1]]
; CHECK-NEXT: [[LOAD:%.*]] = load atomic i64, ptr addrspace(5) [[GEP]] syncscope("somescope") acquire, align 8
; CHECK-NEXT: ret i64 [[LOAD]]
;
diff --git a/llvm/test/Transforms/InstCombine/align-addr.ll b/llvm/test/Transforms/InstCombine/align-addr.ll
index b77037e592b54..a6fb1823cf1d3 100644
--- a/llvm/test/Transforms/InstCombine/align-addr.ll
+++ b/llvm/test/Transforms/InstCombine/align-addr.ll
@@ -12,14 +12,15 @@ define void @test0(ptr %b, i64 %n, i64 %u, i64 %y) nounwind {
; CHECK-NEXT: [[D:%.*]] = and i64 [[C]], -16
; CHECK-NEXT: [[E:%.*]] = inttoptr i64 [[D]] to ptr
; CHECK-NEXT: [[V:%.*]] = shl i64 [[U:%.*]], 1
-; CHECK-NEXT: [[Z:%.*]] = and i64 [[Y:%.*]], -2
+; CHECK-NEXT: [[Z:%.*]] = and i64 [[Y:%.*]], 2305843009213693950
; CHECK-NEXT: [[T1421:%.*]] = icmp eq i64 [[N:%.*]], 0
; CHECK-NEXT: br i1 [[T1421]], label [[RETURN:%.*]], label [[BB:%.*]]
; CHECK: bb:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[INDVAR_NEXT:%.*]], [[BB]] ], [ 20, [[ENTRY:%.*]] ]
; CHECK-NEXT: [[J:%.*]] = mul i64 [[I]], [[V]]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr double, ptr [[E]], i64 [[J]]
-; CHECK-NEXT: [[T8:%.*]] = getelementptr double, ptr [[TMP0]], i64 [[Z]]
+; CHECK-NEXT: [[H:%.*]] = add i64 [[J]], [[Z]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[H]], 3
+; CHECK-NEXT: [[T8:%.*]] = getelementptr i8, ptr [[E]], i64 [[TMP0]]
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr [[T8]], align 8
; CHECK-NEXT: [[INDVAR_NEXT]] = add i64 [[I]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVAR_NEXT]], [[N]]
diff --git a/llvm/test/Transforms/InstCombine/and-or-icmps.ll b/llvm/test/Transforms/InstCombine/and-or-icmps.ll
index 4cd089ca524c2..c6f935bf4382c 100644
--- a/llvm/test/Transforms/InstCombine/and-or-icmps.ll
+++ b/llvm/test/Transforms/InstCombine/and-or-icmps.ll
@@ -407,8 +407,8 @@ define void @simplify_before_foldAndOfICmps2(ptr %p, ptr %A8) "instcombine-no-ve
; CHECK-NEXT: [[L7:%.*]] = load i16, ptr [[A8:%.*]], align 2
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i16 [[L7]], -1
; CHECK-NEXT: [[B11:%.*]] = zext i1 [[TMP1]] to i16
-; CHECK-NEXT: [[TMP2:%.*]] = zext i1 [[TMP1]] to i64
-; CHECK-NEXT: [[G4:%.*]] = getelementptr i16, ptr [[A8]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[TMP1]], i64 2, i64 0
+; CHECK-NEXT: [[G4:%.*]] = getelementptr i8, ptr [[A8]], i64 [[TMP2]]
; CHECK-NEXT: [[L2:%.*]] = load i16, ptr [[G4]], align 2
; CHECK-NEXT: [[L4:%.*]] = load i16, ptr [[A8]], align 2
; CHECK-NEXT: [[B21:%.*]] = sdiv i16 [[L7]], [[L4]]
@@ -429,7 +429,7 @@ define void @simplify_before_foldAndOfICmps2(ptr %p, ptr %A8) "instcombine-no-ve
; CHECK-NEXT: [[TMP4:%.*]] = xor i1 [[C11]], true
; CHECK-NEXT: [[C18:%.*]] = or i1 [[C10]], [[TMP4]]
; CHECK-NEXT: [[TMP3:%.*]] = sext i1 [[C4]] to i64
-; CHECK-NEXT: [[G26:%.*]] = getelementptr i1, ptr null, i64 [[TMP3]]
+; CHECK-NEXT: [[G26:%.*]] = getelementptr i8, ptr null, i64 [[TMP3]]
; CHECK-NEXT: store i16 [[B33]], ptr [[P:%.*]], align 2
; CHECK-NEXT: store i1 [[C18]], ptr [[P]], align 1
; CHECK-NEXT: store ptr [[G26]], ptr [[P]], align 8
diff --git a/llvm/test/Transforms/InstCombine/array.ll b/llvm/test/Transforms/InstCombine/array.ll
index ec1a39dff47bf..e0f0163b65d4e 100644
--- a/llvm/test/Transforms/InstCombine/array.ll
+++ b/llvm/test/Transforms/InstCombine/array.ll
@@ -5,9 +5,10 @@ define void @test(ptr %ptr, i32 %a, i32 %b) {
; CHECK-LABEL: define void @test(
; CHECK-SAME: 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 i8, ptr [[TMP1]], i64 40
+; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[A]], 10
+; CHECK-NEXT: [[IDX:%.*]] = sext i32 [[ADD]] to i64
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP0]]
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret void
;
@@ -25,7 +26,8 @@ define i32 @test_add_res_moreoneuse(ptr %ptr, i32 %a, i32 %b) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[A]], 5
; CHECK-NEXT: [[IDX:%.*]] = sext i32 [[ADD]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP0]]
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret i32 [[ADD]]
;
@@ -43,7 +45,8 @@ define void @test_addop_nonsw_flag(ptr %ptr, i32 %a, i32 %b) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[ADD:%.*]] = add i32 [[A]], 10
; CHECK-NEXT: [[IDX:%.*]] = sext i32 [[ADD]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP0]]
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret void
;
@@ -61,7 +64,8 @@ define void @test_add_op2_not_constant(ptr %ptr, i32 %a, i32 %b) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[ADD:%.*]] = add i32 [[A]], [[B]]
; CHECK-NEXT: [[IDX:%.*]] = sext i32 [[ADD]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP0]]
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret void
;
@@ -77,9 +81,10 @@ define void @test_zext_nneg(ptr %ptr, i32 %a, i32 %b) {
; CHECK-LABEL: define void @test_zext_nneg(
; CHECK-SAME: 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 i8, ptr [[TMP1]], i64 40
+; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[A]], 10
+; CHECK-NEXT: [[IDX:%.*]] = zext nneg i32 [[ADD]] to i64
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[TMP0]]
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret void
;
@@ -97,7 +102,8 @@ define void @test_zext_missing_nneg(ptr %ptr, i32 %a, i32 %b) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[A]], 10
; CHECK-NEXT: [[IDX:%.*]] = zext i32 [[ADD]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[PTR]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[TMP0]]
; CHECK-NEXT: store i32 [[B]], ptr [[GEP]], align 4
; CHECK-NEXT: ret void
;
@@ -112,8 +118,9 @@ entry:
define ptr @gep_inbounds_nuwaddlike(ptr %ptr, i64 %a, i64 %b) {
; CHECK-LABEL: define ptr @gep_inbounds_nuwaddlike(
; CHECK-SAME: ptr [[PTR:%.*]], i64 [[A:%.*]], i64 [[B:%.*]]) {
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = or disjoint i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%add = or disjoint i64 %a, %b
@@ -124,8 +131,9 @@ define ptr @gep_inbounds_nuwaddlike(ptr %ptr, i64 %a, i64 %b) {
define ptr @gep_inbounds_add_nuw(ptr %ptr, i64 %a, i64 %b) {
; CHECK-LABEL: define ptr @gep_inbounds_add_nuw(
; CHECK-SAME: ptr [[PTR:%.*]], i64 [[A:%.*]], i64 [[B:%.*]]) {
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nuw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%add = add nuw i64 %a, %b
@@ -136,8 +144,9 @@ define ptr @gep_inbounds_add_nuw(ptr %ptr, i64 %a, i64 %b) {
define ptr @gep_inbounds_add_nusw_nuw(ptr %ptr, i64 %a, i64 %b) {
; CHECK-LABEL: define ptr @gep_inbounds_add_nusw_nuw(
; CHECK-SAME: ptr [[PTR:%.*]], i64 [[A:%.*]], i64 [[B:%.*]]) {
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr nusw nuw i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr nusw nuw i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nuw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr nusw nuw i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%add = add nuw i64 %a, %b
@@ -148,8 +157,9 @@ define ptr @gep_inbounds_add_nusw_nuw(ptr %ptr, i64 %a, i64 %b) {
define ptr @gep_add_nuw(ptr %ptr, i64 %a, i64 %b) {
; CHECK-LABEL: define ptr @gep_add_nuw(
; CHECK-SAME: ptr [[PTR:%.*]], i64 [[A:%.*]], i64 [[B:%.*]]) {
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr nuw i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr nuw i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nuw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr nuw i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%add = add nuw i64 %a, %b
@@ -164,8 +174,9 @@ define ptr @gep_inbounds_add_nsw_nonneg(ptr %ptr, i64 %a, i64 %b) {
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
; CHECK-NEXT: [[B_NNEG:%.*]] = icmp sgt i64 [[B]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[B_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nuw nsw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%a.nneg = icmp sgt i64 %a, -1
@@ -182,8 +193,9 @@ define ptr @gep_inbounds_add_nsw_not_nonneg1(ptr %ptr, i64 %a, i64 %b) {
; CHECK-SAME: ptr [[PTR:%.*]], i64 [[A:%.*]], i64 [[B:%.*]]) {
; CHECK-NEXT: [[A_NNEG:%.*]] = icmp sgt i64 [[A]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%a.nneg = icmp sgt i64 %a, -1
@@ -198,8 +210,9 @@ define ptr @gep_inbounds_add_nsw_not_nonneg2(ptr %ptr, i64 %a, i64 %b) {
; CHECK-SAME: ptr [[PTR:%.*]], i64 [[A:%.*]], i64 [[B:%.*]]) {
; CHECK-NEXT: [[B_NNEG:%.*]] = icmp sgt i64 [[B]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[B_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%b.nneg = icmp sgt i64 %b, -1
@@ -216,8 +229,9 @@ define ptr @gep_not_inbounds_add_nsw_nonneg(ptr %ptr, i64 %a, i64 %b) {
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
; CHECK-NEXT: [[B_NNEG:%.*]] = icmp sgt i64 [[B]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[B_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nuw nsw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%a.nneg = icmp sgt i64 %a, -1
@@ -236,8 +250,9 @@ define ptr @gep_inbounds_add_not_nsw_nonneg(ptr %ptr, i64 %a, i64 %b) {
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
; CHECK-NEXT: [[B_NNEG:%.*]] = icmp sgt i64 [[B]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[B_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[A]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[B]]
+; CHECK-NEXT: [[ADD:%.*]] = add nuw i64 [[A]], [[B]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[ADD]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%a.nneg = icmp sgt i64 %a, -1
@@ -254,9 +269,10 @@ define ptr @gep_inbounds_sext_add_nonneg(ptr %ptr, i32 %a) {
; CHECK-SAME: ptr [[PTR:%.*]], i32 [[A:%.*]]) {
; CHECK-NEXT: [[A_NNEG:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[A]] to i64
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[PTR]], i64 [[TMP1]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 40
+; CHECK-NEXT: [[ADD:%.*]] = add nuw nsw i32 [[A]], 10
+; CHECK-NEXT: [[IDX:%.*]] = zext nneg i32 [[ADD]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%a.nneg = icmp sgt i32 %a, -1
@@ -272,9 +288,10 @@ define ptr @gep_inbounds_sext_addlike_nonneg(ptr %ptr, i32 %a) {
; CHECK-SAME: ptr [[PTR:%.*]], i32 [[A:%.*]]) {
; CHECK-NEXT: [[A_NNEG:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
-; CHECK-NEXT: [[IDX:%.*]] = zext nneg i32 [[A]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[PTR]], i64 [[IDX]]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP]], i64 40
+; CHECK-NEXT: [[ADD:%.*]] = or disjoint i32 [[A]], 10
+; CHECK-NEXT: [[IDX:%.*]] = zext nneg i32 [[ADD]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds nuw i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP1]]
;
%a.nneg = icmp sgt i32 %a, -1
@@ -290,9 +307,10 @@ define ptr @gep_inbounds_sext_add_not_nonneg_1(ptr %ptr, i32 %a) {
; CHECK-SAME: ptr [[PTR:%.*]], i32 [[A:%.*]]) {
; CHECK-NEXT: [[A_NNEG:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[A]] to i64
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[TMP1]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 -40
+; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[A]], -10
+; CHECK-NEXT: [[IDX:%.*]] = sext i32 [[ADD]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%a.nneg = icmp sgt i32 %a, -1
@@ -306,9 +324,10 @@ define ptr @gep_inbounds_sext_add_not_nonneg_1(ptr %ptr, i32 %a) {
define ptr @gep_inbounds_sext_add_not_nonneg_2(ptr %ptr, i32 %a) {
; CHECK-LABEL: define ptr @gep_inbounds_sext_add_not_nonneg_2(
; CHECK-SAME: ptr [[PTR:%.*]], i32 [[A:%.*]]) {
-; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[A]] to i64
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[TMP1]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 40
+; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[A]], 10
+; CHECK-NEXT: [[IDX:%.*]] = sext i32 [[ADD]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%add = add nsw i32 %a, 10
@@ -322,9 +341,10 @@ define ptr @gep_not_inbounds_sext_add_nonneg(ptr %ptr, i32 %a) {
; CHECK-SAME: ptr [[PTR:%.*]], i32 [[A:%.*]]) {
; CHECK-NEXT: [[A_NNEG:%.*]] = icmp sgt i32 [[A]], -1
; CHECK-NEXT: call void @llvm.assume(i1 [[A_NNEG]])
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[A]] to i64
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[PTR]], i64 [[TMP1]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[TMP2]], i64 40
+; CHECK-NEXT: [[ADD:%.*]] = add nuw nsw i32 [[A]], 10
+; CHECK-NEXT: [[IDX:%.*]] = zext nneg i32 [[ADD]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[PTR]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%a.nneg = icmp sgt i32 %a, -1
diff --git a/llvm/test/Transforms/InstCombine/assume-loop-align.ll b/llvm/test/Transforms/InstCombine/assume-loop-align.ll
index f6805ba4894f0..196ee91713d82 100644
--- a/llvm/test/Transforms/InstCombine/assume-loop-align.ll
+++ b/llvm/test/Transforms/InstCombine/assume-loop-align.ll
@@ -15,10 +15,12 @@ define void @foo(ptr %a, ptr %b) #0 {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 16
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-gep-constglob.ll b/llvm/test/Transforms/InstCombine/canonicalize-gep-constglob.ll
index 129da3f9110ad..d5a5dfc0df6de 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-gep-constglob.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-gep-constglob.ll
@@ -35,9 +35,12 @@ define ptr @xzy(i64 %x, i64 %y, i64 %z) {
; CHECK-LABEL: define ptr @xzy(
; CHECK-SAME: i64 [[X:%.*]], i64 [[Y:%.*]], i64 [[Z:%.*]]) {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr getelementptr inbounds nuw (i8, ptr @glob, i64 40), i64 [[X]]
-; CHECK-NEXT: [[GEP_SPLIT1:%.*]] = getelementptr inbounds [10 x i32], ptr [[GEP_SPLIT]], i64 [[Z]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[GEP_SPLIT1]], i64 [[Y]]
+; CHECK-NEXT: [[TMP0:%.*]] = mul nsw i64 [[X]], 400
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @glob, i64 40), i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul nsw i64 [[Z]], 40
+; CHECK-NEXT: [[GEP_SPLIT1:%.*]] = getelementptr inbounds i8, ptr [[GEP_SPLIT]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Y]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[GEP_SPLIT1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP]]
;
entry:
@@ -49,7 +52,8 @@ define ptr @zerox(i64 %x) {
; CHECK-LABEL: define ptr @zerox(
; CHECK-SAME: i64 [[X:%.*]]) {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @glob, i64 32), i64 [[X]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @glob, i64 32), i64 [[TMP0]]
; CHECK-NEXT: ret ptr [[GEP]]
;
entry:
diff --git a/llvm/test/Transforms/InstCombine/canonicalize-gep-mul.ll b/llvm/test/Transforms/InstCombine/canonicalize-gep-mul.ll
index 1b33e0c2087f1..7fcc02a99c136 100644
--- a/llvm/test/Transforms/InstCombine/canonicalize-gep-mul.ll
+++ b/llvm/test/Transforms/InstCombine/canonicalize-gep-mul.ll
@@ -62,7 +62,8 @@ define ptr @usemul(ptr %p, i64 %x) {
; CHECK-SAME: ptr [[P:%.*]], i64 [[X:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[MUL:%.*]] = mul i64 [[X]], 5
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[P]], i64 [[MUL]]
+; CHECK-NEXT: [[TMP0:%.*]] = mul i64 [[X]], 20
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP0]]
; CHECK-NEXT: call void @use(i64 [[MUL]])
; CHECK-NEXT: ret ptr [[GEP]]
;
diff --git a/llvm/test/Transforms/InstCombine/cast.ll b/llvm/test/Transforms/InstCombine/cast.ll
index 46deb294b9d45..fb5a050c182a8 100644
--- a/llvm/test/Transforms/InstCombine/cast.ll
+++ b/llvm/test/Transforms/InstCombine/cast.ll
@@ -1090,7 +1090,8 @@ define double @test73(ptr %p, i128 %i) {
define double @test74(ptr %p, i64 %i) {
; ALL-LABEL: @test74(
-; ALL-NEXT: [[PP:%.*]] = getelementptr inbounds i64, ptr [[P:%.*]], i64 [[I:%.*]]
+; ALL-NEXT: [[TMP1:%.*]] = shl nsw i64 [[I:%.*]], 3
+; ALL-NEXT: [[PP:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; ALL-NEXT: [[L:%.*]] = load double, ptr [[PP]], align 8
; ALL-NEXT: ret double [[L]]
;
diff --git a/llvm/test/Transforms/InstCombine/cast_phi.ll b/llvm/test/Transforms/InstCombine/cast_phi.ll
index a3f818e013208..47a610648ff1c 100644
--- a/llvm/test/Transforms/InstCombine/cast_phi.ll
+++ b/llvm/test/Transforms/InstCombine/cast_phi.ll
@@ -9,9 +9,11 @@ define void @MainKernel(i32 %iNumSteps, i32 %tid, i32 %base) {
; CHECK-NEXT: [[CALLB:%.*]] = alloca [258 x float], align 4
; CHECK-NEXT: [[CONV_I:%.*]] = uitofp i32 [[INUMSTEPS:%.*]] to float
; CHECK-NEXT: [[CONV_I12:%.*]] = zext i32 [[TID:%.*]] to i64
-; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw float, ptr [[CALLA]], i64 [[CONV_I12]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw i64 [[CONV_I12]], 2
+; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw i8, ptr [[CALLA]], i64 [[TMP12]]
; CHECK-NEXT: store float [[CONV_I]], ptr [[ARRAYIDX3]], align 4
-; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[CALLB]], i64 [[CONV_I12]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nuw nsw i64 [[CONV_I12]], 2
+; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i8, ptr [[CALLB]], i64 [[TMP13]]
; CHECK-NEXT: store float [[CONV_I]], ptr [[ARRAYIDX6]], align 4
; CHECK-NEXT: [[CMP7:%.*]] = icmp eq i32 [[TID]], 0
; CHECK-NEXT: br i1 [[CMP7]], label [[DOTBB1:%.*]], label [[DOTBB2:%.*]]
@@ -31,8 +33,10 @@ define void @MainKernel(i32 %iNumSteps, i32 %tid, i32 %base) {
; CHECK-NEXT: [[TMP3:%.*]] = icmp ugt i32 [[I12_06]], [[BASE:%.*]]
; CHECK-NEXT: [[ADD:%.*]] = add nuw i32 [[I12_06]], 1
; CHECK-NEXT: [[CONV_I9:%.*]] = sext i32 [[ADD]] to i64
-; CHECK-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds float, ptr [[CALLA]], i64 [[CONV_I9]]
-; CHECK-NEXT: [[ARRAYIDX24:%.*]] = getelementptr inbounds float, ptr [[CALLB]], i64 [[CONV_I9]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[CONV_I9]], 2
+; CHECK-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds i8, ptr [[CALLA]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[CONV_I9]], 2
+; CHECK-NEXT: [[ARRAYIDX24:%.*]] = getelementptr inbounds i8, ptr [[CALLB]], i64 [[TMP15]]
; CHECK-NEXT: [[CMP40:%.*]] = icmp ult i32 [[I12_06]], [[BASE]]
; CHECK-NEXT: br i1 [[TMP3]], label [[DOTBB4:%.*]], label [[DOTBB5:%.*]]
; CHECK: .bb4:
diff --git a/llvm/test/Transforms/InstCombine/cast_ptr.ll b/llvm/test/Transforms/InstCombine/cast_ptr.ll
index 1cf3e32556b38..a145afc724b6a 100644
--- a/llvm/test/Transforms/InstCombine/cast_ptr.ll
+++ b/llvm/test/Transforms/InstCombine/cast_ptr.ll
@@ -435,9 +435,9 @@ define i32 @ptr_add_in_int_extra_use2(i32 %x) {
define i32 @ptrtoint_of_inttoptr_multiple_gep(i32 %x, i32 %y, i32 %z) {
; CHECK-LABEL: @ptrtoint_of_inttoptr_multiple_gep(
-; CHECK-NEXT: [[PTR2_IDX:%.*]] = shl nuw i32 [[Y:%.*]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = add nuw i32 [[X:%.*]], [[PTR2_IDX]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw i32 [[Y:%.*]], 1
; CHECK-NEXT: [[PTR3_IDX:%.*]] = shl i32 [[Z:%.*]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = add nuw i32 [[X:%.*]], [[TMP2]]
; CHECK-NEXT: [[R:%.*]] = add i32 [[TMP1]], [[PTR3_IDX]]
; CHECK-NEXT: ret i32 [[R]]
;
@@ -451,9 +451,11 @@ define i32 @ptrtoint_of_inttoptr_multiple_gep(i32 %x, i32 %y, i32 %z) {
define i32 @ptrtoint_of_inttoptr_multiple_gep_extra_use(i32 %x, i32 %y, i32 %z) {
; CHECK-LABEL: @ptrtoint_of_inttoptr_multiple_gep_extra_use(
; CHECK-NEXT: [[PTR:%.*]] = inttoptr i32 [[X:%.*]] to ptr
-; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i16, ptr [[PTR]], i32 [[Y:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[Y:%.*]], 1
+; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i8, ptr [[PTR]], i32 [[TMP1]]
; CHECK-NEXT: call void @use_ptr(ptr [[PTR2]])
-; CHECK-NEXT: [[PTR3:%.*]] = getelementptr i32, ptr [[PTR2]], i32 [[Z:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i32 [[Z:%.*]], 2
+; CHECK-NEXT: [[PTR3:%.*]] = getelementptr i8, ptr [[PTR2]], i32 [[TMP2]]
; CHECK-NEXT: [[R:%.*]] = ptrtoint ptr [[PTR3]] to i32
; CHECK-NEXT: ret i32 [[R]]
;
@@ -468,7 +470,8 @@ define i32 @ptrtoint_of_inttoptr_multiple_gep_extra_use(i32 %x, i32 %y, i32 %z)
define i32 @ptrtoint_of_inttoptr_index_type(i32 %x, i16 %y) {
; CHECK-LABEL: @ptrtoint_of_inttoptr_index_type(
; CHECK-NEXT: [[PTR:%.*]] = inttoptr i32 [[X:%.*]] to ptr addrspace(3)
-; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i16, ptr addrspace(3) [[PTR]], i16 [[Y:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i16 [[Y:%.*]], 1
+; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i8, ptr addrspace(3) [[PTR]], i16 [[TMP1]]
; CHECK-NEXT: [[R:%.*]] = ptrtoint ptr addrspace(3) [[PTR2]] to i32
; CHECK-NEXT: ret i32 [[R]]
;
@@ -481,9 +484,9 @@ define i32 @ptrtoint_of_inttoptr_index_type(i32 %x, i16 %y) {
define i32 @ptrtoint_of_null_multiple_gep(i32 %x, i32 %y, i32 %z) {
; CHECK-LABEL: @ptrtoint_of_null_multiple_gep(
; CHECK-NEXT: [[PTR2_IDX:%.*]] = shl i32 [[X:%.*]], 1
-; CHECK-NEXT: [[PTR3_IDX:%.*]] = shl nuw i32 [[Y:%.*]], 2
-; CHECK-NEXT: [[TMP1:%.*]] = add nuw i32 [[PTR2_IDX]], [[PTR3_IDX]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw i32 [[Y:%.*]], 2
; CHECK-NEXT: [[PTR4_IDX:%.*]] = shl i32 [[Z:%.*]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = add nuw i32 [[PTR2_IDX]], [[TMP2]]
; CHECK-NEXT: [[R:%.*]] = add i32 [[TMP1]], [[PTR4_IDX]]
; CHECK-NEXT: ret i32 [[R]]
;
@@ -496,10 +499,13 @@ define i32 @ptrtoint_of_null_multiple_gep(i32 %x, i32 %y, i32 %z) {
define i32 @ptrtoint_of_null_multiple_gep_extra_use(i32 %x, i32 %y, i32 %z) {
; CHECK-LABEL: @ptrtoint_of_null_multiple_gep_extra_use(
-; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i16, ptr null, i32 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[X:%.*]], 1
+; CHECK-NEXT: [[PTR2:%.*]] = getelementptr i8, ptr null, i32 [[TMP1]]
; CHECK-NEXT: call void @use_ptr(ptr [[PTR2]])
-; CHECK-NEXT: [[PTR3:%.*]] = getelementptr nuw i32, ptr [[PTR2]], i32 [[Y:%.*]]
-; CHECK-NEXT: [[PTR4:%.*]] = getelementptr i64, ptr [[PTR3]], i32 [[Z:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw i32 [[Y:%.*]], 2
+; CHECK-NEXT: [[PTR3:%.*]] = getelementptr nuw i8, ptr [[PTR2]], i32 [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl i32 [[Z:%.*]], 3
+; CHECK-NEXT: [[PTR4:%.*]] = getelementptr i8, ptr [[PTR3]], i32 [[TMP3]]
; CHECK-NEXT: [[R:%.*]] = ptrtoint ptr [[PTR4]] to i32
; CHECK-NEXT: ret i32 [[R]]
;
@@ -524,7 +530,8 @@ define i32 @ptrtoint_of_null_index_type(i16 %x) {
define <2 x i32> @ptrtoint_of_null_splat(<2 x i16> %x) {
; CHECK-LABEL: @ptrtoint_of_null_splat(
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr i16, ptr addrspace(3) null, <2 x i16> [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <2 x i16> [[X:%.*]], splat (i16 1)
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr addrspace(3) null, <2 x i16> [[TMP1]]
; CHECK-NEXT: [[R:%.*]] = ptrtoint <2 x ptr addrspace(3)> [[PTR]] to <2 x i32>
; CHECK-NEXT: ret <2 x i32> [[R]]
;
diff --git a/llvm/test/Transforms/InstCombine/constant-fold-address-space-pointer.ll b/llvm/test/Transforms/InstCombine/constant-fold-address-space-pointer.ll
index 81052194b6b95..82078a4e203fe 100644
--- a/llvm/test/Transforms/InstCombine/constant-fold-address-space-pointer.ll
+++ b/llvm/test/Transforms/InstCombine/constant-fold-address-space-pointer.ll
@@ -266,7 +266,8 @@ define i32 @constant_through_array_as_ptrs() {
define float @canonicalize_addrspacecast(i32 %i) {
; CHECK-LABEL: @canonicalize_addrspacecast(
-; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds float, ptr addrspacecast (ptr addrspace(3) @shared_mem to ptr), i32 [[I:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[I:%.*]], 2
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr addrspacecast (ptr addrspace(3) @shared_mem to ptr), i32 [[TMP1]]
; CHECK-NEXT: [[V:%.*]] = load float, ptr [[P]], align 4
; CHECK-NEXT: ret float [[V]]
;
diff --git a/llvm/test/Transforms/InstCombine/dependent-ivs.ll b/llvm/test/Transforms/InstCombine/dependent-ivs.ll
index bfd01c6b68941..96f5fd3e54933 100644
--- a/llvm/test/Transforms/InstCombine/dependent-ivs.ll
+++ b/llvm/test/Transforms/InstCombine/dependent-ivs.ll
@@ -606,7 +606,8 @@ define void @ptr_iv_non_i8_type(ptr %base, i64 %end) {
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[LOOP]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[IV_PTR:%.*]] = getelementptr i32, ptr [[BASE]], i64 [[IV]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[IV]], 2
+; CHECK-NEXT: [[IV_PTR:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[TMP0]]
; CHECK-NEXT: call void @use.p0(ptr [[IV_PTR]])
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 4
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[IV_NEXT]], [[END]]
diff --git a/llvm/test/Transforms/InstCombine/fmul.ll b/llvm/test/Transforms/InstCombine/fmul.ll
index d19fa41a5b0cd..040b858a74f8e 100644
--- a/llvm/test/Transforms/InstCombine/fmul.ll
+++ b/llvm/test/Transforms/InstCombine/fmul.ll
@@ -1099,7 +1099,8 @@ define void @fmul_loop_invariant_fdiv(ptr %a, float %x) {
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[IDXPROM:%.*]] = zext nneg i32 [[I_08]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[F:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[M:%.*]] = fdiv fast float [[F]], [[X:%.*]]
; CHECK-NEXT: store float [[M]], ptr [[ARRAYIDX]], align 4
diff --git a/llvm/test/Transforms/InstCombine/gep-addrspace.ll b/llvm/test/Transforms/InstCombine/gep-addrspace.ll
index fdd3b2e25dbfc..e93a5ac5b12d9 100644
--- a/llvm/test/Transforms/InstCombine/gep-addrspace.ll
+++ b/llvm/test/Transforms/InstCombine/gep-addrspace.ll
@@ -24,8 +24,10 @@ define void @func(ptr addrspace(1) nocapture %p) nounwind {
define void @keep_necessary_addrspacecast(i64 %i, ptr %out0, ptr %out1) {
; CHECK-LABEL: @keep_necessary_addrspacecast(
-; CHECK-NEXT: [[T1:%.*]] = getelementptr float, ptr addrspacecast (ptr addrspace(3) @array to ptr), i64 [[I:%.*]]
-; CHECK-NEXT: [[T2:%.*]] = getelementptr float, ptr addrspacecast (ptr addrspace(3) @scalar to ptr), i64 [[I]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[I:%.*]], 2
+; CHECK-NEXT: [[T1:%.*]] = getelementptr i8, ptr addrspacecast (ptr addrspace(3) @array to ptr), i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[I]], 2
+; CHECK-NEXT: [[T2:%.*]] = getelementptr i8, ptr addrspacecast (ptr addrspace(3) @scalar to ptr), i64 [[TMP2]]
; CHECK-NEXT: store ptr [[T1]], ptr [[OUT1:%.*]], align 4
; CHECK-NEXT: store ptr [[T2]], ptr [[OUT2:%.*]], align 4
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/InstCombine/gep-alias.ll b/llvm/test/Transforms/InstCombine/gep-alias.ll
index 0b573aa7c6967..206173b36ae3b 100644
--- a/llvm/test/Transforms/InstCombine/gep-alias.ll
+++ b/llvm/test/Transforms/InstCombine/gep-alias.ll
@@ -10,7 +10,8 @@ define i32 @f(i64 %i) {
; CHECK-LABEL: define i32 @f(
; CHECK-SAME: i64 [[I:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*:]]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr @x, i64 [[I]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr @x, i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: ret i32 [[TMP0]]
;
diff --git a/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll b/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll
index 8a4898ae923a7..d76dc29bf9739 100644
--- a/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll
+++ b/llvm/test/Transforms/InstCombine/gep-canonicalize-constant-indices.ll
@@ -14,8 +14,10 @@ declare void @use(i1)
define ptr @basic(ptr %p, i64 %a, i64 %b) {
; CHECK-LABEL: @basic(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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: [[TMP2:%.*]] = shl nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[B:%.*]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i64 [[TMP4]]
; CHECK-NEXT: ret ptr [[TMP3]]
;
%1 = getelementptr inbounds i32, ptr %p, i64 1
@@ -27,7 +29,8 @@ define ptr @basic(ptr %p, i64 %a, i64 %b) {
; GEP with the last index being a constant should also be swapped.
define ptr @partialConstant1(ptr %p, i64 %a, i64 %b) {
; CHECK-LABEL: @partialConstant1(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[B:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[B:%.*]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[TMP1]]
;
%1 = getelementptr inbounds [4 x i32], ptr %p, i64 %a, i64 1
@@ -38,7 +41,8 @@ define ptr @partialConstant1(ptr %p, i64 %a, i64 %b) {
; Negative test. GEP should not be swapped if the last index is not a constant.
define ptr @partialConstant2(ptr %p, i64 %a, i64 %b) {
; CHECK-LABEL: @partialConstant2(
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[B:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[B:%.*]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[TMP1]]
;
%1 = getelementptr inbounds [4 x i32], ptr %p, i64 1, i64 %a
@@ -50,7 +54,8 @@ 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 i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP1]], i64 12
; CHECK-NEXT: ret ptr [[TMP2]]
;
@@ -68,7 +73,8 @@ define ptr @nested(ptr %p, i64 %a, i64 %b) {
; CHECK-LABEL: @nested(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[A:%.*]]
; CHECK-NEXT: [[TMP2:%.*]] = mul i64 [[A]], [[B:%.*]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i16, ptr [[TMP1]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP2]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[TMP3]], i64 160
; CHECK-NEXT: ret ptr [[TMP4]]
;
@@ -86,7 +92,8 @@ define ptr @multipleUses1(ptr %p) {
; CHECK-LABEL: @multipleUses1(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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: [[TMP4:%.*]] = shl nsw i64 [[TMP2]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[TMP4]]
; CHECK-NEXT: ret ptr [[TMP3]]
;
%1 = getelementptr inbounds i32, ptr %p, i64 1
@@ -99,7 +106,8 @@ define ptr @multipleUses1(ptr %p) {
define ptr @multipleUses2(ptr %p, i64 %a) {
; CHECK-LABEL: @multipleUses2(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[P:%.*]], i64 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[TMP3]]
; CHECK-NEXT: call void @use(ptr nonnull [[TMP2]])
; CHECK-NEXT: ret ptr [[TMP2]]
;
@@ -114,7 +122,8 @@ define ptr @multipleUses3(ptr %p) {
; CHECK-LABEL: @multipleUses3(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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: [[TMP4:%.*]] = shl nsw i64 [[TMP2]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[TMP4]]
; CHECK-NEXT: ret ptr [[TMP3]]
;
%1 = getelementptr inbounds i32, ptr %p, i64 1
@@ -125,7 +134,8 @@ define ptr @multipleUses3(ptr %p) {
define ptr @merge_nuw(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_nuw(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr nuw i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -137,7 +147,8 @@ define ptr @merge_nuw(ptr %p, i64 %a) {
define ptr @merge_nuw_inbounds(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_nuw_inbounds(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds nuw i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -151,7 +162,8 @@ define ptr @merge_nuw_inbounds(ptr %p, i64 %a) {
; overflow.
define ptr @merge_nuw_nusw(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_nuw_nusw(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nusw nuw i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nusw nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr nuw i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -164,7 +176,8 @@ define ptr @merge_nuw_nusw(ptr %p, i64 %a) {
; Can't preserve nusw on the final GEP
define ptr @merge_nuw_nusw_overflow(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_nuw_nusw_overflow(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nusw nuw i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nusw nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr nuw i8, ptr [[GEP2]], i64 -2305843009213693952
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -176,7 +189,8 @@ define ptr @merge_nuw_nusw_overflow(ptr %p, i64 %a) {
define ptr @merge_missing_nuw1(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_missing_nuw1(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -188,7 +202,8 @@ define ptr @merge_missing_nuw1(ptr %p, i64 %a) {
define ptr @merge_missing_nuw2(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_missing_nuw2(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -200,7 +215,8 @@ define ptr @merge_missing_nuw2(ptr %p, i64 %a) {
define ptr @merge_missing_nuw3(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_missing_nuw3(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -212,7 +228,8 @@ define ptr @merge_missing_nuw3(ptr %p, i64 %a) {
define ptr @merge_nuw_missing_inbounds(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_nuw_missing_inbounds(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr nuw i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -224,7 +241,8 @@ define ptr @merge_nuw_missing_inbounds(ptr %p, i64 %a) {
define ptr @merge_nuw_missing_nusw(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_nuw_missing_nusw(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr nuw i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -236,7 +254,8 @@ define ptr @merge_nuw_missing_nusw(ptr %p, i64 %a) {
define ptr @merge_inbounds_missing_nuw(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_inbounds_missing_nuw(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
@@ -248,7 +267,8 @@ define ptr @merge_inbounds_missing_nuw(ptr %p, i64 %a) {
define ptr @merge_nusw_missing_nuw(ptr %p, i64 %a) {
; CHECK-LABEL: @merge_nusw_missing_nuw(
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[A:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[GEP2]], i64 5
; CHECK-NEXT: ret ptr [[GEP3]]
;
diff --git a/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll b/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
index 9b266b097057c..b1c3a178d12ab 100644
--- a/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
+++ b/llvm/test/Transforms/InstCombine/gep-combine-loop-invariant.ll
@@ -30,7 +30,8 @@ define i32 @foo(ptr nocapture readnone %match, i32 %cur_match, i32 %best_len, i3
; CHECK-NEXT: [[CHAIN_LENGTH_ADDR_08:%.*]] = phi i32 [ [[CHAIN_LENGTH:%.*]], [[IF_THEN_LR_PH]] ], [ [[DEC:%.*]], [[DO_BODY]] ]
; CHECK-NEXT: [[AND:%.*]] = and i32 [[CUR_MATCH_ADDR_09]], [[WMASK:%.*]]
; CHECK-NEXT: [[IDXPROM:%.*]] = zext i32 [[AND]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[PREV:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[PREV:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[I4]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CMP4:%.*]] = icmp ugt i32 [[I4]], [[LIMIT:%.*]]
; CHECK-NEXT: br i1 [[CMP4]], label [[LAND_LHS_TRUE:%.*]], label [[DO_END]]
@@ -93,9 +94,8 @@ define void @PR37005(ptr %base, ptr %in) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
-; CHECK-NEXT: [[E1:%.*]] = getelementptr inbounds ptr, ptr [[IN:%.*]], i64 undef
-; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds nuw i8, ptr [[E1]], i64 48
-; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds ptr, ptr [[E2]], <2 x i64> <i64 0, i64 1>
+; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds nuw i8, ptr [[E1:%.*]], i64 48
+; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds i8, ptr [[E2]], <2 x i64> <i64 0, i64 8>
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint <2 x ptr> [[E4]] to <2 x i64>
; CHECK-NEXT: [[TMP0:%.*]] = lshr <2 x i64> [[PI1]], splat (i64 14)
; CHECK-NEXT: [[SL1:%.*]] = and <2 x i64> [[TMP0]], splat (i64 1125899906842496)
@@ -125,8 +125,7 @@ define void @PR37005_2(ptr %base, ptr %in) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
-; CHECK-NEXT: [[E1:%.*]] = getelementptr inbounds ptr, ptr [[IN:%.*]], i64 undef
-; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds nuw i8, ptr [[E1]], i64 48
+; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds nuw 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
@@ -155,9 +154,8 @@ define void @PR37005_3(<2 x ptr> %base, ptr %in) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
-; CHECK-NEXT: [[E1:%.*]] = getelementptr inbounds ptr, ptr [[IN:%.*]], i64 undef
-; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds nuw i8, ptr [[E1]], i64 48
-; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds ptr, ptr [[E2]], <2 x i64> <i64 0, i64 1>
+; CHECK-NEXT: [[E2:%.*]] = getelementptr inbounds nuw i8, ptr [[E1:%.*]], i64 48
+; CHECK-NEXT: [[E4:%.*]] = getelementptr inbounds i8, ptr [[E2]], <2 x i64> <i64 0, i64 8>
; CHECK-NEXT: [[PI1:%.*]] = ptrtoint <2 x ptr> [[E4]] to <2 x i64>
; CHECK-NEXT: [[TMP0:%.*]] = lshr <2 x i64> [[PI1]], splat (i64 14)
; CHECK-NEXT: [[SL1:%.*]] = and <2 x i64> [[TMP0]], splat (i64 1125899906842496)
@@ -212,7 +210,8 @@ define float @gep_cross_loop(ptr %_arg_, ptr %_arg_3, float %_arg_8) {
; CHECK-LABEL: @gep_cross_loop(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[I:%.*]] = load i64, ptr [[_ARG_:%.*]], align 8
-; CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds float, ptr [[_ARG_3:%.*]], i64 [[I]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i8, ptr [[_ARG_3:%.*]], i64 [[TMP0]]
; CHECK-NEXT: br label [[FOR_COND_I:%.*]]
; CHECK: for.cond.i:
; CHECK-NEXT: [[IDX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[ADD11_I:%.*]], [[FOR_BODY_I:%.*]] ]
@@ -222,7 +221,8 @@ define float @gep_cross_loop(ptr %_arg_, ptr %_arg_3, float %_arg_8) {
; CHECK: for.cond.i.i.i.preheader:
; CHECK-NEXT: ret float [[SUM]]
; CHECK: for.body.i:
-; CHECK-NEXT: [[ARRAYIDX_I84_I:%.*]] = getelementptr inbounds nuw float, ptr [[ADD_PTR]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[ARRAYIDX_I84_I:%.*]] = getelementptr inbounds nuw i8, ptr [[ADD_PTR]], i64 [[TMP1]]
; CHECK-NEXT: [[I1:%.*]] = load float, ptr [[ARRAYIDX_I84_I]], align 4
; CHECK-NEXT: [[ADD_I]] = fadd fast float [[SUM]], [[I1]]
; CHECK-NEXT: [[ADD11_I]] = add nuw nsw i64 [[IDX]], 1
diff --git a/llvm/test/Transforms/InstCombine/gep-custom-dl.ll b/llvm/test/Transforms/InstCombine/gep-custom-dl.ll
index 4cd711a2582ae..e44251c5316e4 100644
--- a/llvm/test/Transforms/InstCombine/gep-custom-dl.ll
+++ b/llvm/test/Transforms/InstCombine/gep-custom-dl.ll
@@ -73,8 +73,10 @@ define void @test_evaluate_gep_as_ptrs_array(ptr addrspace(2) %B) {
define ptr @test4(ptr %I, i32 %C, i32 %D) {
; CHECK-LABEL: @test4(
-; CHECK-NEXT: [[A:%.*]] = getelementptr i32, ptr [[I:%.*]], i32 [[C:%.*]]
-; CHECK-NEXT: [[B:%.*]] = getelementptr i32, ptr [[A]], i32 [[D:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[C:%.*]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i32 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i32 [[D:%.*]], 2
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP2]]
; CHECK-NEXT: ret ptr [[B]]
;
%A = getelementptr i32, ptr %I, i32 %C
@@ -137,7 +139,8 @@ define ptr @test7(i16 %Idx) {
@Array = external global [40 x i32]
define ptr @test8(i32 %X) {
; CHECK-LABEL: @test8(
-; CHECK-NEXT: [[A:%.*]] = getelementptr i32, ptr @Array, i32 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[X:%.*]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr @Array, i32 [[TMP1]]
; CHECK-NEXT: ret ptr [[A]]
;
%A = getelementptr i32, ptr @Array, i32 %X
@@ -147,7 +150,8 @@ define ptr @test8(i32 %X) {
define ptr @test9(ptr %base, i8 %ind) {
; CHECK-LABEL: @test9(
; CHECK-NEXT: [[TMP1:%.*]] = sext i8 [[IND:%.*]] to i32
-; CHECK-NEXT: [[RES:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i32 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 2
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i32 [[TMP2]]
; CHECK-NEXT: ret ptr [[RES]]
;
%res = getelementptr i32, ptr %base, i8 %ind
diff --git a/llvm/test/Transforms/InstCombine/gep-fold-chained-const-select.ll b/llvm/test/Transforms/InstCombine/gep-fold-chained-const-select.ll
index 5dd220abda648..fea13dc52875b 100644
--- a/llvm/test/Transforms/InstCombine/gep-fold-chained-const-select.ll
+++ b/llvm/test/Transforms/InstCombine/gep-fold-chained-const-select.ll
@@ -88,10 +88,9 @@ define <2 x ptr> @src_splat_scalar_ptr(i32 %arg0, ptr %arg1) {
; Fail 1: Different GEP type
define ptr @src_fail_different_type(i32 %arg0, ptr %arg1) {
; CHECK-LABEL: @src_fail_different_type(
-; CHECK-NEXT: [[V0:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG1:%.*]], i64 8148
; CHECK-NEXT: [[V1:%.*]] = icmp sgt i32 [[ARG0:%.*]], 3
-; CHECK-NEXT: [[V2:%.*]] = select i1 [[V1]], i64 55104, i64 21304
-; CHECK-NEXT: [[V3:%.*]] = getelementptr i16, ptr [[V0]], i64 [[V2]]
+; CHECK-NEXT: [[TMP1:%.*]] = select i1 [[V1]], i64 118356, i64 50756
+; CHECK-NEXT: [[V3:%.*]] = getelementptr i8, ptr [[ARG1:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[V3]]
;
%v0 = getelementptr inbounds nuw i8, ptr %arg1, i64 8148
diff --git a/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll b/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll
index ecebaa7d9ae5d..9ee89f76d698f 100644
--- a/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll
+++ b/llvm/test/Transforms/InstCombine/gep-merge-constant-indices.ll
@@ -118,7 +118,8 @@ define ptr @structStruct(ptr %p) {
; result = (ptr) &((struct.B*) p)[i].member1 + 2
define ptr @appendIndex(ptr %p, i64 %i) {
; CHECK-LABEL: @appendIndex(
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[STRUCT_B:%.*]], ptr [[P:%.*]], i64 [[I:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = mul nsw i64 [[I:%.*]], 140
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 6
; CHECK-NEXT: ret ptr [[TMP1]]
;
@@ -174,7 +175,8 @@ define ptr @partialConstant3(ptr %p) {
; result = &((struct.C*) p + a).member2
define ptr @partialConstantMemberAliasing1(ptr %p, i64 %a) {
; CHECK-LABEL: @partialConstantMemberAliasing1(
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = mul nsw i64 [[A:%.*]], 12
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 8
; CHECK-NEXT: ret ptr [[TMP1]]
;
@@ -187,7 +189,8 @@ define ptr @partialConstantMemberAliasing1(ptr %p, i64 %a) {
; address of another member.
define ptr @partialConstantMemberAliasing2(ptr %p, i64 %a) {
; CHECK-LABEL: @partialConstantMemberAliasing2(
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul nsw i64 [[A:%.*]], 12
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 5
; CHECK-NEXT: ret ptr [[TMP2]]
;
@@ -200,7 +203,8 @@ define ptr @partialConstantMemberAliasing2(ptr %p, i64 %a) {
; range of the object currently pointed by the non-constant GEP.
define ptr @partialConstantMemberAliasing3(ptr %p, i64 %a) {
; CHECK-LABEL: @partialConstantMemberAliasing3(
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[A:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul nsw i64 [[A:%.*]], 12
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 12
; CHECK-NEXT: ret ptr [[TMP2]]
;
diff --git a/llvm/test/Transforms/InstCombine/gep-sext.ll b/llvm/test/Transforms/InstCombine/gep-sext.ll
index c1998776bde2a..b39180e1bafbf 100644
--- a/llvm/test/Transforms/InstCombine/gep-sext.ll
+++ b/llvm/test/Transforms/InstCombine/gep-sext.ll
@@ -8,7 +8,8 @@ define void @test(ptr %p, i32 %index) {
; CHECK-LABEL: define void @test(
; CHECK-SAME: ptr [[P:%.*]], i32 [[INDEX:%.*]]) {
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[INDEX]] to i64
-; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[P]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP2]]
; CHECK-NEXT: [[VAL:%.*]] = load i32, ptr [[ADDR]], align 4
; CHECK-NEXT: call void @use(i32 [[VAL]])
; CHECK-NEXT: ret void
@@ -23,7 +24,8 @@ define void @test2(ptr %p, i32 %index) {
; CHECK-LABEL: define void @test2(
; CHECK-SAME: ptr [[P:%.*]], i32 [[INDEX:%.*]]) {
; CHECK-NEXT: [[I:%.*]] = zext i32 [[INDEX]] to i64
-; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I]], 2
+; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP1]]
; CHECK-NEXT: [[VAL:%.*]] = load i32, ptr [[ADDR]], align 4
; CHECK-NEXT: call void @use(i32 [[VAL]])
; CHECK-NEXT: ret void
@@ -43,7 +45,8 @@ define void @test3(ptr %p, i32 %index) {
; CHECK-NEXT: [[ADDR_FIXED:%.*]] = getelementptr i8, ptr [[P]], i64 352
; CHECK-NEXT: [[VAL_FIXED:%.*]] = load i32, ptr [[ADDR_FIXED]], align 4, !range [[RNG0:![0-9]+]]
; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[VAL_FIXED]] to i64
-; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ADDR_BEGIN]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i8, ptr [[ADDR_BEGIN]], i64 [[TMP2]]
; CHECK-NEXT: [[VAL:%.*]] = load i32, ptr [[ADDR]], align 4
; CHECK-NEXT: call void @use(i32 [[VAL]])
; CHECK-NEXT: ret void
@@ -64,7 +67,8 @@ define void @test4(ptr %p, i32 %index) {
; CHECK-NEXT: [[ADDR_FIXED:%.*]] = getelementptr i8, ptr [[P]], i64 352
; CHECK-NEXT: [[VAL_FIXED:%.*]] = load i32, ptr [[ADDR_FIXED]], align 4, !range [[RNG0]]
; CHECK-NEXT: [[I:%.*]] = zext nneg i32 [[VAL_FIXED]] to i64
-; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i32, ptr [[ADDR_BEGIN]], i64 [[I]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I]], 2
+; CHECK-NEXT: [[ADDR:%.*]] = getelementptr i8, ptr [[ADDR_BEGIN]], i64 [[TMP1]]
; CHECK-NEXT: [[VAL:%.*]] = load i32, ptr [[ADDR]], align 4
; CHECK-NEXT: call void @use(i32 [[VAL]])
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/InstCombine/gep-srem-to-and-deref.ll b/llvm/test/Transforms/InstCombine/gep-srem-to-and-deref.ll
index 9f529dad5d3d1..d9e16897abf41 100644
--- a/llvm/test/Transforms/InstCombine/gep-srem-to-and-deref.ll
+++ b/llvm/test/Transforms/InstCombine/gep-srem-to-and-deref.ll
@@ -40,8 +40,9 @@ define ptr @pos_pow2_8(ptr %foo, i64 %x) {
define ptr @pos_not_i8(ptr %foo, i64 %x) {
; CHECK-LABEL: define ptr @pos_not_i8(
; CHECK-SAME: ptr [[FOO:%.*]], i64 [[X:%.*]]) {
-; CHECK-NEXT: [[IDX1:%.*]] = and i64 [[X]], 3
-; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds nuw i32, ptr [[FOO]], i64 [[IDX1]]
+; CHECK-NEXT: [[IDX:%.*]] = srem i64 [[X]], 4
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds nuw i8, ptr [[FOO]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
%idx = srem i64 %x, 4
diff --git a/llvm/test/Transforms/InstCombine/gep-vector-indices.ll b/llvm/test/Transforms/InstCombine/gep-vector-indices.ll
index ce584e0400cd5..c2a0d01e84484 100644
--- a/llvm/test/Transforms/InstCombine/gep-vector-indices.ll
+++ b/llvm/test/Transforms/InstCombine/gep-vector-indices.ll
@@ -51,7 +51,8 @@ define ptr @vector_splat_indices_nxv2i64_ext0_nusw(ptr %a) {
define ptr @vector_indices_v2i64_ext0(ptr %a, <2 x i64> %indices) {
; CHECK-LABEL: @vector_indices_v2i64_ext0(
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x i64> [[INDICES:%.*]], i64 0
-; CHECK-NEXT: [[RES:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[TMP1]], 2
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[RES]]
;
%gep = getelementptr i32, ptr %a, <2 x i64> %indices
@@ -62,7 +63,8 @@ define ptr @vector_indices_v2i64_ext0(ptr %a, <2 x i64> %indices) {
define ptr @vector_indices_nxv1i64_ext0(ptr %a, <vscale x 1 x i64> %indices) {
; CHECK-LABEL: @vector_indices_nxv1i64_ext0(
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <vscale x 1 x i64> [[INDICES:%.*]], i64 0
-; CHECK-NEXT: [[RES:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[TMP1]], 2
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[RES]]
;
%gep = getelementptr i32, ptr %a, <vscale x 1 x i64> %indices
@@ -73,7 +75,8 @@ define ptr @vector_indices_nxv1i64_ext0(ptr %a, <vscale x 1 x i64> %indices) {
define ptr @vector_splat_ptrs_v2i64_ext0(ptr %a, i64 %index) {
; CHECK-LABEL: @vector_splat_ptrs_v2i64_ext0(
-; CHECK-NEXT: [[RES:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[INDEX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[INDEX:%.*]], 2
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[RES]]
;
%tmp = insertelement <2 x ptr> poison, ptr %a, i32 0
@@ -86,7 +89,8 @@ define ptr @vector_splat_ptrs_v2i64_ext0(ptr %a, i64 %index) {
define ptr @vector_splat_ptrs_nxv2i64_ext0(ptr %a, i64 %index) {
; CHECK-LABEL: @vector_splat_ptrs_nxv2i64_ext0(
-; CHECK-NEXT: [[RES:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[INDEX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[INDEX:%.*]], 2
+; CHECK-NEXT: [[RES:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[RES]]
;
%tmp = insertelement <vscale x 2 x ptr> poison, ptr %a, i32 0
@@ -123,7 +127,8 @@ define ptr @vector_struct2_splat_indices_v4i64_ext1(ptr %a) {
define ptr @vector_indices_nxv2i64_ext3(ptr %a, <vscale x 2 x i64> %indices) {
; CHECK-LABEL: @vector_indices_nxv2i64_ext3(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[A:%.*]], <vscale x 2 x i64> [[INDICES:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <vscale x 2 x i64> [[INDICES:%.*]], splat (i64 2)
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[A:%.*]], <vscale x 2 x i64> [[TMP1]]
; CHECK-NEXT: [[RES:%.*]] = extractelement <vscale x 2 x ptr> [[GEP]], i64 3
; CHECK-NEXT: ret ptr [[RES]]
;
@@ -134,7 +139,8 @@ define ptr @vector_indices_nxv2i64_ext3(ptr %a, <vscale x 2 x i64> %indices) {
define ptr @vector_indices_nxv2i64_extN(ptr %a, <vscale x 2 x i64> %indices, i32 %N) {
; CHECK-LABEL: @vector_indices_nxv2i64_extN(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[A:%.*]], <vscale x 2 x i64> [[INDICES:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <vscale x 2 x i64> [[INDICES:%.*]], splat (i64 2)
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[A:%.*]], <vscale x 2 x i64> [[TMP1]]
; CHECK-NEXT: [[RES:%.*]] = extractelement <vscale x 2 x ptr> [[GEP]], i32 [[N:%.*]]
; CHECK-NEXT: ret ptr [[RES]]
;
@@ -145,7 +151,8 @@ define ptr @vector_indices_nxv2i64_extN(ptr %a, <vscale x 2 x i64> %indices, i32
define void @vector_indices_nxv2i64_mulitple_use(ptr %a, <vscale x 2 x i64> %indices, ptr %b, ptr %c) {
; CHECK-LABEL: @vector_indices_nxv2i64_mulitple_use(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[A:%.*]], <vscale x 2 x i64> [[INDICES:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <vscale x 2 x i64> [[INDICES:%.*]], splat (i64 2)
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[A:%.*]], <vscale x 2 x i64> [[TMP1]]
; CHECK-NEXT: [[LANE0:%.*]] = extractelement <vscale x 2 x ptr> [[GEP]], i64 0
; CHECK-NEXT: [[LANE1:%.*]] = extractelement <vscale x 2 x ptr> [[GEP]], i64 1
; CHECK-NEXT: store ptr [[LANE0]], ptr [[B:%.*]], align 8
@@ -162,7 +169,8 @@ define void @vector_indices_nxv2i64_mulitple_use(ptr %a, <vscale x 2 x i64> %ind
define ptr @vector_ptrs_and_indices_ext0(<vscale x 4 x ptr> %a, <vscale x 4 x i64> %indices) {
; CHECK-LABEL: @vector_ptrs_and_indices_ext0(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, <vscale x 4 x ptr> [[A:%.*]], <vscale x 4 x i64> [[INDICES:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <vscale x 4 x i64> [[INDICES:%.*]], splat (i64 2)
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, <vscale x 4 x ptr> [[A:%.*]], <vscale x 4 x i64> [[TMP1]]
; CHECK-NEXT: [[RES:%.*]] = extractelement <vscale x 4 x ptr> [[GEP]], i64 0
; CHECK-NEXT: ret ptr [[RES]]
;
diff --git a/llvm/test/Transforms/InstCombine/gep-vector.ll b/llvm/test/Transforms/InstCombine/gep-vector.ll
index 920147b3576d2..70e282804e52a 100644
--- a/llvm/test/Transforms/InstCombine/gep-vector.ll
+++ b/llvm/test/Transforms/InstCombine/gep-vector.ll
@@ -31,8 +31,10 @@ define <2 x ptr> @vectorindex3() {
define ptr @bitcast_vec_to_array_gep(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @bitcast_vec_to_array_gep(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr [7 x i32], ptr [[GEP_SPLIT:%.*]], i64 [[Z:%.*]]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[GEP]], i64 [[Z1:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul i64 [[Y:%.*]], 28
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP1]]
;
%gep = getelementptr [7 x i32], ptr %x, i64 %y, i64 %z
@@ -43,8 +45,10 @@ define ptr @bitcast_vec_to_array_gep(ptr %x, i64 %y, i64 %z) {
define ptr @bitcast_array_to_vec_gep(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @bitcast_array_to_vec_gep(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds <3 x i32>, ptr [[GEP_SPLIT:%.*]], i64 [[Z:%.*]]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[GEP]], i64 [[Z1:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[Y:%.*]], 4
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP1]]
;
%gep = getelementptr inbounds <3 x i32>, ptr %x, i64 %y, i64 %z
@@ -55,8 +59,10 @@ define ptr @bitcast_array_to_vec_gep(ptr %x, i64 %y, i64 %z) {
define ptr @bitcast_vec_to_array_gep_matching_alloc_size(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @bitcast_vec_to_array_gep_matching_alloc_size(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr [4 x i32], ptr [[GEP_SPLIT:%.*]], i64 [[Z:%.*]]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[GEP]], i64 [[Z1:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[Y:%.*]], 4
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP1]]
;
%gep = getelementptr [4 x i32], ptr %x, i64 %y, i64 %z
@@ -67,8 +73,10 @@ define ptr @bitcast_vec_to_array_gep_matching_alloc_size(ptr %x, i64 %y, i64 %z)
define ptr @bitcast_array_to_vec_gep_matching_alloc_size(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @bitcast_array_to_vec_gep_matching_alloc_size(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds <4 x i32>, ptr [[GEP_SPLIT:%.*]], i64 [[Z:%.*]]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[GEP]], i64 [[Z1:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[Y:%.*]], 4
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP1]]
;
%gep = getelementptr inbounds <4 x i32>, ptr %x, i64 %y, i64 %z
@@ -80,8 +88,10 @@ define ptr @bitcast_array_to_vec_gep_matching_alloc_size(ptr %x, i64 %y, i64 %z)
define ptr addrspace(3) @bitcast_vec_to_array_addrspace(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @bitcast_vec_to_array_addrspace(
; CHECK-NEXT: [[ASC:%.*]] = addrspacecast ptr [[X:%.*]] to ptr addrspace(3)
-; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr [7 x i32], ptr addrspace(3) [[ASC]], i64 [[Y:%.*]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr addrspace(3) [[GEP_SPLIT]], i64 [[Z:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul i64 [[Y:%.*]], 28
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr i8, ptr addrspace(3) [[ASC]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr addrspace(3) [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr addrspace(3) [[GEP]]
;
%asc = addrspacecast ptr %x to ptr addrspace(3)
@@ -94,8 +104,10 @@ define ptr addrspace(3) @bitcast_vec_to_array_addrspace(ptr %x, i64 %y, i64 %z)
define ptr addrspace(3) @inbounds_bitcast_vec_to_array_addrspace(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @inbounds_bitcast_vec_to_array_addrspace(
; CHECK-NEXT: [[ASC:%.*]] = addrspacecast ptr [[X:%.*]] to ptr addrspace(3)
-; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds [7 x i32], ptr addrspace(3) [[ASC]], i64 [[Y:%.*]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr addrspace(3) [[GEP_SPLIT]], i64 [[Z:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul nsw i64 [[Y:%.*]], 28
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds i8, ptr addrspace(3) [[ASC]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr addrspace(3) [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr addrspace(3) [[GEP]]
;
%asc = addrspacecast ptr %x to ptr addrspace(3)
@@ -108,8 +120,10 @@ define ptr addrspace(3) @inbounds_bitcast_vec_to_array_addrspace(ptr %x, i64 %y,
define ptr addrspace(3) @bitcast_vec_to_array_addrspace_matching_alloc_size(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @bitcast_vec_to_array_addrspace_matching_alloc_size(
; CHECK-NEXT: [[ASC:%.*]] = addrspacecast ptr [[X:%.*]] to ptr addrspace(3)
-; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr [4 x i32], ptr addrspace(3) [[ASC]], i64 [[Y:%.*]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr addrspace(3) [[GEP_SPLIT]], i64 [[Z:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[Y:%.*]], 4
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr i8, ptr addrspace(3) [[ASC]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr addrspace(3) [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr addrspace(3) [[GEP]]
;
%asc = addrspacecast ptr %x to ptr addrspace(3)
@@ -122,8 +136,10 @@ define ptr addrspace(3) @bitcast_vec_to_array_addrspace_matching_alloc_size(ptr
define ptr addrspace(3) @inbounds_bitcast_vec_to_array_addrspace_matching_alloc_size(ptr %x, i64 %y, i64 %z) {
; CHECK-LABEL: @inbounds_bitcast_vec_to_array_addrspace_matching_alloc_size(
; CHECK-NEXT: [[ASC:%.*]] = addrspacecast ptr [[X:%.*]] to ptr addrspace(3)
-; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds [4 x i32], ptr addrspace(3) [[ASC]], i64 [[Y:%.*]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr addrspace(3) [[GEP_SPLIT]], i64 [[Z:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[Y:%.*]], 4
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds i8, ptr addrspace(3) [[ASC]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Z:%.*]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr addrspace(3) [[GEP_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr addrspace(3) [[GEP]]
;
%asc = addrspacecast ptr %x to ptr addrspace(3)
diff --git a/llvm/test/Transforms/InstCombine/gepofconstgepi8.ll b/llvm/test/Transforms/InstCombine/gepofconstgepi8.ll
index 42a4619645994..01a628a59b6f3 100644
--- a/llvm/test/Transforms/InstCombine/gepofconstgepi8.ll
+++ b/llvm/test/Transforms/InstCombine/gepofconstgepi8.ll
@@ -8,7 +8,8 @@ define ptr @test_zero(ptr %base, i64 %a) {
; CHECK-LABEL: define ptr @test_zero(
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[BASE]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -23,7 +24,8 @@ define ptr @test_nonzero(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[BASE]], i64 4
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -37,7 +39,8 @@ define ptr @test_or_disjoint(ptr %base, i64 %a) {
; CHECK-LABEL: define ptr @test_or_disjoint(
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[BASE]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -53,7 +56,8 @@ define ptr @test_zero_multiuse_index(ptr %base, i64 %a) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[INDEX:%.*]] = add i64 [[A]], 1
; CHECK-NEXT: call void @use64(i64 [[INDEX]])
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[BASE]], i64 [[A]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[TMP0]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -70,7 +74,8 @@ define ptr @test_zero_multiuse_ptr(ptr %base, i64 %a) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
; CHECK-NEXT: call void @useptr(ptr [[P1]])
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[BASE]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -85,8 +90,11 @@ define ptr @test_zero_sext_add_nsw(ptr %base, i32 %a) {
; CHECK-LABEL: define ptr @test_zero_sext_add_nsw(
; CHECK-SAME: ptr [[BASE:%.*]], i32 [[A:%.*]]) {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[A]] to i64
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[BASE]], i64 [[TMP0]]
+; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
+; CHECK-NEXT: [[INDEX:%.*]] = add nsw i32 [[A]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[P1]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -101,7 +109,8 @@ define ptr @test_zero_trunc_add(ptr %base, i128 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i128 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = trunc i128 [[A]] to i64
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[BASE]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[TMP0]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -116,7 +125,8 @@ define ptr @test_non_i8(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i32, ptr [[P1]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[TMP0]]
;
entry:
@@ -131,7 +141,8 @@ define ptr @test_non_const(ptr %base, i64 %a, i64 %b) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]], i64 [[B:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[B]]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i32, ptr [[P1]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 4
; CHECK-NEXT: ret ptr [[P2]]
;
@@ -148,7 +159,8 @@ define ptr @test_too_many_indices(ptr %base, i64 %a, i64 %b) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 [[B]]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P1]], i64 36
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -163,8 +175,9 @@ define ptr @test_wrong_op(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
-; CHECK-NEXT: [[INDEX:%.*]] = xor i64 [[A]], 1
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[P1]], i64 [[INDEX]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = xor i64 [[INDEX]], 4
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[P1]], i64 [[TMP0]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -181,7 +194,8 @@ define ptr @test_sext_add_without_nsw(ptr %base, i32 %a) {
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
; CHECK-NEXT: [[INDEX:%.*]] = add i32 [[A]], 1
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[P1]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[P1]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -196,8 +210,9 @@ define ptr @test_or_without_disjoint(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
-; CHECK-NEXT: [[INDEX:%.*]] = or i64 [[A]], 1
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[P1]], i64 [[INDEX]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = or i64 [[INDEX]], 4
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[P1]], i64 [[TMP0]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -212,7 +227,8 @@ define ptr @test_smul_overflow(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -12
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i32, ptr [[P1]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[TMP0]]
;
entry:
@@ -227,7 +243,8 @@ define ptr @test_sadd_overflow(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -9223372036854775808
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i32, ptr [[P1]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[TMP0]]
;
entry:
@@ -244,7 +261,8 @@ define ptr @test_nonzero_multiuse_index(ptr %base, i64 %a) {
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
; CHECK-NEXT: [[INDEX:%.*]] = add i64 [[A]], 2
; CHECK-NEXT: call void @use64(i64 [[INDEX]])
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[P1]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[P1]], i64 [[TMP0]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -261,7 +279,8 @@ define ptr @test_nonzero_multiuse_ptr(ptr %base, i64 %a) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[P1:%.*]] = getelementptr i8, ptr [[BASE]], i64 -4
; CHECK-NEXT: call void @useptr(ptr [[P1]])
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i32, ptr [[P1]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P1]], i64 [[INDEX]]
; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 8
; CHECK-NEXT: ret ptr [[P2]]
;
@@ -297,7 +316,8 @@ define ptr @test_all_nuw(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nuw i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i8, ptr [[TMP0]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -312,7 +332,8 @@ define ptr @test_all_partial_nuw1(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -327,7 +348,8 @@ define ptr @test_all_partial_nuw2(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -342,7 +364,8 @@ define ptr @test_all_partial_nuw3(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -356,8 +379,9 @@ define ptr @test_all_nuw_disjoint(ptr %base, i64 %a) {
; CHECK-LABEL: define ptr @test_all_nuw_disjoint(
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nuw i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[P1:%.*]] = getelementptr nuw i8, ptr [[BASE]], i64 9
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i8, ptr [[P1]], i64 [[TMP0]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -372,7 +396,8 @@ define ptr @test_all_inbounds_nuw(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -387,7 +412,8 @@ define ptr @test_all_partial_inbounds1(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nuw i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i8, ptr [[TMP0]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -402,7 +428,8 @@ define ptr @test_all_partial_inbounds2(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nuw i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i8, ptr [[TMP0]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -417,7 +444,8 @@ define ptr @test_all_inbounds_partial_nuw1(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[BASE]], i64 7
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -432,7 +460,8 @@ define ptr @test_all_inbounds_partial_nuw2(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -447,7 +476,8 @@ define ptr @test_all_inbounds_partial_nuw3(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[INDEX:%.*]] = shl i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
@@ -462,7 +492,8 @@ define ptr @test_all_nusw_nuw(ptr %base, i64 %a) {
; CHECK-SAME: ptr [[BASE:%.*]], i64 [[A:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr nuw i8, ptr [[BASE]], i64 9
-; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i32, ptr [[TMP0]], i64 [[A]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[A]], 2
+; CHECK-NEXT: [[P2:%.*]] = getelementptr nuw i8, ptr [[TMP0]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[P2]]
;
entry:
diff --git a/llvm/test/Transforms/InstCombine/gepphigep.ll b/llvm/test/Transforms/InstCombine/gepphigep.ll
index 6a6b7708bbfc9..41c11e7e36205 100644
--- a/llvm/test/Transforms/InstCombine/gepphigep.ll
+++ b/llvm/test/Transforms/InstCombine/gepphigep.ll
@@ -12,16 +12,18 @@ define i32 @test1(ptr %dm, i1 %c, i64 %idx1, i64 %idx2) {
; CHECK-NEXT: [[INST1:%.*]] = load ptr, ptr [[DM:%.*]], align 8
; CHECK-NEXT: br i1 [[C:%.*]], label [[BB1:%.*]], label [[BB2:%.*]]
; CHECK: bb1:
-; CHECK-NEXT: [[INST10:%.*]] = getelementptr inbounds [[STRUCT2:%.*]], ptr [[INST1]], i64 [[IDX1:%.*]]
+; CHECK-NEXT: [[IDX1:%.*]] = shl nsw i64 [[IDX3:%.*]], 3
+; CHECK-NEXT: [[INST10:%.*]] = getelementptr inbounds i8, ptr [[INST1]], i64 [[IDX1]]
; CHECK-NEXT: store i32 0, ptr [[INST10]], align 4
; CHECK-NEXT: br label [[BB3:%.*]]
; CHECK: bb2:
-; CHECK-NEXT: [[INST20:%.*]] = getelementptr inbounds [[STRUCT2]], ptr [[INST1]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[IDX2:%.*]] = shl nsw i64 [[IDX4:%.*]], 3
+; CHECK-NEXT: [[INST20:%.*]] = getelementptr inbounds i8, ptr [[INST1]], i64 [[IDX2]]
; CHECK-NEXT: store i32 0, ptr [[INST20]], align 4
; CHECK-NEXT: br label [[BB3]]
; CHECK: bb3:
; CHECK-NEXT: [[TMP0:%.*]] = phi i64 [ [[IDX1]], [[BB1]] ], [ [[IDX2]], [[BB2]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT2]], ptr [[INST1]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[INST1]], i64 [[TMP0]]
; CHECK-NEXT: [[INST24:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 4
; CHECK-NEXT: [[INST25:%.*]] = load i32, ptr [[INST24]], align 4
; CHECK-NEXT: ret i32 [[INST25]]
@@ -51,9 +53,11 @@ define i32 @test2(ptr %dm, i64 %idx1, i64 %idx2) {
; CHECK-LABEL: @test2(
; CHECK-NEXT: bb:
; CHECK-NEXT: [[INST1:%.*]] = load ptr, ptr [[DM:%.*]], align 8
-; CHECK-NEXT: [[INST10:%.*]] = getelementptr inbounds [[STRUCT2:%.*]], ptr [[INST1]], i64 [[IDX1:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IDX1:%.*]], 3
+; CHECK-NEXT: [[INST10:%.*]] = getelementptr inbounds i8, ptr [[INST1]], i64 [[TMP0]]
; CHECK-NEXT: store i32 0, ptr [[INST10]], align 4
-; CHECK-NEXT: [[INST20:%.*]] = getelementptr inbounds [[STRUCT2]], ptr [[INST1]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX2:%.*]], 3
+; CHECK-NEXT: [[INST20:%.*]] = getelementptr inbounds i8, ptr [[INST1]], i64 [[TMP1]]
; CHECK-NEXT: store i32 0, ptr [[INST20]], align 4
; CHECK-NEXT: [[INST24:%.*]] = getelementptr inbounds nuw i8, ptr [[INST10]], i64 4
; CHECK-NEXT: [[INST25:%.*]] = load i32, ptr [[INST24]], align 4
@@ -77,12 +81,14 @@ define i32 @test3(ptr %dm, i1 %c, i64 %idx1, i64 %idx2, i64 %idx3) personality p
; CHECK-NEXT: bb:
; CHECK-NEXT: br i1 [[C:%.*]], label [[BB1:%.*]], label [[BB2:%.*]]
; CHECK: bb1:
-; CHECK-NEXT: [[INST1_SPLIT:%.*]] = getelementptr inbounds [[STRUCT3:%.*]], ptr [[DM:%.*]], i64 [[IDX1:%.*]]
+; CHECK-NEXT: [[IDX1:%.*]] = mul nsw i64 [[IDX4:%.*]], 36
+; CHECK-NEXT: [[INST1_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[DM:%.*]], i64 [[IDX1]]
; CHECK-NEXT: [[INST1:%.*]] = getelementptr inbounds nuw i8, ptr [[INST1_SPLIT]], i64 4
; CHECK-NEXT: store i32 0, ptr [[INST1]], align 4
; CHECK-NEXT: br label [[BB3:%.*]]
; CHECK: bb2:
-; CHECK-NEXT: [[INST2_SPLIT:%.*]] = getelementptr inbounds [[STRUCT3]], ptr [[DM]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[IDX2:%.*]] = mul nsw i64 [[IDX5:%.*]], 36
+; CHECK-NEXT: [[INST2_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[DM]], i64 [[IDX2]]
; CHECK-NEXT: [[INST12:%.*]] = getelementptr inbounds nuw i8, ptr [[INST2_SPLIT]], i64 8
; CHECK-NEXT: store i32 0, ptr [[INST12]], align 4
; CHECK-NEXT: br label [[BB3]]
@@ -95,8 +101,9 @@ define i32 @test3(ptr %dm, i1 %c, i64 %idx1, i64 %idx2, i64 %idx3) personality p
; CHECK: bb5:
; CHECK-NEXT: [[INST27:%.*]] = landingpad { ptr, i32 }
; CHECK-NEXT: catch ptr @_ZTIi
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT3]], ptr [[DM]], i64 [[TMP0]]
-; CHECK-NEXT: [[INST34_SPLIT:%.*]] = getelementptr [[STRUCT4:%.*]], ptr [[TMP1]], i64 [[IDX3:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[DM]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[IDX3:%.*]], 4
+; CHECK-NEXT: [[INST34_SPLIT:%.*]] = getelementptr i8, ptr [[TMP3]], i64 [[TMP4]]
; CHECK-NEXT: [[INST35:%.*]] = getelementptr i8, ptr [[INST34_SPLIT]], i64 16
; CHECK-NEXT: [[INST25:%.*]] = load i32, ptr [[INST35]], align 4
; CHECK-NEXT: ret i32 [[INST25]]
@@ -215,7 +222,8 @@ define void @test5(ptr %idx, ptr %in) #0 {
; CHECK: while.cond.57:
; CHECK-NEXT: [[TMP3:%.*]] = load i8, ptr [[INCDEC_PTR34]], align 1
; CHECK-NEXT: [[TMP4:%.*]] = zext i8 [[TMP3]] to i64
-; CHECK-NEXT: [[ARRAYIDX61:%.*]] = getelementptr inbounds nuw i16, ptr [[IDX:%.*]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP4]], 1
+; CHECK-NEXT: [[ARRAYIDX61:%.*]] = getelementptr inbounds nuw i8, ptr [[IDX:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP5:%.*]] = load i16, ptr [[ARRAYIDX61]], align 2
; CHECK-NEXT: [[AND63:%.*]] = and i16 [[TMP5]], 2048
; CHECK-NEXT: [[TOBOOL64:%.*]] = icmp eq i16 [[AND63]], 0
diff --git a/llvm/test/Transforms/InstCombine/getelementptr.ll b/llvm/test/Transforms/InstCombine/getelementptr.ll
index 92b76c5d1b46a..530baa390fce9 100644
--- a/llvm/test/Transforms/InstCombine/getelementptr.ll
+++ b/llvm/test/Transforms/InstCombine/getelementptr.ll
@@ -125,8 +125,10 @@ define void @test_overaligned_vec(i8 %B) {
define ptr @test7(ptr %I, i64 %C, i64 %D) {
; CHECK-LABEL: @test7(
-; CHECK-NEXT: [[A:%.*]] = getelementptr i32, ptr [[I:%.*]], i64 [[C:%.*]]
-; CHECK-NEXT: [[B:%.*]] = getelementptr i32, ptr [[A]], i64 [[D:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[C:%.*]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[D:%.*]], 2
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[B]]
;
%A = getelementptr i32, ptr %I, i64 %C
@@ -233,9 +235,9 @@ define <2 x i1> @test13_vector(<2 x i64> %X, <2 x ptr> %P) nounwind {
define <2 x i1> @test13_vector2(i64 %X, <2 x ptr> %P) nounwind {
; CHECK-LABEL: @test13_vector2(
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[X:%.*]], i64 0
-; CHECK-NEXT: [[TMP1:%.*]] = shl <2 x i64> [[DOTSPLATINSERT]], <i64 2, i64 0>
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i64> [[TMP1]], <i64 -4, i64 poison>
+; CHECK-NEXT: [[X:%.*]] = shl nsw i64 [[X1:%.*]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[X]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i64> [[DOTSPLATINSERT]], <i64 -4, i64 poison>
; CHECK-NEXT: [[C:%.*]] = shufflevector <2 x i1> [[TMP2]], <2 x i1> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: ret <2 x i1> [[C]]
;
@@ -247,10 +249,12 @@ define <2 x i1> @test13_vector2(i64 %X, <2 x ptr> %P) nounwind {
define <2 x i1> @test13_fixed_fixed(i64 %X, ptr %P, <2 x i64> %y) nounwind {
; CHECK-LABEL: @test13_fixed_fixed(
-; CHECK-NEXT: [[A1:%.*]] = getelementptr inbounds i64, ptr [[P:%.*]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 3
+; CHECK-NEXT: [[A1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x ptr> poison, ptr [[A1]], i64 0
; CHECK-NEXT: [[DOTSPLAT:%.*]] = shufflevector <2 x ptr> [[DOTSPLATINSERT]], <2 x ptr> poison, <2 x i32> zeroinitializer
-; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds <2 x i64>, ptr [[P]], <2 x i64> [[Y:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw <2 x i64> [[Y:%.*]], splat (i64 4)
+; CHECK-NEXT: [[B:%.*]] = getelementptr inbounds i8, ptr [[P]], <2 x i64> [[TMP2]]
; CHECK-NEXT: [[C:%.*]] = icmp eq <2 x ptr> [[DOTSPLAT]], [[B]]
; CHECK-NEXT: ret <2 x i1> [[C]]
;
@@ -301,9 +305,9 @@ define <vscale x 2 x i1> @test13_scalable_scalable(i64 %X, ptr %P, <vscale x 2 x
; This is a test of icmp + shl nuw in disguise - 4611... is 0x3fff...
define <2 x i1> @test13_vector3(i64 %X, <2 x ptr> %P) nounwind {
; CHECK-LABEL: @test13_vector3(
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[X:%.*]], i64 0
-; CHECK-NEXT: [[TMP1:%.*]] = shl <2 x i64> [[DOTSPLATINSERT]], <i64 2, i64 0>
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i64> [[TMP1]], <i64 4, i64 poison>
+; CHECK-NEXT: [[X:%.*]] = shl nsw i64 [[X1:%.*]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[X]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i64> [[DOTSPLATINSERT]], <i64 4, i64 poison>
; CHECK-NEXT: [[C:%.*]] = shufflevector <2 x i1> [[TMP2]], <2 x i1> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: ret <2 x i1> [[C]]
;
@@ -383,7 +387,8 @@ define ptr @test14(i32 %idx) {
@Array = external global [40 x i32]
define ptr @test15(i64 %X) {
; CHECK-LABEL: @test15(
-; CHECK-NEXT: [[A:%.*]] = getelementptr i32, ptr @Array, i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[X:%.*]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr @Array, i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[A]]
;
%A = getelementptr i32, ptr @Array, i64 %X
@@ -393,7 +398,8 @@ define ptr @test15(i64 %X) {
define ptr @test_index_canon(ptr %X, i32 %Idx) {
; CHECK-LABEL: @test_index_canon(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[IDX:%.*]] to i64
-; CHECK-NEXT: [[R:%.*]] = getelementptr i32, ptr [[X:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[R:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[R]]
;
%R = getelementptr i32, ptr %X, i32 %Idx
@@ -403,7 +409,8 @@ define ptr @test_index_canon(ptr %X, i32 %Idx) {
define ptr @test_index_canon_inbounds(ptr %X, i32 %Idx) {
; CHECK-LABEL: @test_index_canon_inbounds(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[IDX:%.*]] to i64
-; CHECK-NEXT: [[R:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[R:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[R]]
;
%R = getelementptr inbounds i32, ptr %X, i32 %Idx
@@ -413,7 +420,8 @@ define ptr @test_index_canon_inbounds(ptr %X, i32 %Idx) {
define ptr @test_index_canon_nusw_nuw(ptr %X, i32 %Idx) {
; CHECK-LABEL: @test_index_canon_nusw_nuw(
; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[IDX:%.*]] to i64
-; CHECK-NEXT: [[R:%.*]] = getelementptr nusw nuw i32, ptr [[X:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[R:%.*]] = getelementptr nusw nuw i8, ptr [[X:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[R]]
;
%R = getelementptr nusw nuw i32, ptr %X, i32 %Idx
@@ -680,7 +688,7 @@ entry:
define i32 @test28() nounwind {
; CHECK-LABEL: @test28(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ORIENTATIONS:%.*]] = alloca [1 x [1 x %struct.x]], align 8
+; 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: br label [[BB10:%.*]]
; CHECK: bb10:
@@ -760,7 +768,8 @@ define i32 @test30(i32 %m, i32 %n) nounwind {
; CHECK-NEXT: [[TMP1:%.*]] = alloca i32, i64 [[TMP0]], align 4
; CHECK-NEXT: call void @test30f(ptr nonnull [[TMP1]]) #[[ATTR0]]
; CHECK-NEXT: [[TMP2:%.*]] = sext i32 [[M:%.*]] to i64
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP2]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
; CHECK-NEXT: ret i32 [[TMP4]]
;
@@ -917,7 +926,8 @@ define i1 @test37() nounwind {
define ptr @test38(ptr %I, i32 %n) {
; CHECK-LABEL: @test38(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[N:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = getelementptr i32, ptr [[I:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[A]]
;
%A = getelementptr i32, ptr %I, i32 %n
@@ -1137,8 +1147,8 @@ define ptr @test43i(ptr %c1, ptr %c2) {
; CHECK-LABEL: @test43i(
; CHECK-NEXT: [[PTRTOINT:%.*]] = ptrtoint ptr [[C1:%.*]] to i64
; CHECK-NEXT: [[SUB:%.*]] = sub i64 0, [[PTRTOINT]]
-; CHECK-NEXT: [[SHR:%.*]] = ashr i64 [[SUB]], 1
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i16, ptr [[C2:%.*]], i64 [[SHR]]
+; CHECK-NEXT: [[SHR:%.*]] = and i64 [[SUB]], -2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[C2:%.*]], i64 [[SHR]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%ptrtoint = ptrtoint ptr %c1 to i64
@@ -1151,10 +1161,12 @@ define ptr @test43i(ptr %c1, ptr %c2) {
define ptr @test44i(ptr %c1, ptr %c2) {
; CHECK-LABEL: @test44i(
-; CHECK-NEXT: [[PTRTOINT:%.*]] = ptrtoint ptr [[C1:%.*]] to i64
+; CHECK-NEXT: [[C1:%.*]] = freeze ptr [[C3:%.*]]
+; CHECK-NEXT: [[PTRTOINT:%.*]] = ptrtoint ptr [[C1]] to i64
; CHECK-NEXT: [[SUB:%.*]] = sub i64 0, [[PTRTOINT]]
-; CHECK-NEXT: [[SHR:%.*]] = sdiv i64 [[SUB]], 7
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds [[STRUCT_C:%.*]], ptr [[C2:%.*]], i64 [[SHR]]
+; CHECK-NEXT: [[TMP1:%.*]] = srem i64 [[SUB]], 7
+; CHECK-NEXT: [[TMP2:%.*]] = sub nsw i64 [[SUB]], [[TMP1]]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[C2:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%ptrtoint = ptrtoint ptr %c1 to i64
@@ -1170,8 +1182,10 @@ define ptr @test45(ptr %c1, ptr %c2) {
; CHECK-NEXT: [[PTRTOINT1:%.*]] = ptrtoint ptr [[C1:%.*]] to i64
; CHECK-NEXT: [[PTRTOINT2:%.*]] = ptrtoint ptr [[C2:%.*]] to i64
; CHECK-NEXT: [[SUB:%.*]] = sub i64 [[PTRTOINT2]], [[PTRTOINT1]]
-; CHECK-NEXT: [[SHR:%.*]] = sdiv i64 [[SUB]], 7
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds [[STRUCT_C:%.*]], ptr [[C1]], i64 [[SHR]]
+; CHECK-NEXT: [[SUB_FR:%.*]] = freeze i64 [[SUB]]
+; CHECK-NEXT: [[TMP1:%.*]] = srem i64 [[SUB_FR]], 7
+; CHECK-NEXT: [[TMP2:%.*]] = sub nsw i64 [[SUB_FR]], [[TMP1]]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[C1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%ptrtoint1 = ptrtoint ptr %c1 to i64
@@ -1187,7 +1201,8 @@ define ptr @test46(ptr %c1, ptr %c2, i64 %N) {
; CHECK-NEXT: [[PTRTOINT:%.*]] = ptrtoint ptr [[C1:%.*]] to i64
; CHECK-NEXT: [[SUB:%.*]] = sub i64 0, [[PTRTOINT]]
; CHECK-NEXT: [[SDIV:%.*]] = sdiv i64 [[SUB]], [[N:%.*]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds [[STRUCT_C:%.*]], ptr [[C2:%.*]], i64 [[SDIV]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul nsw i64 [[SDIV]], 7
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[C2:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%ptrtoint = ptrtoint ptr %c1 to i64
@@ -1200,7 +1215,11 @@ define ptr @test46(ptr %c1, ptr %c2, i64 %N) {
define ptr @test47(ptr %I, i64 %C, i64 %D) {
; CHECK-LABEL: @test47(
-; CHECK-NEXT: [[B:%.*]] = getelementptr i32, ptr [[I:%.*]], i64 [[D:%.*]]
+; CHECK-NEXT: [[SUB:%.*]] = sub i64 [[D:%.*]], [[C:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[C]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[SUB]], 2
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[B]]
;
%sub = sub i64 %D, %C
@@ -1211,7 +1230,11 @@ define ptr @test47(ptr %I, i64 %C, i64 %D) {
define ptr @test48(ptr %I, i64 %C, i64 %D) {
; CHECK-LABEL: @test48(
-; CHECK-NEXT: [[B:%.*]] = getelementptr i32, ptr [[I:%.*]], i64 [[D:%.*]]
+; CHECK-NEXT: [[SUB:%.*]] = sub i64 [[D:%.*]], [[C:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[SUB]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[C]], 2
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[B]]
;
%sub = sub i64 %D, %C
@@ -1222,7 +1245,11 @@ define ptr @test48(ptr %I, i64 %C, i64 %D) {
define ptr @test49(ptr %I, i64 %C) {
; CHECK-LABEL: @test49(
-; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 -4
+; CHECK-NEXT: [[NOTC:%.*]] = xor i64 [[C:%.*]], -1
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[C]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[I:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[NOTC]], 2
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[B]]
;
%notC = xor i64 -1, %C
@@ -1253,7 +1280,7 @@ define ptr addrspace(1) @ascast_0_0_gep(ptr %p) nounwind {
define <2 x ptr> @PR32414(ptr %ptr) {
; CHECK-LABEL: @PR32414(
-; CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 1>
+; CHECK-NEXT: [[T1:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 4>
; CHECK-NEXT: ret <2 x ptr> [[T1]]
;
%t1 = getelementptr inbounds i32, ptr %ptr, <2 x i64> <i64 0, i64 1>
@@ -1262,7 +1289,8 @@ define <2 x ptr> @PR32414(ptr %ptr) {
define ptr @test_bitcast_nzgep(ptr %base, i64 %idx) {
; CHECK-LABEL: @test_bitcast_nzgep(
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%ptr = getelementptr inbounds i32, ptr %base, i64 %idx
@@ -1271,7 +1299,8 @@ define ptr @test_bitcast_nzgep(ptr %base, i64 %idx) {
define ptr @test_zgep_nzgep(ptr %base, i64 %idx) {
; CHECK-LABEL: @test_zgep_nzgep(
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%ptr = getelementptr inbounds i32, ptr %base, i64 %idx
@@ -1280,7 +1309,8 @@ define ptr @test_zgep_nzgep(ptr %base, i64 %idx) {
define ptr @test_nzgep_zgep(ptr %base, i64 %idx) {
; CHECK-LABEL: @test_nzgep_zgep(
-; CHECK-NEXT: [[BASE2:%.*]] = getelementptr inbounds [1 x i32], ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[BASE2:%.*]] = getelementptr inbounds i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[BASE2]]
;
%base2 = getelementptr inbounds [1 x i32], ptr %base, i64 %idx
@@ -1365,7 +1395,8 @@ define ptr @gep_of_gep_multiuse_const_and_const(ptr %p, i64 %idx) {
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: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: call void @use(ptr [[GEP1]])
; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[GEP1]], i64 4
; CHECK-NEXT: ret ptr [[GEP2]]
@@ -1378,9 +1409,11 @@ define ptr @gep_of_gep_multiuse_var_and_const(ptr %p, i64 %idx) {
define ptr @gep_of_gep_multiuse_var_and_var(ptr %p, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_of_gep_multiuse_var_and_var(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr [4 x i32], ptr [[P:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 4
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: call void @use(ptr [[GEP1]])
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[GEP1]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[GEP1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP2]]
;
%gep1 = getelementptr [4 x i32], ptr %p, i64 %idx
@@ -1568,7 +1601,8 @@ define ptr @gep_i8(ptr %p, i64 %off) {
define ptr @gep_sdiv_mismatched_size(ptr %p, i64 %off) {
; CHECK-LABEL: @gep_sdiv_mismatched_size(
; CHECK-NEXT: [[INDEX:%.*]] = sdiv exact i64 [[OFF:%.*]], 20
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul nsw i64 [[INDEX]], 7
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%index = sdiv exact i64 %off, 20
@@ -1579,7 +1613,8 @@ define ptr @gep_sdiv_mismatched_size(ptr %p, i64 %off) {
define ptr @gep_udiv_mismatched_size(ptr %p, i64 %off) {
; CHECK-LABEL: @gep_udiv_mismatched_size(
; CHECK-NEXT: [[INDEX:%.*]] = udiv exact i64 [[OFF:%.*]], 20
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = mul nuw nsw i64 [[INDEX]], 7
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%index = udiv exact i64 %off, 20
@@ -1589,8 +1624,10 @@ define ptr @gep_udiv_mismatched_size(ptr %p, i64 %off) {
define ptr @gep_sdiv_without_exact(ptr %p, i64 %off) {
; CHECK-LABEL: @gep_sdiv_without_exact(
-; CHECK-NEXT: [[INDEX:%.*]] = sdiv i64 [[OFF:%.*]], 7
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[OFF_FR:%.*]] = freeze i64 [[OFF:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = srem i64 [[OFF_FR]], 7
+; CHECK-NEXT: [[TMP2:%.*]] = sub nsw i64 [[OFF_FR]], [[TMP1]]
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%index = sdiv i64 %off, 7
@@ -1600,8 +1637,10 @@ define ptr @gep_sdiv_without_exact(ptr %p, i64 %off) {
define ptr @gep_udiv_without_exact(ptr %p, i64 %off) {
; CHECK-LABEL: @gep_udiv_without_exact(
-; CHECK-NEXT: [[INDEX:%.*]] = udiv i64 [[OFF:%.*]], 7
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr [[STRUCT_C:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[OFF_FR:%.*]] = freeze i64 [[OFF:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = urem i64 [[OFF_FR]], 7
+; CHECK-NEXT: [[TMP2:%.*]] = sub nuw i64 [[OFF_FR]], [[TMP1]]
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%index = udiv i64 %off, 7
@@ -1611,8 +1650,8 @@ define ptr @gep_udiv_without_exact(ptr %p, i64 %off) {
define ptr @gep_ashr_without_exact(ptr %p, i64 %off) {
; CHECK-LABEL: @gep_ashr_without_exact(
-; CHECK-NEXT: [[INDEX:%.*]] = ashr i64 [[OFF:%.*]], 2
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[INDEX:%.*]] = and i64 [[OFF:%.*]], -4
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%index = ashr i64 %off, 2
@@ -1622,8 +1661,8 @@ define ptr @gep_ashr_without_exact(ptr %p, i64 %off) {
define ptr @gep_lshr_without_exact(ptr %p, i64 %off) {
; CHECK-LABEL: @gep_lshr_without_exact(
-; CHECK-NEXT: [[INDEX:%.*]] = lshr i64 [[OFF:%.*]], 2
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[INDEX:%.*]] = and i64 [[OFF:%.*]], -4
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[INDEX]]
; CHECK-NEXT: ret ptr [[PTR]]
;
%index = lshr i64 %off, 2
@@ -1762,7 +1801,8 @@ define ptr @gep_of_udiv(ptr %p, i64 %x) {
define ptr @gep_of_udiv_fail_not_divisible(ptr %p, i64 %x) {
; CHECK-LABEL: @gep_of_udiv_fail_not_divisible(
; CHECK-NEXT: [[IDX:%.*]] = udiv exact i64 [[X:%.*]], 13
-; CHECK-NEXT: [[R:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[R:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[R]]
;
%idx = udiv exact i64 %x, 13
@@ -1786,7 +1826,8 @@ define ptr @gep_of_sdiv(ptr %p, i64 %x) {
define ptr @gep_of_sdiv_fail_not_divisible(ptr %p, i64 %x) {
; CHECK-LABEL: @gep_of_sdiv_fail_not_divisible(
; CHECK-NEXT: [[IDX:%.*]] = sdiv exact i64 [[X:%.*]], -35
-; CHECK-NEXT: [[R:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX]], 2
+; CHECK-NEXT: [[R:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[R]]
;
%idx = sdiv exact i64 %x, -35
@@ -1797,7 +1838,8 @@ define ptr @gep_of_sdiv_fail_not_divisible(ptr %p, i64 %x) {
define ptr @gep_of_sdiv_fail_ub(ptr %p, i64 %x) {
; CHECK-LABEL: @gep_of_sdiv_fail_ub(
; CHECK-NEXT: [[IDX:%.*]] = sdiv i64 [[X:%.*]], -4
-; CHECK-NEXT: [[R:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX]], 2
+; CHECK-NEXT: [[R:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[R]]
;
%idx = sdiv i64 %x, -4
@@ -1831,7 +1873,8 @@ define ptr @gep_of_ashr_fail_multiuse(ptr %p, i64 %x) {
; CHECK-LABEL: @gep_of_ashr_fail_multiuse(
; CHECK-NEXT: [[IDX:%.*]] = ashr exact i64 [[X:%.*]], 3
; CHECK-NEXT: call void @use.i64(i64 [[IDX]])
-; CHECK-NEXT: [[R:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = ashr exact i64 [[X]], 1
+; CHECK-NEXT: [[R:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[R]]
;
%idx = ashr exact i64 %x, 3
@@ -1842,8 +1885,9 @@ define ptr @gep_of_ashr_fail_multiuse(ptr %p, i64 %x) {
define ptr @gep_of_lshr_fail_missing_exact(ptr %p, i64 %x) {
; CHECK-LABEL: @gep_of_lshr_fail_missing_exact(
-; CHECK-NEXT: [[IDX:%.*]] = lshr i64 [[X:%.*]], 3
-; CHECK-NEXT: [[R:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = lshr i64 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP2:%.*]] = and i64 [[TMP1]], 9223372036854775804
+; CHECK-NEXT: [[R:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[R]]
;
%idx = lshr i64 %x, 3
@@ -1853,8 +1897,8 @@ define ptr @gep_of_lshr_fail_missing_exact(ptr %p, i64 %x) {
define ptr @gep_of_ashr_fail_not_divisible(ptr %p, i64 %x) {
; CHECK-LABEL: @gep_of_ashr_fail_not_divisible(
-; CHECK-NEXT: [[IDX:%.*]] = ashr exact i64 [[X:%.*]], 1
-; CHECK-NEXT: [[R:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[X:%.*]], 1
+; CHECK-NEXT: [[R:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[R]]
;
%idx = ashr exact i64 %x, 1
@@ -1922,8 +1966,10 @@ define ptr @gep_merge_nusw(ptr %p, i64 %x, i64 %y) {
define ptr @gep_merge_nuw_add_zero(ptr %p, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_merge_nuw_add_zero(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr nuw [2 x i32], ptr [[GEP_SPLIT:%.*]], i64 [[IDX2:%.*]]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i32, ptr [[GEP]], i64 [[IDX3:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i8, ptr [[GEP2]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP1]]
;
%gep1 = getelementptr nuw [2 x i32], ptr %p, i64 %idx
@@ -1936,8 +1982,10 @@ define ptr @gep_merge_nuw_add_zero(ptr %p, i64 %idx, i64 %idx2) {
; after the merge.
define ptr @gep_merge_nusw_add_zero(ptr %p, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_merge_nusw_add_zero(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nusw [2 x i32], ptr [[P:%.*]], i64 [[IDX:%.*]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr nusw i32, ptr [[GEP1]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nusw i8, ptr [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr nusw i8, ptr [[GEP1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[GEP]]
;
%gep1 = getelementptr nusw [2 x i32], ptr %p, i64 %idx
@@ -1947,7 +1995,8 @@ define ptr @gep_merge_nusw_add_zero(ptr %p, i64 %idx, i64 %idx2) {
define ptr @gep_merge_nuw_const(ptr %p, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_merge_nuw_const(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw [2 x i32], ptr [[P:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr nuw i8, ptr [[GEP1]], i64 4
; CHECK-NEXT: ret ptr [[GEP]]
;
@@ -1958,7 +2007,8 @@ define ptr @gep_merge_nuw_const(ptr %p, i64 %idx, i64 %idx2) {
define ptr @gep_merge_nuw_const_neg(ptr %p, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_merge_nuw_const_neg(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw [2 x i32], ptr [[P:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr nuw i8, ptr [[GEP1]], i64 -4
; CHECK-NEXT: ret ptr [[GEP]]
;
@@ -1971,7 +2021,8 @@ define ptr @gep_merge_nuw_const_neg(ptr %p, i64 %idx, i64 %idx2) {
; does not overflow.
define ptr @gep_merge_nusw_const(ptr %p, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_merge_nusw_const(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nusw [2 x i32], ptr [[P:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nusw i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr nusw nuw i8, ptr [[GEP1]], i64 4
; CHECK-NEXT: ret ptr [[GEP]]
;
diff --git a/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll b/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll
index f873551512718..a36ecfb4dca3a 100644
--- a/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-custom-dl.ll
@@ -116,7 +116,8 @@ define i1 @test60_addrspacecast_larger(ptr addrspace(1) %foo, i32 %i, i16 %j) {
define i1 @test61(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test61(
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[I:%.*]] to i32
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[FOO:%.*]], i32 [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl i32 [[TMP1]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[FOO:%.*]], i32 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = trunc i64 [[J:%.*]] to i32
; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[FOO]], i32 [[TMP2]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[GEP1]], [[GEP2]]
@@ -131,9 +132,10 @@ define i1 @test61(ptr %foo, i64 %i, i64 %j) {
define i1 @test61_as1(ptr addrspace(1) %foo, i16 %i, i16 %j) {
; CHECK-LABEL: @test61_as1(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr addrspace(1) [[FOO:%.*]], i16 [[I:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr addrspace(1) [[FOO]], i16 [[J:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr addrspace(1) [[GEP1]], [[GEP2]]
+; CHECK-NEXT: [[J:%.*]] = shl i16 [[I:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr addrspace(1) [[FOO:%.*]], i16 [[J]]
+; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr addrspace(1) [[FOO]], i16 [[J1:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr addrspace(1) [[GEP2]], [[GEP3]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep1 = getelementptr i32, ptr addrspace(1) %foo, i16 %i
diff --git a/llvm/test/Transforms/InstCombine/icmp-gep.ll b/llvm/test/Transforms/InstCombine/icmp-gep.ll
index 10d7b6df8eaf3..807cefda5c2a1 100644
--- a/llvm/test/Transforms/InstCombine/icmp-gep.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-gep.ll
@@ -410,7 +410,8 @@ define i1 @test60_nusw_nuw(ptr %foo, i64 %i, i64 %j) {
define i1 @test60_nusw_nuw_mix(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test60_nusw_nuw_mix(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i32, ptr [[FOO:%.*]], i64 [[I:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[I:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i8, ptr [[FOO:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nusw i8, ptr [[FOO]], i64 [[J:%.*]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[GEP1]], [[GEP2]]
; CHECK-NEXT: ret i1 [[CMP]]
@@ -423,9 +424,10 @@ define i1 @test60_nusw_nuw_mix(ptr %foo, i64 %i, i64 %j) {
define i1 @test_gep_ult_no_inbounds(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test_gep_ult_no_inbounds(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[FOO:%.*]], i64 [[I:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[FOO]], i64 [[J:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[GEP1]], [[GEP2]]
+; CHECK-NEXT: [[J:%.*]] = shl i64 [[I:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[FOO:%.*]], i64 [[J]]
+; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[FOO]], i64 [[J1:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[GEP2]], [[GEP3]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep1 = getelementptr i32, ptr %foo, i64 %i
@@ -508,9 +510,10 @@ define i1 @test60_addrspacecast_larger(ptr addrspace(1) %foo, i32 %i, i16 %j) {
define i1 @test61(ptr %foo, i64 %i, i64 %j) {
; CHECK-LABEL: @test61(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr [[FOO:%.*]], i64 [[I:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[FOO]], i64 [[J:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[GEP1]], [[GEP2]]
+; CHECK-NEXT: [[J:%.*]] = shl i64 [[I:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[FOO:%.*]], i64 [[J]]
+; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[FOO]], i64 [[J1:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[GEP2]], [[GEP3]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep1 = getelementptr i32, ptr %foo, i64 %i
@@ -522,9 +525,10 @@ define i1 @test61(ptr %foo, i64 %i, i64 %j) {
define i1 @test61_as1(ptr addrspace(1) %foo, i16 %i, i16 %j) {
; CHECK-LABEL: @test61_as1(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i32, ptr addrspace(1) [[FOO:%.*]], i16 [[I:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr addrspace(1) [[FOO]], i16 [[J:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr addrspace(1) [[GEP1]], [[GEP2]]
+; CHECK-NEXT: [[J:%.*]] = shl i16 [[I:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr addrspace(1) [[FOO:%.*]], i16 [[J]]
+; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr addrspace(1) [[FOO]], i16 [[J1:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr addrspace(1) [[GEP2]], [[GEP3]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep1 = getelementptr i32, ptr addrspace(1) %foo, i16 %i
@@ -813,8 +817,8 @@ define i1 @gep_multiple_eq_commuted(ptr %base, i64 %idx, i64 %idx2) {
define i1 @gep_mugtiple_ugt_nuw(ptr %base, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_mugtiple_ugt_nuw(
-; CHECK-NEXT: [[GEP1_IDX1:%.*]] = sub i64 0, [[IDX2:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = icmp ne i64 [[IDX:%.*]], [[GEP1_IDX1]]
+; CHECK-NEXT: [[TMP1:%.*]] = sub i64 0, [[IDX2:%.*]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ne i64 [[IDX:%.*]], [[TMP1]]
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep1 = getelementptr nuw i32, ptr %base, i64 %idx
@@ -825,8 +829,10 @@ define i1 @gep_mugtiple_ugt_nuw(ptr %base, i64 %idx, i64 %idx2) {
define i1 @gep_mugtiple_ugt_not_all_nuw(ptr %base, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_mugtiple_ugt_not_all_nuw(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[GEP1]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i8, ptr [[BASE:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[GEP1]], i64 [[TMP2]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt ptr [[GEP2]], [[BASE]]
; CHECK-NEXT: ret i1 [[CMP]]
;
@@ -838,8 +844,10 @@ define i1 @gep_mugtiple_ugt_not_all_nuw(ptr %base, i64 %idx, i64 %idx2) {
define i1 @gep_mugtiple_ugt_inbounds_nusw(ptr %base, i64 %idx, i64 %idx2) {
; CHECK-LABEL: @gep_mugtiple_ugt_inbounds_nusw(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nusw i32, ptr [[GEP1]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[BASE:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nusw i8, ptr [[GEP1]], i64 [[TMP2]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt ptr [[GEP2]], [[BASE]]
; CHECK-NEXT: ret i1 [[CMP]]
;
@@ -935,11 +943,13 @@ define i1 @gep_multiple_multi_above_below_limit_consts(ptr %base, i64 %idx1, i64
; CHECK-LABEL: @gep_multiple_multi_above_below_limit_consts(
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 16
; CHECK-NEXT: call void @use(ptr [[GEP1]])
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i32, ptr [[GEP1]], i64 [[IDX1:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX1:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[GEP1]], i64 [[TMP1]]
; CHECK-NEXT: call void @use(ptr [[GEP2]])
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[GEP2]], i64 16
; CHECK-NEXT: call void @use(ptr [[GEP3]])
-; CHECK-NEXT: [[GEP4:%.*]] = getelementptr i32, ptr [[GEP3]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP4:%.*]] = getelementptr i8, ptr [[GEP3]], i64 [[TMP2]]
; CHECK-NEXT: call void @use(ptr [[GEP4]])
; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[GEP4]], [[BASE]]
; CHECK-NEXT: ret i1 [[CMP]]
@@ -958,13 +968,17 @@ define i1 @gep_multiple_multi_above_below_limit_consts(ptr %base, i64 %idx1, i64
define i1 @gep_multiple_multi_use_above_limit(ptr %base, i64 %idx1, i64 %idx2, i64 %idx3, i64 %idx4) {
; CHECK-LABEL: @gep_multiple_multi_use_above_limit(
-; CHECK-NEXT: [[GEP4:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 [[IDX1:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX1:%.*]], 2
+; CHECK-NEXT: [[GEP4:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: call void @use(ptr [[GEP4]])
-; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i32, ptr [[GEP4]], i64 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i8, ptr [[GEP4]], i64 [[TMP2]]
; CHECK-NEXT: call void @use(ptr [[GEP3]])
-; CHECK-NEXT: [[GEP5:%.*]] = getelementptr i32, ptr [[GEP3]], i64 [[IDX3:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl i64 [[IDX3:%.*]], 2
+; CHECK-NEXT: [[GEP5:%.*]] = getelementptr i8, ptr [[GEP3]], i64 [[TMP3]]
; CHECK-NEXT: call void @use(ptr [[GEP5]])
-; CHECK-NEXT: [[GEP6:%.*]] = getelementptr i32, ptr [[GEP5]], i64 [[IDX4:%.*]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[IDX4:%.*]], 2
+; CHECK-NEXT: [[GEP6:%.*]] = getelementptr i8, ptr [[GEP5]], i64 [[TMP4]]
; CHECK-NEXT: call void @use(ptr [[GEP6]])
; CHECK-NEXT: [[CMP:%.*]] = icmp eq ptr [[GEP6]], [[BASE]]
; CHECK-NEXT: ret i1 [[CMP]]
@@ -1087,10 +1101,14 @@ define i1 @gep_gep_multiple_ult_nuw(ptr %base, i64 %idx1, i64 %idx2, i64 %idx3,
define i1 @gep_gep_multiple_ult_missing_nuw(ptr %base, i64 %idx1, i64 %idx2, i64 %idx3, i64 %idx4) {
; CHECK-LABEL: @gep_gep_multiple_ult_missing_nuw(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i32, ptr [[BASE:%.*]], i64 [[IDX1:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i32, ptr [[GEP1]], i64 [[IDX2:%.*]]
-; CHECK-NEXT: [[GEP3:%.*]] = getelementptr nuw i32, ptr [[BASE]], i64 [[IDX3:%.*]]
-; CHECK-NEXT: [[GEP4:%.*]] = getelementptr i32, ptr [[GEP3]], i64 [[IDX4:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[IDX1:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nuw i8, ptr [[BASE:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw i64 [[IDX2:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr nuw i8, ptr [[GEP1]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw i64 [[IDX3:%.*]], 2
+; CHECK-NEXT: [[GEP3:%.*]] = getelementptr nuw i8, ptr [[BASE]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[IDX4:%.*]], 2
+; CHECK-NEXT: [[GEP4:%.*]] = getelementptr i8, ptr [[GEP3]], i64 [[TMP4]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ult ptr [[GEP2]], [[GEP4]]
; CHECK-NEXT: ret i1 [[CMP]]
;
diff --git a/llvm/test/Transforms/InstCombine/icmp-trunc.ll b/llvm/test/Transforms/InstCombine/icmp-trunc.ll
index 431fc2b4f062c..e75276f65eadb 100644
--- a/llvm/test/Transforms/InstCombine/icmp-trunc.ll
+++ b/llvm/test/Transforms/InstCombine/icmp-trunc.ll
@@ -59,7 +59,13 @@ define i1 @ult_2_use(i32 %x) {
define i1 @PR52260(i32 %x) {
; CHECK-LABEL: @PR52260(
-; CHECK-NEXT: ret i1 true
+; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[X:%.*]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[IDX:%.*]] = getelementptr inbounds i8, ptr @a, i64 [[TMP1]]
+; CHECK-NEXT: [[T1:%.*]] = load i32, ptr [[IDX]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = and i32 [[T1]], 254
+; CHECK-NEXT: [[TOBOOL:%.*]] = icmp eq i32 [[TMP2]], 0
+; CHECK-NEXT: ret i1 [[TOBOOL]]
;
%idxprom = sext i32 %x to i64
%idx = getelementptr inbounds [3 x i32], ptr @a, i64 0, i64 %idxprom
diff --git a/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll b/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll
index 07486ff21d54d..dca52ec10d380 100644
--- a/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll
+++ b/llvm/test/Transforms/InstCombine/indexed-gep-compares.ll
@@ -131,7 +131,8 @@ bb2:
define ptr @test3_no_inbounds1(ptr %A, i32 %Offset) {
; CHECK-LABEL: @test3_no_inbounds1(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP:%.*]] = getelementptr i32, ptr [[A:%.*]], i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i32 [[OFFSET:%.*]], 2
+; CHECK-NEXT: [[TMP:%.*]] = getelementptr i8, ptr [[A:%.*]], i32 [[TMP0]]
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb:
; CHECK-NEXT: [[RHS:%.*]] = phi ptr [ [[RHS_NEXT:%.*]], [[BB]] ], [ [[TMP]], [[ENTRY:%.*]] ]
@@ -160,7 +161,8 @@ bb2:
define ptr @test3_no_inbounds2(ptr %A, i32 %Offset) {
; CHECK-LABEL: @test3_no_inbounds2(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[OFFSET:%.*]], 2
+; CHECK-NEXT: [[TMP:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP0]]
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb:
; CHECK-NEXT: [[RHS:%.*]] = phi ptr [ [[RHS_NEXT:%.*]], [[BB]] ], [ [[TMP]], [[ENTRY:%.*]] ]
@@ -189,7 +191,8 @@ bb2:
define ptr @test3_no_inbounds3(ptr %A, i32 %Offset) {
; CHECK-LABEL: @test3_no_inbounds3(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[OFFSET:%.*]], 2
+; CHECK-NEXT: [[TMP:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP0]]
; CHECK-NEXT: br label [[BB:%.*]]
; CHECK: bb:
; CHECK-NEXT: [[RHS:%.*]] = phi ptr [ [[RHS_NEXT:%.*]], [[BB]] ], [ [[TMP]], [[ENTRY:%.*]] ]
diff --git a/llvm/test/Transforms/InstCombine/known-phi-recurse.ll b/llvm/test/Transforms/InstCombine/known-phi-recurse.ll
index ac44e6c9df412..1a53528220865 100644
--- a/llvm/test/Transforms/InstCombine/known-phi-recurse.ll
+++ b/llvm/test/Transforms/InstCombine/known-phi-recurse.ll
@@ -158,7 +158,8 @@ define i32 @knownbits_phi_select_test1(ptr %p1, ptr %p2, i8 %x) {
; CHECK-NEXT: ]
; CHECK: default:
; CHECK-NEXT: [[EXT:%.*]] = sext i8 [[INDVAR1]] to i64
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i16, ptr [[P2:%.*]], i64 [[EXT]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[EXT]], 1
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[P2:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[LOAD1:%.*]] = load i16, ptr [[GEP1]], align 2
; CHECK-NEXT: [[MASK:%.*]] = and i16 [[LOAD1]], 8192
; CHECK-NEXT: [[CMP1:%.*]] = icmp eq i16 [[MASK]], 0
diff --git a/llvm/test/Transforms/InstCombine/load-bitcast-select.ll b/llvm/test/Transforms/InstCombine/load-bitcast-select.ll
index 3c43618b13569..7f1cf7529c9e8 100644
--- a/llvm/test/Transforms/InstCombine/load-bitcast-select.ll
+++ b/llvm/test/Transforms/InstCombine/load-bitcast-select.ll
@@ -16,8 +16,10 @@ define void @_Z3foov() {
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[TMP0:%.*]] = zext nneg i32 [[I_0]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr @a, i64 [[TMP0]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw float, ptr @b, i64 [[TMP0]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr @a, i64 [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr @b, i64 [[TMP4]]
; CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[CMP_I:%.*]] = fcmp fast olt float [[TMP1]], [[TMP2]]
diff --git a/llvm/test/Transforms/InstCombine/load-cmp.ll b/llvm/test/Transforms/InstCombine/load-cmp.ll
index a5a0681ac3af3..05b4b9d02325d 100644
--- a/llvm/test/Transforms/InstCombine/load-cmp.ll
+++ b/llvm/test/Transforms/InstCombine/load-cmp.ll
@@ -25,7 +25,10 @@
define i1 @test1(i32 %X) {
; CHECK-LABEL: @test1(
-; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[X:%.*]], 9
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @G16, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp eq i16 [[Q]], 0
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [10 x i16], ptr @G16, i32 0, i32 %X
@@ -36,8 +39,10 @@ define i1 @test1(i32 %X) {
define i1 @test1_noinbounds(i32 %X) {
; CHECK-LABEL: @test1_noinbounds(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[X:%.*]], 2147483647
-; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[TMP1]], 9
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[X:%.*]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr @G16, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp eq i16 [[Q]], 0
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr [10 x i16], ptr @G16, i32 0, i32 %X
@@ -48,8 +53,11 @@ define i1 @test1_noinbounds(i32 %X) {
define i1 @test1_noinbounds_i64(i64 %X) {
; CHECK-LABEL: @test1_noinbounds_i64(
-; CHECK-NEXT: [[TMP1:%.*]] = and i64 [[X:%.*]], 2147483647
-; CHECK-NEXT: [[R:%.*]] = icmp eq i64 [[TMP1]], 9
+; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl i32 [[TMP1]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr @G16, i32 [[TMP2]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp eq i16 [[Q]], 0
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr [10 x i16], ptr @G16, i64 0, i64 %X
@@ -60,8 +68,11 @@ define i1 @test1_noinbounds_i64(i64 %X) {
define i1 @test1_noinbounds_as1(i32 %x) {
; CHECK-LABEL: @test1_noinbounds_as1(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[X:%.*]], 32767
-; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[TMP1]], 9
+; CHECK-NEXT: [[TMP1:%.*]] = trunc i32 [[X:%.*]] to i16
+; CHECK-NEXT: [[TMP2:%.*]] = shl i16 [[TMP1]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr addrspace(1) @G16_as1, i16 [[TMP2]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr addrspace(1) [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp eq i16 [[Q]], 0
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr [10 x i16], ptr addrspace(1) @G16_as1, i16 0, i32 %x
@@ -72,8 +83,11 @@ define i1 @test1_noinbounds_as1(i32 %x) {
define i1 @test1_noinbounds_as2(i64 %x) {
; CHECK-LABEL: @test1_noinbounds_as2(
-; CHECK-NEXT: [[TMP1:%.*]] = and i64 [[X:%.*]], 2147483647
-; CHECK-NEXT: [[R:%.*]] = icmp eq i64 [[TMP1]], 9
+; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl i32 [[TMP1]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr addrspace(2) @G16_as2, i32 [[TMP2]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr addrspace(2) [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp eq i16 [[Q]], 0
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr [10 x i16], ptr addrspace(2) @G16_as2, i16 0, i64 %x
@@ -84,7 +98,10 @@ define i1 @test1_noinbounds_as2(i64 %x) {
define i1 @test1_noarrayty(i32 %X) {
; CHECK-LABEL: @test1_noarrayty(
-; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[X:%.*]], 9
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @G16, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp eq i16 [[Q]], 0
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds i16, ptr @G16, i32 %X
@@ -95,7 +112,10 @@ define i1 @test1_noarrayty(i32 %X) {
define i1 @test2(i32 %X) {
; CHECK-LABEL: @test2(
-; CHECK-NEXT: [[R:%.*]] = icmp ne i32 [[X:%.*]], 4
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @G16, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp slt i16 [[Q]], 85
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [10 x i16], ptr @G16, i32 0, i32 %X
@@ -106,7 +126,10 @@ define i1 @test2(i32 %X) {
define i1 @test3(i32 %X) {
; CHECK-LABEL: @test3(
-; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 3
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @GD, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load double, ptr [[P]], align 8
+; CHECK-NEXT: [[R:%.*]] = fcmp oeq double [[Q]], 1.000000e+00
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [6 x double], ptr @GD, i32 0, i32 %X
@@ -117,7 +140,10 @@ define i1 @test3(i32 %X) {
define i1 @test3_noarrayty(i32 %X) {
; CHECK-LABEL: @test3_noarrayty(
-; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 3
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @GD, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load double, ptr [[P]], align 8
+; CHECK-NEXT: [[R:%.*]] = fcmp oeq double [[Q]], 1.000000e+00
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds double, ptr @GD, i32 %X
@@ -128,8 +154,10 @@ define i1 @test3_noarrayty(i32 %X) {
define i1 @test4(i32 %X) {
; CHECK-LABEL: @test4(
-; CHECK-NEXT: [[TMP1:%.*]] = lshr i32 933, [[X:%.*]]
-; CHECK-NEXT: [[R:%.*]] = trunc i32 [[TMP1]] to i1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @G16, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp slt i16 [[Q]], 74
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [10 x i16], ptr @G16, i32 0, i32 %X
@@ -140,9 +168,11 @@ define i1 @test4(i32 %X) {
define i1 @test4_i16(i16 %X) {
; CHECK-LABEL: @test4_i16(
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i16 [[X:%.*]] to i32
-; CHECK-NEXT: [[TMP2:%.*]] = lshr i32 933, [[TMP1]]
-; CHECK-NEXT: [[R:%.*]] = trunc i32 [[TMP2]] to i1
+; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @G16, i32 [[TMP2]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp slt i16 [[Q]], 74
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [10 x i16], ptr @G16, i32 0, i16 %X
@@ -153,9 +183,10 @@ define i1 @test4_i16(i16 %X) {
define i1 @test5(i32 %X) {
; CHECK-LABEL: @test5(
-; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i32 [[X:%.*]], 2
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i32 [[X]], 7
-; CHECK-NEXT: [[R:%.*]] = or i1 [[TMP1]], [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @G16, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = icmp eq i16 [[Q]], 69
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [10 x i16], ptr @G16, i32 0, i32 %X
@@ -166,8 +197,10 @@ define i1 @test5(i32 %X) {
define i1 @test6(i32 %X) {
; CHECK-LABEL: @test6(
-; CHECK-NEXT: [[TMP1:%.*]] = add i32 [[X:%.*]], -1
-; CHECK-NEXT: [[R:%.*]] = icmp ult i32 [[TMP1]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 3
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @GD, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load double, ptr [[P]], align 8
+; CHECK-NEXT: [[R:%.*]] = fcmp ogt double [[Q]], 0.000000e+00
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [6 x double], ptr @GD, i32 0, i32 %X
@@ -178,8 +211,10 @@ define i1 @test6(i32 %X) {
define i1 @test7(i32 %X) {
; CHECK-LABEL: @test7(
-; CHECK-NEXT: [[TMP1:%.*]] = add i32 [[X:%.*]], -4
-; CHECK-NEXT: [[R:%.*]] = icmp ult i32 [[TMP1]], -3
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 3
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @GD, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load double, ptr [[P]], align 8
+; CHECK-NEXT: [[R:%.*]] = fcmp olt double [[Q]], 0.000000e+00
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [6 x double], ptr @GD, i32 0, i32 %X
@@ -190,8 +225,11 @@ define i1 @test7(i32 %X) {
define i1 @test8(i32 %X) {
; CHECK-LABEL: @test8(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[X:%.*]], -2
-; CHECK-NEXT: [[S:%.*]] = icmp eq i32 [[TMP1]], 8
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 1
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @G16, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i16, ptr [[P]], align 2
+; CHECK-NEXT: [[R:%.*]] = and i16 [[Q]], 3
+; CHECK-NEXT: [[S:%.*]] = icmp eq i16 [[R]], 0
; CHECK-NEXT: ret i1 [[S]]
;
%P = getelementptr inbounds [10 x i16], ptr @G16, i32 0, i32 %X
@@ -210,8 +248,11 @@ define i1 @test8(i32 %X) {
define i1 @test9(i32 %X) {
; CHECK-LABEL: @test9(
-; CHECK-NEXT: [[TMP1:%.*]] = add i32 [[X:%.*]], -1
-; CHECK-NEXT: [[R:%.*]] = icmp ult i32 [[TMP1]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 3
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr inbounds i8, ptr @GA, i32 [[TMP1]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds nuw i8, ptr [[P_SPLIT]], i32 4
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 1
; CHECK-NEXT: ret i1 [[R]]
;
%P = getelementptr inbounds [4 x { i32, i32 } ], ptr @GA, i32 0, i32 %X, i32 1
@@ -222,7 +263,11 @@ define i1 @test9(i32 %X) {
define i1 @test10_struct(i32 %x) {
; CHECK-LABEL: @test10_struct(
-; CHECK-NEXT: ret i1 false
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 4
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @GS, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
+; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr inbounds %Foo, ptr @GS, i32 %x, i32 0
%q = load i32, ptr %p
@@ -232,7 +277,11 @@ define i1 @test10_struct(i32 %x) {
define i1 @test10_struct_noinbounds(i32 %x) {
; CHECK-LABEL: @test10_struct_noinbounds(
-; CHECK-NEXT: ret i1 false
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[X:%.*]], 4
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr @GS, i32 [[TMP1]]
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
+; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr %Foo, ptr @GS, i32 %x, i32 0
%q = load i32, ptr %p
@@ -244,7 +293,12 @@ define i1 @test10_struct_noinbounds(i32 %x) {
; Index < ptr size
define i1 @test10_struct_i16(i16 %x){
; CHECK-LABEL: @test10_struct_i16(
-; CHECK-NEXT: ret i1 false
+; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 4
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @GS, i32 [[TMP2]]
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 0
+; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr inbounds %Foo, ptr @GS, i16 %x, i32 0
%q = load i32, ptr %p
@@ -256,7 +310,12 @@ define i1 @test10_struct_i16(i16 %x){
; Index > ptr size
define i1 @test10_struct_i64(i64 %x){
; CHECK-LABEL: @test10_struct_i64(
-; CHECK-NEXT: ret i1 false
+; CHECK-NEXT: [[TMP1:%.*]] = trunc nsw i64 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 4
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds i8, ptr @GS, i32 [[TMP2]]
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 0
+; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr inbounds %Foo, ptr @GS, i64 %x, i32 0
%q = load i32, ptr %p
@@ -266,7 +325,12 @@ define i1 @test10_struct_i64(i64 %x){
define i1 @test10_struct_noinbounds_i16(i16 %x) {
; CHECK-LABEL: @test10_struct_noinbounds_i16(
-; CHECK-NEXT: ret i1 false
+; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 4
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr @GS, i32 [[TMP2]]
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 0
+; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr %Foo, ptr @GS, i16 %x, i32 0
%q = load i32, ptr %p
@@ -276,7 +340,11 @@ define i1 @test10_struct_noinbounds_i16(i16 %x) {
define i1 @test10_struct_arr(i32 %x) {
; CHECK-LABEL: @test10_struct_arr(
-; CHECK-NEXT: [[R:%.*]] = icmp ne i32 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 4
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr inbounds i8, ptr @GStructArr, i32 [[TMP1]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds nuw i8, ptr [[P_SPLIT]], i32 8
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr inbounds [4 x %Foo], ptr @GStructArr, i32 0, i32 %x, i32 2
@@ -287,8 +355,11 @@ define i1 @test10_struct_arr(i32 %x) {
define i1 @test10_struct_arr_noinbounds(i32 %x) {
; CHECK-LABEL: @test10_struct_arr_noinbounds(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[X:%.*]], 268435455
-; CHECK-NEXT: [[R:%.*]] = icmp ne i32 [[TMP1]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[X:%.*]], 4
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr i8, ptr @GStructArr, i32 [[TMP1]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr [[P_SPLIT]], i32 8
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr [4 x %Foo], ptr @GStructArr, i32 0, i32 %x, i32 2
@@ -299,7 +370,12 @@ define i1 @test10_struct_arr_noinbounds(i32 %x) {
define i1 @test10_struct_arr_i16(i16 %x) {
; CHECK-LABEL: @test10_struct_arr_i16(
-; CHECK-NEXT: [[R:%.*]] = icmp ne i16 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 4
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr inbounds i8, ptr @GStructArr, i32 [[TMP2]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds nuw i8, ptr [[P_SPLIT]], i32 8
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr inbounds [4 x %Foo], ptr @GStructArr, i16 0, i16 %x, i32 2
@@ -310,7 +386,12 @@ define i1 @test10_struct_arr_i16(i16 %x) {
define i1 @test10_struct_arr_i64(i64 %x) {
; CHECK-LABEL: @test10_struct_arr_i64(
-; CHECK-NEXT: [[R:%.*]] = icmp ne i64 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = trunc nsw i64 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 4
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr inbounds i8, ptr @GStructArr, i32 [[TMP2]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds nuw i8, ptr [[P_SPLIT]], i32 8
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr inbounds [4 x %Foo], ptr @GStructArr, i64 0, i64 %x, i32 2
@@ -321,7 +402,12 @@ define i1 @test10_struct_arr_i64(i64 %x) {
define i1 @test10_struct_arr_noinbounds_i16(i16 %x) {
; CHECK-LABEL: @test10_struct_arr_noinbounds_i16(
-; CHECK-NEXT: [[R:%.*]] = icmp ne i16 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 4
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr i8, ptr @GStructArr, i32 [[TMP2]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr [[P_SPLIT]], i32 8
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr [4 x %Foo], ptr @GStructArr, i32 0, i16 %x, i32 2
@@ -332,8 +418,12 @@ define i1 @test10_struct_arr_noinbounds_i16(i16 %x) {
define i1 @test10_struct_arr_noinbounds_i64(i64 %x) {
; CHECK-LABEL: @test10_struct_arr_noinbounds_i64(
-; CHECK-NEXT: [[TMP1:%.*]] = and i64 [[X:%.*]], 268435455
-; CHECK-NEXT: [[R:%.*]] = icmp ne i64 [[TMP1]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[X:%.*]] to i32
+; CHECK-NEXT: [[TMP2:%.*]] = shl i32 [[TMP1]], 4
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr i8, ptr @GStructArr, i32 [[TMP2]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr [[P_SPLIT]], i32 8
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr [4 x %Foo], ptr @GStructArr, i32 0, i64 %x, i32 2
@@ -344,7 +434,11 @@ define i1 @test10_struct_arr_noinbounds_i64(i64 %x) {
define i1 @test10_struct_arr_noarrayty(i32 %x) {
; CHECK-LABEL: @test10_struct_arr_noarrayty(
-; CHECK-NEXT: [[R:%.*]] = icmp ne i32 [[X:%.*]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[X:%.*]], 4
+; CHECK-NEXT: [[P_SPLIT:%.*]] = getelementptr inbounds i8, ptr @GStructArr, i32 [[TMP1]]
+; CHECK-NEXT: [[P:%.*]] = getelementptr inbounds nuw i8, ptr [[P_SPLIT]], i32 8
+; CHECK-NEXT: [[Q:%.*]] = load i32, ptr [[P]], align 4
+; CHECK-NEXT: [[R:%.*]] = icmp eq i32 [[Q]], 9
; CHECK-NEXT: ret i1 [[R]]
;
%p = getelementptr inbounds %Foo, ptr @GStructArr, i32 %x, i32 2
@@ -359,7 +453,8 @@ define i1 @test10_struct_arr_noarrayty(i32 %x) {
define i1 @pr93017(i64 %idx) {
; CHECK-LABEL: @pr93017(
; CHECK-NEXT: [[TMP1:%.*]] = trunc nsw i64 [[IDX:%.*]] to i32
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds ptr, ptr @table, i32 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[TMP1]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr @table, i32 [[TMP2]]
; CHECK-NEXT: [[V:%.*]] = load ptr, ptr [[GEP]], align 4
; CHECK-NEXT: [[CMP:%.*]] = icmp ne ptr [[V]], null
; CHECK-NEXT: ret i1 [[CMP]]
@@ -375,8 +470,10 @@ define i1 @pr93017(i64 %idx) {
; Mask is 0b10101010
define i1 @load_vs_array_type_mismatch1(i32 %idx) {
; CHECK-LABEL: @load_vs_array_type_mismatch1(
-; CHECK-NEXT: [[TMP1:%.*]] = lshr i32 170, [[IDX:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = trunc i32 [[TMP1]] to i1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr @g_i32_lo, i32 [[TMP1]]
+; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep = getelementptr inbounds i16, ptr @g_i32_lo, i32 %idx
@@ -390,8 +487,10 @@ define i1 @load_vs_array_type_mismatch1(i32 %idx) {
; Mask is 0b01010101
define i1 @load_vs_array_type_mismatch2(i32 %idx) {
; CHECK-LABEL: @load_vs_array_type_mismatch2(
-; CHECK-NEXT: [[TMP1:%.*]] = lshr i32 85, [[IDX:%.*]]
-; CHECK-NEXT: [[CMP:%.*]] = trunc i32 [[TMP1]] to i1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr @g_i32_hi, i32 [[TMP1]]
+; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep = getelementptr inbounds i16, ptr @g_i32_hi, i32 %idx
@@ -405,8 +504,11 @@ define i1 @load_vs_array_type_mismatch2(i32 %idx) {
; idx == 1 || idx == 3
define i1 @load_vs_array_type_mismatch_offset1(i32 %idx) {
; CHECK-LABEL: @load_vs_array_type_mismatch_offset1(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[IDX:%.*]], -3
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[TMP1]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 2
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds i8, ptr @g_i16_1, i32 [[TMP1]]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP_SPLIT]], i32 2
+; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep = getelementptr inbounds {i16, i16}, ptr @g_i16_1, i32 %idx, i32 1
@@ -417,8 +519,11 @@ define i1 @load_vs_array_type_mismatch_offset1(i32 %idx) {
define i1 @load_vs_array_type_mismatch_offset1_separate_gep(i32 %idx) {
; CHECK-LABEL: @load_vs_array_type_mismatch_offset1_separate_gep(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[IDX:%.*]], -3
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[TMP1]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr @g_i16_1, i32 [[TMP1]]
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP1]], i32 2
+; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP2]], align 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep1 = getelementptr inbounds {i16, i16}, ptr @g_i16_1, i32 %idx
@@ -430,8 +535,10 @@ define i1 @load_vs_array_type_mismatch_offset1_separate_gep(i32 %idx) {
define i1 @load_vs_array_type_mismatch_offset1_separate_gep_swapped(i32 %idx) {
; CHECK-LABEL: @load_vs_array_type_mismatch_offset1_separate_gep_swapped(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[IDX:%.*]], -3
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[TMP1]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @g_i16_1, i32 2), i32 [[TMP1]]
+; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP2]], align 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep1 = getelementptr inbounds i8, ptr @g_i16_1, i32 2
@@ -446,8 +553,11 @@ define i1 @load_vs_array_type_mismatch_offset1_separate_gep_swapped(i32 %idx) {
; idx == 0 || idx == 2
define i1 @load_vs_array_type_mismatch_offset2(i32 %idx) {
; CHECK-LABEL: @load_vs_array_type_mismatch_offset2(
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[IDX:%.*]], -3
-; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[TMP1]], 0
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 2
+; CHECK-NEXT: [[GEP_SPLIT:%.*]] = getelementptr inbounds i8, ptr @g_i16_2, i32 [[TMP1]]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP_SPLIT]], i32 2
+; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
; CHECK-NEXT: ret i1 [[CMP]]
;
%gep = getelementptr inbounds {i16, i16}, ptr @g_i16_2, i32 %idx, i32 1
@@ -458,7 +568,8 @@ define i1 @load_vs_array_type_mismatch_offset2(i32 %idx) {
define i1 @offset_larger_than_stride(i32 %idx) {
; CHECK-LABEL: @offset_larger_than_stride(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i16, ptr getelementptr inbounds nuw (i8, ptr @g_i16_1, i32 4), i32 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[IDX:%.*]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr getelementptr inbounds nuw (i8, ptr @g_i16_1, i32 4), i32 [[TMP1]]
; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
; CHECK-NEXT: ret i1 [[CMP]]
@@ -487,9 +598,10 @@ define i1 @load_size_larger_stride(i32 %idx) {
define i1 @cmp_load_constant_array_messy(i32 %x){
; CHECK-LABEL: @cmp_load_constant_array_messy(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[TMP1:%.*]] = and i32 [[TMP0:%.*]], 1073741823
-; CHECK-NEXT: [[TMP2:%.*]] = lshr i32 373, [[TMP1]]
-; CHECK-NEXT: [[COND:%.*]] = trunc i32 [[TMP2]] to i1
+; CHECK-NEXT: [[TMP0:%.*]] = shl i32 [[X:%.*]], 2
+; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr i8, ptr @CG_MESSY, i32 [[TMP0]]
+; CHECK-NEXT: [[ISOK:%.*]] = load i32, ptr [[ISOK_PTR]], align 4
+; CHECK-NEXT: [[COND:%.*]] = icmp slt i32 [[ISOK]], 5
; CHECK-NEXT: ret i1 [[COND]]
;
@@ -502,9 +614,10 @@ entry:
define i1 @cmp_diff_load_constant_array_messy0(i32 %x){
; CHECK-LABEL: @cmp_diff_load_constant_array_messy0(
-; CHECK-NEXT: [[TMP2:%.*]] = and i32 [[TMP1:%.*]], 1073741823
-; CHECK-NEXT: [[TMP3:%.*]] = lshr i32 373, [[TMP2]]
-; CHECK-NEXT: [[COND:%.*]] = trunc i32 [[TMP3]] to i1
+; CHECK-NEXT: [[TMP1:%.*]] = shl i32 [[X:%.*]], 2
+; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr i8, ptr @CG_MESSY, i32 [[TMP1]]
+; CHECK-NEXT: [[ISOK:%.*]] = load i16, ptr [[ISOK_PTR]], align 2
+; CHECK-NEXT: [[COND:%.*]] = icmp slt i16 [[ISOK]], 5
; CHECK-NEXT: ret i1 [[COND]]
;
%isOK_ptr = getelementptr i32, ptr @CG_MESSY, i32 %x
@@ -516,7 +629,7 @@ define i1 @cmp_diff_load_constant_array_messy0(i32 %x){
; Load size larger than store size currently not supported.
define i1 @cmp_diff_load_constant_array_messy1(i32 %x){
; CHECK-LABEL: @cmp_diff_load_constant_array_messy1(
-; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr i6, ptr @CG_MESSY, i32 [[TMP1:%.*]]
+; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr i8, ptr @CG_MESSY, i32 [[X:%.*]]
; CHECK-NEXT: [[ISOK:%.*]] = load i16, ptr [[ISOK_PTR]], align 2
; CHECK-NEXT: [[COND:%.*]] = icmp slt i16 [[ISOK]], 5
; CHECK-NEXT: ret i1 [[COND]]
@@ -530,7 +643,8 @@ define i1 @cmp_diff_load_constant_array_messy1(i32 %x){
define i1 @cmp_load_constant_array_variable_icmp(i32 %x, i32 %y) {
; CHECK-LABEL: @cmp_load_constant_array_variable_icmp(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr inbounds i32, ptr @CG_MESSY, i32 [[X:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[X:%.*]], 2
+; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr inbounds i8, ptr @CG_MESSY, i32 [[TMP0]]
; CHECK-NEXT: [[ISOK:%.*]] = load i32, ptr [[ISOK_PTR]], align 4
; CHECK-NEXT: [[COND:%.*]] = icmp ult i32 [[ISOK]], [[Y:%.*]]
; CHECK-NEXT: ret i1 [[COND]]
@@ -548,7 +662,8 @@ entry:
define i1 @cmp_load_constant_additional_positive_offset(i32 %x) {
; CHECK-LABEL: @cmp_load_constant_additional_positive_offset(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @CG_CLEAR, i32 20), i32 [[X:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[X:%.*]], 2
+; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @CG_CLEAR, i32 20), i32 [[TMP0]]
; CHECK-NEXT: [[ISOK:%.*]] = load i32, ptr [[ISOK_PTR]], align 4
; CHECK-NEXT: [[COND:%.*]] = icmp ult i32 [[ISOK]], 5
; CHECK-NEXT: ret i1 [[COND]]
@@ -563,7 +678,8 @@ entry:
define i1 @cmp_load_constant_additional_negative_offset(i32 %x) {
; CHECK-LABEL: @cmp_load_constant_additional_negative_offset(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ISOK_PTR_SPLIT:%.*]] = getelementptr inbounds [1 x i32], ptr @CG_CLEAR, i32 [[X:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[X:%.*]], 2
+; CHECK-NEXT: [[ISOK_PTR_SPLIT:%.*]] = getelementptr inbounds i8, ptr @CG_CLEAR, i32 [[TMP0]]
; CHECK-NEXT: [[ISOK_PTR:%.*]] = getelementptr inbounds i8, ptr [[ISOK_PTR_SPLIT]], i32 -20
; CHECK-NEXT: [[ISOK:%.*]] = load i32, ptr [[ISOK_PTR]], align 4
; CHECK-NEXT: [[COND:%.*]] = icmp ult i32 [[ISOK]], 5
@@ -578,8 +694,10 @@ entry:
define i1 @cmp_load_multiple_indices(i32 %idx, i32 %idx2) {
; CHECK-LABEL: @cmp_load_multiple_indices(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i16, ptr @g_i16_1, i32 [[IDX:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i16, ptr [[GEP1]], i32 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 1
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr @g_i16_1, i32 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[IDX2:%.*]], 1
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[GEP1]], i32 [[TMP2]]
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP2]], i32 2
; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP3]], align 2
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
@@ -595,8 +713,10 @@ define i1 @cmp_load_multiple_indices(i32 %idx, i32 %idx2) {
define i1 @cmp_load_multiple_indices2(i32 %idx, i32 %idx2) {
; CHECK-LABEL: @cmp_load_multiple_indices2(
-; CHECK-NEXT: [[GEP1_SPLIT:%.*]] = getelementptr inbounds [1 x i16], ptr @g_i16_1, i32 [[IDX:%.*]]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i16, ptr [[GEP1_SPLIT]], i32 [[IDX2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[IDX:%.*]], 1
+; CHECK-NEXT: [[GEP1_SPLIT:%.*]] = getelementptr inbounds i8, ptr @g_i16_1, i32 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[IDX2:%.*]], 1
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[GEP1_SPLIT]], i32 [[TMP2]]
; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP1]], i32 2
; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP2]], align 2
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 0
diff --git a/llvm/test/Transforms/InstCombine/load.ll b/llvm/test/Transforms/InstCombine/load.ll
index 736a31fc6ae0a..7576c23ec2784 100644
--- a/llvm/test/Transforms/InstCombine/load.ll
+++ b/llvm/test/Transforms/InstCombine/load.ll
@@ -91,7 +91,8 @@ define i32 @load_gep_null_not_inbounds(i64 %X) {
define i32 @test7_no_null_opt(i32 %X) #0 {
; CHECK-LABEL: @test7_no_null_opt(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[X:%.*]] to i64
-; CHECK-NEXT: [[V:%.*]] = getelementptr i32, ptr null, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[V:%.*]] = getelementptr i8, ptr null, i64 [[TMP2]]
; CHECK-NEXT: [[R:%.*]] = load i32, ptr [[V]], align 4
; CHECK-NEXT: ret i32 [[R]]
;
@@ -455,7 +456,8 @@ define i32 @load_select_with_null_gep(i1 %cond, ptr %p, i64 %off) {
define i16 @load_select_with_null_gep2(i1 %cond, ptr %p, i64 %x) {
; CHECK-LABEL: @load_select_with_null_gep2(
; CHECK-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr i8, ptr [[SEL:%.*]], i64 -2
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i16, ptr [[INVARIANT_GEP]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[X:%.*]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[INVARIANT_GEP]], i64 [[TMP1]]
; CHECK-NEXT: [[RES:%.*]] = load i16, ptr [[GEP]], align 2
; CHECK-NEXT: ret i16 [[RES]]
;
@@ -469,8 +471,10 @@ define i16 @load_select_with_null_gep2(i1 %cond, ptr %p, i64 %x) {
define i16 @load_select_with_null_gep3(i1 %cond, ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @load_select_with_null_gep3(
; CHECK-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr i8, ptr [[SEL:%.*]], i64 -2
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i16, ptr [[INVARIANT_GEP]], i64 [[X:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i16, ptr [[GEP]], i64 [[Y:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[X:%.*]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr [[INVARIANT_GEP]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[Y:%.*]], 1
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[GEP]], i64 [[TMP2]]
; CHECK-NEXT: [[RES:%.*]] = load i16, ptr [[GEP2]], align 2
; CHECK-NEXT: ret i16 [[RES]]
;
diff --git a/llvm/test/Transforms/InstCombine/loadstore-alignment.ll b/llvm/test/Transforms/InstCombine/loadstore-alignment.ll
index e6a752575901f..e3a28e8d3cfb7 100644
--- a/llvm/test/Transforms/InstCombine/loadstore-alignment.ll
+++ b/llvm/test/Transforms/InstCombine/loadstore-alignment.ll
@@ -20,7 +20,8 @@ define <2 x i64> @static_hem() {
define <2 x i64> @hem(i32 %i) {
; CHECK-LABEL: @hem(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[I:%.*]] to i64
-; CHECK-NEXT: [[T:%.*]] = getelementptr <2 x i64>, ptr @x, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 4
+; CHECK-NEXT: [[T:%.*]] = getelementptr i8, ptr @x, i64 [[TMP2]]
; CHECK-NEXT: [[L:%.*]] = load <2 x i64>, ptr [[T]], align 1
; CHECK-NEXT: ret <2 x i64> [[L]]
;
@@ -33,8 +34,10 @@ define <2 x i64> @hem_2d(i32 %i, i32 %j) {
; CHECK-LABEL: @hem_2d(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[I:%.*]] to i64
; CHECK-NEXT: [[TMP2:%.*]] = sext i32 [[J:%.*]] to i64
-; CHECK-NEXT: [[T_SPLIT:%.*]] = getelementptr [13 x <2 x i64>], ptr @xx, i64 [[TMP1]]
-; CHECK-NEXT: [[T:%.*]] = getelementptr <2 x i64>, ptr [[T_SPLIT]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = mul nsw i64 [[TMP1]], 208
+; CHECK-NEXT: [[T_SPLIT:%.*]] = getelementptr i8, ptr @xx, i64 [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP2]], 4
+; CHECK-NEXT: [[T:%.*]] = getelementptr i8, ptr [[T_SPLIT]], i64 [[TMP4]]
; CHECK-NEXT: [[L:%.*]] = load <2 x i64>, ptr [[T]], align 1
; CHECK-NEXT: ret <2 x i64> [[L]]
;
@@ -78,7 +81,8 @@ define void @static_hem_store(<2 x i64> %y) {
define void @hem_store(i32 %i, <2 x i64> %y) {
; CHECK-LABEL: @hem_store(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[I:%.*]] to i64
-; CHECK-NEXT: [[T:%.*]] = getelementptr <2 x i64>, ptr @x, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 4
+; CHECK-NEXT: [[T:%.*]] = getelementptr i8, ptr @x, i64 [[TMP2]]
; CHECK-NEXT: store <2 x i64> [[Y:%.*]], ptr [[T]], align 1
; CHECK-NEXT: ret void
;
@@ -91,8 +95,10 @@ define void @hem_2d_store(i32 %i, i32 %j, <2 x i64> %y) {
; CHECK-LABEL: @hem_2d_store(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[I:%.*]] to i64
; CHECK-NEXT: [[TMP2:%.*]] = sext i32 [[J:%.*]] to i64
-; CHECK-NEXT: [[T_SPLIT:%.*]] = getelementptr [13 x <2 x i64>], ptr @xx, i64 [[TMP1]]
-; CHECK-NEXT: [[T:%.*]] = getelementptr <2 x i64>, ptr [[T_SPLIT]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = mul nsw i64 [[TMP1]], 208
+; CHECK-NEXT: [[T_SPLIT:%.*]] = getelementptr i8, ptr @xx, i64 [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP2]], 4
+; CHECK-NEXT: [[T:%.*]] = getelementptr i8, ptr [[T_SPLIT]], i64 [[TMP4]]
; CHECK-NEXT: store <2 x i64> [[Y:%.*]], ptr [[T]], align 1
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/loadstore-metadata.ll b/llvm/test/Transforms/InstCombine/loadstore-metadata.ll
index 64f9f2797a439..36905b90aec0f 100644
--- a/llvm/test/Transforms/InstCombine/loadstore-metadata.ll
+++ b/llvm/test/Transforms/InstCombine/loadstore-metadata.ll
@@ -128,9 +128,11 @@ define void @test_load_cast_combine_loop(ptr %src, ptr %dst, i32 %n) {
; CHECK: [[LOOP]]:
; CHECK-NEXT: [[I:%.*]] = phi i32 [ 0, %[[ENTRY]] ], [ [[I_NEXT:%.*]], %[[LOOP]] ]
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[I]] to i64
-; CHECK-NEXT: [[SRC_GEP:%.*]] = getelementptr inbounds float, ptr [[SRC]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[SRC_GEP:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[I]] to i64
-; CHECK-NEXT: [[DST_GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[DST_GEP:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP3]]
; CHECK-NEXT: [[L1:%.*]] = load i32, ptr [[SRC_GEP]], align 4, !llvm.access.group [[ACC_GRP9:![0-9]+]]
; CHECK-NEXT: store i32 [[L1]], ptr [[DST_GEP]], align 4
; CHECK-NEXT: [[I_NEXT]] = add i32 [[I]], 1
diff --git a/llvm/test/Transforms/InstCombine/masked_intrinsics-inseltpoison.ll b/llvm/test/Transforms/InstCombine/masked_intrinsics-inseltpoison.ll
index 02539d17fac04..56821c0de897b 100644
--- a/llvm/test/Transforms/InstCombine/masked_intrinsics-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/masked_intrinsics-inseltpoison.ll
@@ -58,7 +58,7 @@ define <2 x double> @load_lane0(ptr %ptr, double %pt) {
define double @load_all(ptr %base, double %pt) {
; CHECK-LABEL: @load_all(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <4 x i64> <i64 0, i64 poison, i64 2, i64 3>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <4 x i64> <i64 0, i64 poison, i64 16, i64 24>
; CHECK-NEXT: [[RES:%.*]] = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 4 [[PTRS]], <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x double> poison)
; CHECK-NEXT: [[ELT:%.*]] = extractelement <4 x double> [[RES]], i64 2
; CHECK-NEXT: ret double [[ELT]]
@@ -200,7 +200,7 @@ define <2 x double> @gather_onemask(<2 x ptr> %ptrs, <2 x double> %passthru) {
define <4 x double> @gather_lane2(ptr %base, double %pt) {
; CHECK-LABEL: @gather_lane2(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <4 x i64> <i64 poison, i64 poison, i64 2, i64 poison>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <4 x i64> <i64 poison, i64 poison, i64 16, i64 poison>
; CHECK-NEXT: [[PT_V1:%.*]] = insertelement <4 x double> poison, double [[PT:%.*]], i64 0
; CHECK-NEXT: [[PT_V2:%.*]] = shufflevector <4 x double> [[PT_V1]], <4 x double> poison, <4 x i32> <i32 0, i32 0, i32 poison, i32 0>
; CHECK-NEXT: [[RES:%.*]] = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 4 [[PTRS]], <4 x i1> <i1 false, i1 false, i1 true, i1 false>, <4 x double> [[PT_V2]])
@@ -215,7 +215,7 @@ define <4 x double> @gather_lane2(ptr %base, double %pt) {
define <2 x double> @gather_lane0_maybe(ptr %base, double %pt, <2 x i1> %mask) {
; CHECK-LABEL: @gather_lane0_maybe(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 1>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 8>
; CHECK-NEXT: [[PT_V1:%.*]] = insertelement <2 x double> poison, double [[PT:%.*]], i64 0
; CHECK-NEXT: [[PT_V2:%.*]] = shufflevector <2 x double> [[PT_V1]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[MASK2:%.*]] = insertelement <2 x i1> [[MASK:%.*]], i1 false, i64 1
@@ -232,7 +232,7 @@ define <2 x double> @gather_lane0_maybe(ptr %base, double %pt, <2 x i1> %mask)
define <2 x double> @gather_lane0_maybe_spec(ptr %base, double %pt, <2 x i1> %mask) {
; CHECK-LABEL: @gather_lane0_maybe_spec(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 1>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 8>
; CHECK-NEXT: [[PT_V1:%.*]] = insertelement <2 x double> poison, double [[PT:%.*]], i64 0
; CHECK-NEXT: [[PT_V2:%.*]] = shufflevector <2 x double> [[PT_V1]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[MASK2:%.*]] = insertelement <2 x i1> [[MASK:%.*]], i1 false, i64 1
@@ -258,7 +258,7 @@ define void @scatter_zeromask(<2 x ptr> %ptrs, <2 x double> %val) {
define void @scatter_demandedelts(ptr %ptr, double %val) {
; CHECK-LABEL: @scatter_demandedelts(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 poison>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 poison>
; CHECK-NEXT: [[VALVEC1:%.*]] = insertelement <2 x double> poison, double [[VAL:%.*]], i64 0
; CHECK-NEXT: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> [[VALVEC1]], <2 x ptr> align 8 [[PTRS]], <2 x i1> <i1 true, i1 false>)
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/InstCombine/masked_intrinsics.ll b/llvm/test/Transforms/InstCombine/masked_intrinsics.ll
index 6aadb08b16991..8efef217b1ca3 100644
--- a/llvm/test/Transforms/InstCombine/masked_intrinsics.ll
+++ b/llvm/test/Transforms/InstCombine/masked_intrinsics.ll
@@ -67,7 +67,7 @@ define <2 x double> @load_lane0(ptr %ptr, double %pt) {
define double @load_all(ptr %base, double %pt) {
; CHECK-LABEL: @load_all(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <4 x i64> <i64 0, i64 poison, i64 2, i64 3>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <4 x i64> <i64 0, i64 poison, i64 16, i64 24>
; CHECK-NEXT: [[RES:%.*]] = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 4 [[PTRS]], <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x double> poison)
; CHECK-NEXT: [[ELT:%.*]] = extractelement <4 x double> [[RES]], i64 2
; CHECK-NEXT: ret double [[ELT]]
@@ -242,7 +242,7 @@ define <2 x double> @gather_one_withpoisonmask(<2 x ptr> %ptrs, <2 x double> %pa
define <4 x double> @gather_lane2(ptr %base, double %pt) {
; CHECK-LABEL: @gather_lane2(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <4 x i64> <i64 poison, i64 poison, i64 2, i64 poison>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <4 x i64> <i64 poison, i64 poison, i64 16, i64 poison>
; CHECK-NEXT: [[PT_V1:%.*]] = insertelement <4 x double> poison, double [[PT:%.*]], i64 0
; CHECK-NEXT: [[PT_V2:%.*]] = shufflevector <4 x double> [[PT_V1]], <4 x double> poison, <4 x i32> <i32 0, i32 0, i32 poison, i32 0>
; CHECK-NEXT: [[RES:%.*]] = call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 4 [[PTRS]], <4 x i1> <i1 false, i1 false, i1 true, i1 false>, <4 x double> [[PT_V2]])
@@ -257,7 +257,7 @@ define <4 x double> @gather_lane2(ptr %base, double %pt) {
define <2 x double> @gather_lane0_maybe(ptr %base, double %pt, <2 x i1> %mask) {
; CHECK-LABEL: @gather_lane0_maybe(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 1>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 8>
; CHECK-NEXT: [[PT_V1:%.*]] = insertelement <2 x double> poison, double [[PT:%.*]], i64 0
; CHECK-NEXT: [[PT_V2:%.*]] = shufflevector <2 x double> [[PT_V1]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[MASK2:%.*]] = insertelement <2 x i1> [[MASK:%.*]], i1 false, i64 1
@@ -274,7 +274,7 @@ define <2 x double> @gather_lane0_maybe(ptr %base, double %pt, <2 x i1> %mask)
define <2 x double> @gather_lane0_maybe_spec(ptr %base, double %pt, <2 x i1> %mask) {
; CHECK-LABEL: @gather_lane0_maybe_spec(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 1>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[BASE:%.*]], <2 x i64> <i64 0, i64 8>
; CHECK-NEXT: [[PT_V1:%.*]] = insertelement <2 x double> poison, double [[PT:%.*]], i64 0
; CHECK-NEXT: [[PT_V2:%.*]] = shufflevector <2 x double> [[PT_V1]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[MASK2:%.*]] = insertelement <2 x i1> [[MASK:%.*]], i1 false, i64 1
@@ -317,7 +317,7 @@ define void @scatter_one_withpoison_mask(<2 x ptr> %ptrs, <2 x double> %val) {
define void @scatter_demandedelts(ptr %ptr, double %val) {
; CHECK-LABEL: @scatter_demandedelts(
-; CHECK-NEXT: [[PTRS:%.*]] = getelementptr double, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 poison>
+; CHECK-NEXT: [[PTRS:%.*]] = getelementptr i8, ptr [[PTR:%.*]], <2 x i64> <i64 0, i64 poison>
; CHECK-NEXT: [[VALVEC1:%.*]] = insertelement <2 x double> poison, double [[VAL:%.*]], i64 0
; CHECK-NEXT: call void @llvm.masked.scatter.v2f64.v2p0(<2 x double> [[VALVEC1]], <2 x ptr> align 8 [[PTRS]], <2 x i1> <i1 true, i1 false>)
; CHECK-NEXT: ret void
diff --git a/llvm/test/Transforms/InstCombine/mem-gep-zidx.ll b/llvm/test/Transforms/InstCombine/mem-gep-zidx.ll
index 8840374de1c1e..b6c28c327223d 100644
--- a/llvm/test/Transforms/InstCombine/mem-gep-zidx.ll
+++ b/llvm/test/Transforms/InstCombine/mem-gep-zidx.ll
@@ -9,7 +9,11 @@ target triple = "powerpc64-unknown-linux-gnu"
define signext i32 @test1(i32 signext %x) #0 {
; CHECK-LABEL: @test1(
-; CHECK-NEXT: ret i32 12
+; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[X:%.*]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr @f.a, i64 [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; CHECK-NEXT: ret i32 [[R]]
;
%idxprom = sext i32 %x to i64
%arrayidx = getelementptr inbounds [1 x i32], ptr @f.a, i64 0, i64 %idxprom
@@ -21,7 +25,10 @@ declare void @foo(ptr %p)
define void @test2(i32 signext %x, i64 %v) #0 {
; CHECK-LABEL: @test2(
; CHECK-NEXT: [[P:%.*]] = alloca i64, align 8
-; CHECK-NEXT: store i64 [[V:%.*]], ptr [[P]], align 8
+; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[X:%.*]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP1]]
+; CHECK-NEXT: store i64 [[V:%.*]], ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: call void @foo(ptr nonnull [[P]]) #[[ATTR1:[0-9]+]]
; CHECK-NEXT: ret void
;
@@ -35,7 +42,11 @@ define void @test2(i32 signext %x, i64 %v) #0 {
define signext i32 @test3(i32 signext %x, i1 %y) #0 {
; CHECK-LABEL: @test3(
-; CHECK-NEXT: [[R:%.*]] = select i1 [[Y:%.*]], i32 12, i32 55
+; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[X:%.*]] to i64
+; CHECK-NEXT: [[P:%.*]] = select i1 [[Y:%.*]], ptr @f.a, ptr @f.b
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP1]]
+; CHECK-NEXT: [[R:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: ret i32 [[R]]
;
%idxprom = sext i32 %x to i64
@@ -48,7 +59,8 @@ define signext i32 @test3(i32 signext %x, i1 %y) #0 {
define signext i32 @test4(i32 signext %x, i1 %y) #0 {
; CHECK-LABEL: @test4(
; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[X:%.*]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr @f.c, i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr @f.c, i64 [[TMP1]]
; CHECK-NEXT: [[R:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: ret i32 [[R]]
;
diff --git a/llvm/test/Transforms/InstCombine/memcpy-addrspace.ll b/llvm/test/Transforms/InstCombine/memcpy-addrspace.ll
index cfec7b8558412..0bc4e6e2266ce 100644
--- a/llvm/test/Transforms/InstCombine/memcpy-addrspace.ll
+++ b/llvm/test/Transforms/InstCombine/memcpy-addrspace.ll
@@ -6,9 +6,11 @@
define void @test_load(ptr addrspace(1) %out, i64 %x) {
; CHECK-LABEL: @test_load(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr addrspace(2) @test.data, i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr addrspace(2) @test.data, i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(2) [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[TMP0]], ptr addrspace(1) [[ARRAYIDX1]], align 4
; CHECK-NEXT: ret void
;
@@ -26,9 +28,11 @@ entry:
define void @test_load_gep_nusw_nuw(ptr addrspace(1) %out, i64 %x) {
; CHECK-LABEL: @test_load_gep_nusw_nuw(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr nusw nuw i32, ptr addrspace(2) @test.data, i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr nusw nuw i8, ptr addrspace(2) @test.data, i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(2) [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr nusw nuw i32, ptr addrspace(1) [[OUT:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[X]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr nusw nuw i8, ptr addrspace(1) [[OUT:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[TMP0]], ptr addrspace(1) [[ARRAYIDX1]], align 4
; CHECK-NEXT: ret void
;
@@ -45,9 +49,11 @@ entry:
define void @test_load_bitcast_chain(ptr addrspace(1) %out, i64 %x) {
; CHECK-LABEL: @test_load_bitcast_chain(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr addrspace(2) @test.data, i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr addrspace(2) @test.data, i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr addrspace(2) [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[TMP0]], ptr addrspace(1) [[ARRAYIDX1]], align 4
; CHECK-NEXT: ret void
;
@@ -66,9 +72,11 @@ define void @test_call(ptr addrspace(1) %out, i64 %x) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[DATA:%.*]] = alloca [8 x i32], align 4
; CHECK-NEXT: call void @llvm.memcpy.p0.p2.i64(ptr noundef nonnull align 4 dereferenceable(32) [[DATA]], ptr addrspace(2) noundef align 4 dereferenceable(32) @test.data, i64 32, i1 false)
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[DATA]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[DATA]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @foo(ptr nonnull [[ARRAYIDX]])
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[TMP0]], ptr addrspace(1) [[ARRAYIDX1]], align 4
; CHECK-NEXT: ret void
;
@@ -87,9 +95,11 @@ define void @test_call_no_null_opt(ptr addrspace(1) %out, i64 %x) #0 {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[DATA:%.*]] = alloca [8 x i32], align 4
; CHECK-NEXT: call void @llvm.memcpy.p0.p2.i64(ptr noundef nonnull align 4 dereferenceable(32) [[DATA]], ptr addrspace(2) noundef align 4 dereferenceable(32) @test.data, i64 32, i1 false)
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[DATA]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[DATA]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @foo(ptr [[ARRAYIDX]])
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[TMP0]], ptr addrspace(1) [[ARRAYIDX1]], align 4
; CHECK-NEXT: ret void
;
@@ -108,12 +118,15 @@ define void @test_load_and_call(ptr addrspace(1) %out, i64 %x, i64 %y) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[DATA:%.*]] = alloca [8 x i32], align 4
; CHECK-NEXT: call void @llvm.memcpy.p0.p2.i64(ptr noundef nonnull align 4 dereferenceable(32) [[DATA]], ptr addrspace(2) noundef align 4 dereferenceable(32) @test.data, i64 32, i1 false)
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[DATA]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[DATA]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[TMP0]], ptr addrspace(1) [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @foo(ptr nonnull [[ARRAYIDX]])
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT]], i64 [[Y:%.*]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[Y:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT]], i64 [[TMP4]]
; CHECK-NEXT: store i32 [[TMP1]], ptr addrspace(1) [[ARRAYIDX2]], align 4
; CHECK-NEXT: ret void
;
@@ -135,12 +148,15 @@ define void @test_load_and_call_no_null_opt(ptr addrspace(1) %out, i64 %x, i64 %
; CHECK-NEXT: entry:
; CHECK-NEXT: [[DATA:%.*]] = alloca [8 x i32], align 4
; CHECK-NEXT: call void @llvm.memcpy.p0.p2.i64(ptr noundef nonnull align 4 dereferenceable(32) [[DATA]], ptr addrspace(2) noundef align 4 dereferenceable(32) @test.data, i64 32, i1 false)
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[DATA]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[DATA]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[TMP0]], ptr addrspace(1) [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @foo(ptr [[ARRAYIDX]])
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr addrspace(1) [[OUT]], i64 [[Y:%.*]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[Y:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[OUT]], i64 [[TMP4]]
; CHECK-NEXT: store i32 [[TMP1]], ptr addrspace(1) [[ARRAYIDX2]], align 4
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/InstCombine/memcpy-from-global.ll b/llvm/test/Transforms/InstCombine/memcpy-from-global.ll
index f10ba1e3d27e6..5e333bb025c48 100644
--- a/llvm/test/Transforms/InstCombine/memcpy-from-global.ll
+++ b/llvm/test/Transforms/InstCombine/memcpy-from-global.ll
@@ -8,25 +8,29 @@ define float @test1(i32 %hash, float %x, float %y, float %z, float %w) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[T3:%.*]] = shl i32 [[HASH:%.*]], 2
; CHECK-NEXT: [[T5:%.*]] = and i32 [[T3]], 124
-; CHECK-NEXT: [[TMP0:%.*]] = zext nneg i32 [[T5]] to i64
-; CHECK-NEXT: [[T753:%.*]] = getelementptr float, ptr @C.0.1248, i64 [[TMP0]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw nsw i32 [[T5]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TMP0]] to i64
+; CHECK-NEXT: [[T753:%.*]] = getelementptr i8, ptr @C.0.1248, i64 [[TMP1]]
; CHECK-NEXT: [[T9:%.*]] = load float, ptr [[T753]], align 4
; CHECK-NEXT: [[T11:%.*]] = fmul float [[T9]], [[X:%.*]]
; CHECK-NEXT: [[T13:%.*]] = fadd float [[T11]], 0.000000e+00
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[T5]] to i64
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr float, ptr @C.0.1248, i64 [[TMP1]]
+; CHECK-NEXT: [[T17_SUM52:%.*]] = shl nuw nsw i32 [[T5]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = zext nneg i32 [[T17_SUM52]] to i64
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr @C.0.1248, i64 [[TMP3]]
; CHECK-NEXT: [[T1851:%.*]] = getelementptr i8, ptr [[TMP2]], i64 4
; CHECK-NEXT: [[T19:%.*]] = load float, ptr [[T1851]], align 4
; CHECK-NEXT: [[T21:%.*]] = fmul float [[T19]], [[Y:%.*]]
; CHECK-NEXT: [[T23:%.*]] = fadd float [[T21]], [[T13]]
-; CHECK-NEXT: [[TMP3:%.*]] = zext nneg i32 [[T5]] to i64
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr float, ptr @C.0.1248, i64 [[TMP3]]
+; CHECK-NEXT: [[T27_SUM50:%.*]] = shl nuw nsw i32 [[T5]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = zext nneg i32 [[T27_SUM50]] to i64
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr @C.0.1248, i64 [[TMP5]]
; CHECK-NEXT: [[T2849:%.*]] = getelementptr i8, ptr [[TMP4]], i64 8
; CHECK-NEXT: [[T29:%.*]] = load float, ptr [[T2849]], align 4
; CHECK-NEXT: [[T31:%.*]] = fmul float [[T29]], [[Z:%.*]]
; CHECK-NEXT: [[T33:%.*]] = fadd float [[T31]], [[T23]]
-; CHECK-NEXT: [[TMP5:%.*]] = zext nneg i32 [[T5]] to i64
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr @C.0.1248, i64 [[TMP5]]
+; CHECK-NEXT: [[T37_SUM48:%.*]] = shl nuw nsw i32 [[T5]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = zext nneg i32 [[T37_SUM48]] to i64
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr @C.0.1248, i64 [[TMP7]]
; CHECK-NEXT: [[T3847:%.*]] = getelementptr i8, ptr [[TMP6]], i64 12
; CHECK-NEXT: [[T39:%.*]] = load float, ptr [[T3847]], align 4
; CHECK-NEXT: [[T41:%.*]] = fmul float [[T39]], [[W:%.*]]
@@ -337,7 +341,8 @@ define float @test11_volatile(i64 %i) {
; CHECK-NEXT: [[A:%.*]] = alloca [4 x float], align 4
; CHECK-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[A]])
; CHECK-NEXT: call void @llvm.memcpy.p0.p1.i64(ptr align 4 [[A]], ptr addrspace(1) align 4 @I, i64 16, i1 true)
-; CHECK-NEXT: [[G:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[I:%.*]], 2
+; CHECK-NEXT: [[G:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP0]]
; CHECK-NEXT: [[R:%.*]] = load float, ptr [[G]], align 4
; CHECK-NEXT: ret float [[R]]
;
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 df3152a6a4353..ebd10f6f9da8e 100644
--- a/llvm/test/Transforms/InstCombine/multi-size-address-space-pointer.ll
+++ b/llvm/test/Transforms/InstCombine/multi-size-address-space-pointer.ll
@@ -120,7 +120,8 @@ define ptr addrspace(3) @shrink_gep_constant_index_64_as3(ptr addrspace(3) %p) {
define ptr addrspace(2) @shrink_gep_variable_index_64_as2(ptr addrspace(2) %p, i64 %idx) {
; CHECK-LABEL: @shrink_gep_variable_index_64_as2(
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[IDX:%.*]] to i8
-; CHECK-NEXT: [[RET:%.*]] = getelementptr i32, ptr addrspace(2) [[P:%.*]], i8 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i8 [[TMP1]], 2
+; CHECK-NEXT: [[RET:%.*]] = getelementptr i8, ptr addrspace(2) [[P:%.*]], i8 [[TMP2]]
; CHECK-NEXT: ret ptr addrspace(2) [[RET]]
;
%ret = getelementptr i32, ptr addrspace(2) %p, i64 %idx
@@ -130,7 +131,8 @@ define ptr addrspace(2) @shrink_gep_variable_index_64_as2(ptr addrspace(2) %p, i
define ptr addrspace(1) @grow_gep_variable_index_8_as1(ptr addrspace(1) %p, i8 %idx) {
; CHECK-LABEL: @grow_gep_variable_index_8_as1(
; CHECK-NEXT: [[TMP1:%.*]] = sext i8 [[IDX:%.*]] to i64
-; CHECK-NEXT: [[RET:%.*]] = getelementptr i32, ptr addrspace(1) [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[RET:%.*]] = getelementptr i8, ptr addrspace(1) [[P:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr addrspace(1) [[RET]]
;
%ret = getelementptr i32, ptr addrspace(1) %p, i8 %idx
diff --git a/llvm/test/Transforms/InstCombine/narrow.ll b/llvm/test/Transforms/InstCombine/narrow.ll
index 5fad7f07090b9..7161365cf9645 100644
--- a/llvm/test/Transforms/InstCombine/narrow.ll
+++ b/llvm/test/Transforms/InstCombine/narrow.ll
@@ -109,7 +109,8 @@ define i1 @searchArray1(i32 %needle, ptr %haystack) {
; CHECK-NEXT: [[INDVAR:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INDVAR_NEXT:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[FOUND:%.*]] = phi i8 [ 0, [[ENTRY]] ], [ [[OR:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDVAR]] to i64
-; CHECK-NEXT: [[IDX:%.*]] = getelementptr i32, ptr [[HAYSTACK:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[IDX:%.*]] = getelementptr i8, ptr [[HAYSTACK:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[LD:%.*]] = load i32, ptr [[IDX]], align 4
; CHECK-NEXT: [[CMP1:%.*]] = icmp eq i32 [[LD]], [[NEEDLE:%.*]]
; CHECK-NEXT: [[ZEXT:%.*]] = zext i1 [[CMP1]] to i8
@@ -152,7 +153,8 @@ define i1 @searchArray2(i32 %hay, ptr %haystack) {
; CHECK: loop:
; CHECK-NEXT: [[INDVAR:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVAR_NEXT:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[FOUND:%.*]] = phi i8 [ 1, [[ENTRY]] ], [ [[AND:%.*]], [[LOOP]] ]
-; CHECK-NEXT: [[IDX:%.*]] = getelementptr i32, ptr [[HAYSTACK:%.*]], i64 [[INDVAR]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[INDVAR]], 2
+; CHECK-NEXT: [[IDX:%.*]] = getelementptr i8, ptr [[HAYSTACK:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[LD:%.*]] = load i32, ptr [[IDX]], align 4
; CHECK-NEXT: [[CMP1:%.*]] = icmp eq i32 [[LD]], [[HAY:%.*]]
; CHECK-NEXT: [[AND]] = select i1 [[CMP1]], i8 [[FOUND]], i8 0
diff --git a/llvm/test/Transforms/InstCombine/opaque-ptr.ll b/llvm/test/Transforms/InstCombine/opaque-ptr.ll
index 1d80724d2355e..b31812b9d1054 100644
--- a/llvm/test/Transforms/InstCombine/opaque-ptr.ll
+++ b/llvm/test/Transforms/InstCombine/opaque-ptr.ll
@@ -223,7 +223,8 @@ define ptr @geps_combinable_different_elem_type5(ptr %a) {
define ptr @geps_combinable_different_elem_type6(ptr %a, i64 %idx) {
; CHECK-LABEL: @geps_combinable_different_elem_type6(
-; CHECK-NEXT: [[A2:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[A2:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A2]], i64 4
; CHECK-NEXT: ret ptr [[A3]]
;
@@ -234,7 +235,8 @@ define ptr @geps_combinable_different_elem_type6(ptr %a, i64 %idx) {
define ptr @geps_combinable_different_elem_type7(ptr %a, i64 %idx) {
; CHECK-LABEL: @geps_combinable_different_elem_type7(
-; CHECK-NEXT: [[A2_SPLIT:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[A2_SPLIT:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A2_SPLIT]], i64 8
; CHECK-NEXT: ret ptr [[A3]]
;
@@ -245,7 +247,8 @@ define ptr @geps_combinable_different_elem_type7(ptr %a, i64 %idx) {
define ptr @geps_combinable_different_elem_type8(ptr %a, i64 %idx) {
; CHECK-LABEL: @geps_combinable_different_elem_type8(
-; CHECK-NEXT: [[A2_SPLIT:%.*]] = getelementptr inbounds { { i32, i32 } }, ptr [[A:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[A2_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[A3:%.*]] = getelementptr inbounds nuw i8, ptr [[A2_SPLIT]], i64 8
; CHECK-NEXT: ret ptr [[A3]]
;
@@ -256,7 +259,8 @@ define ptr @geps_combinable_different_elem_type8(ptr %a, i64 %idx) {
define ptr @geps_combinable_different_elem_type9(ptr %a, i64 %idx) {
; CHECK-LABEL: @geps_combinable_different_elem_type9(
-; CHECK-NEXT: [[A3:%.*]] = getelementptr inbounds { { i32, i32 } }, ptr [[A:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[A3:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[A3]]
;
%a2 = getelementptr inbounds { { i32, i32 } }, ptr %a, i64 %idx, i32 0, i32 1
@@ -281,7 +285,8 @@ define ptr @geps_combinable_different_elem_type_extra_use1(ptr %a) {
define ptr @geps_combinable_different_elem_type_extra_use2(ptr %a, i64 %idx) {
; CHECK-LABEL: @geps_combinable_different_elem_type_extra_use2(
-; CHECK-NEXT: [[A2:%.*]] = getelementptr { i32, i32 }, ptr [[A:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[A2:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: call void @use(ptr [[A2]])
; CHECK-NEXT: [[A3:%.*]] = getelementptr i8, ptr [[A2]], i64 4
; CHECK-NEXT: ret ptr [[A3]]
@@ -320,7 +325,11 @@ define ptr @geps_combinable_scalable_vector_array(ptr %a, i64 %idx) {
define i1 @compare_geps_same_indices(ptr %a, ptr %b, i64 %idx) {
; CHECK-LABEL: @compare_geps_same_indices(
-; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[A:%.*]], [[B:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr [[A1:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[IDX]], 2
+; CHECK-NEXT: [[B:%.*]] = getelementptr i8, ptr [[B1:%.*]], i64 [[TMP2]]
+; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[A]], [[B]]
; CHECK-NEXT: ret i1 [[C]]
;
%a2 = getelementptr i32, ptr %a, i64 %idx
@@ -331,8 +340,10 @@ define i1 @compare_geps_same_indices(ptr %a, ptr %b, i64 %idx) {
define i1 @compare_geps_same_indices_different_types(ptr %a, ptr %b, i64 %idx) {
; CHECK-LABEL: @compare_geps_same_indices_different_types(
-; CHECK-NEXT: [[A2:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[IDX:%.*]]
-; CHECK-NEXT: [[B2:%.*]] = getelementptr i64, ptr [[B:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[A2:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[IDX]], 3
+; CHECK-NEXT: [[B2:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[C:%.*]] = icmp eq ptr [[A2]], [[B2]]
; CHECK-NEXT: ret i1 [[C]]
;
@@ -354,9 +365,9 @@ define i1 @compare_gep_with_base(ptr %p, i64 %idx) {
define <2 x i1> @compare_gep_with_base_vector1(<2 x ptr> %p, i64 %idx) {
; CHECK-LABEL: @compare_gep_with_base_vector1(
-; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[IDX:%.*]], i64 0
-; CHECK-NEXT: [[TMP1:%.*]] = shl <2 x i64> [[DOTSPLATINSERT]], <i64 2, i64 0>
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i64> [[TMP1]], zeroinitializer
+; CHECK-NEXT: [[IDX:%.*]] = shl nsw i64 [[IDX1:%.*]], 2
+; CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[IDX]], i64 0
+; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i64> [[DOTSPLATINSERT]], zeroinitializer
; CHECK-NEXT: [[C:%.*]] = shufflevector <2 x i1> [[TMP2]], <2 x i1> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: ret <2 x i1> [[C]]
;
@@ -377,8 +388,8 @@ define <2 x i1> @compare_gep_with_base_vector2(<2 x ptr> %p, <2 x i64> %idx) {
define <4 x i1> @compare_geps_same_indices_scalar_vector_base_mismatch(ptr %ptr, <4 x ptr> %ptrs) {
; CHECK-LABEL: @compare_geps_same_indices_scalar_vector_base_mismatch(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i16, <4 x ptr> [[PTRS:%.*]], <4 x i64> <i64 1, i64 2, i64 3, i64 4>
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i16, ptr [[PTR:%.*]], <4 x i64> <i64 1, i64 2, i64 3, i64 4>
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, <4 x ptr> [[PTRS:%.*]], <4 x i64> <i64 2, i64 4, i64 6, i64 8>
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[PTR:%.*]], <4 x i64> <i64 2, i64 4, i64 6, i64 8>
; CHECK-NEXT: [[CMP:%.*]] = icmp eq <4 x ptr> [[GEP1]], [[GEP2]]
; CHECK-NEXT: ret <4 x i1> [[CMP]]
;
@@ -544,7 +555,11 @@ define i1 @cmp_load_gep_global_different_load_type(i64 %idx) {
define i1 @cmp_load_gep_global_different_gep_type(i64 %idx) {
; CHECK-LABEL: @cmp_load_gep_global_different_gep_type(
-; CHECK-NEXT: ret i1 false
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, ptr @ary, i64 [[TMP1]]
+; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp eq i16 [[LOAD]], 3
+; CHECK-NEXT: ret i1 [[CMP]]
;
%gep = getelementptr [4 x i16], ptr @ary, i64 0, i64 %idx
%load = load i16, ptr %gep
diff --git a/llvm/test/Transforms/InstCombine/phi.ll b/llvm/test/Transforms/InstCombine/phi.ll
index 1c8b21ceb029b..a97fa90715941 100644
--- a/llvm/test/Transforms/InstCombine/phi.ll
+++ b/llvm/test/Transforms/InstCombine/phi.ll
@@ -1750,7 +1750,8 @@ for.end:
define i1 @pr57488_icmp_of_phi(ptr %ptr.base, i64 %len) {
; CHECK-LABEL: @pr57488_icmp_of_phi(
; CHECK-NEXT: start:
-; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds i64, ptr [[PTR_BASE:%.*]], i64 [[LEN:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[LEN:%.*]], 3
+; CHECK-NEXT: [[END:%.*]] = getelementptr inbounds i8, ptr [[PTR_BASE:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[LEN_ZERO:%.*]] = icmp eq i64 [[LEN]], 0
; CHECK-NEXT: br i1 [[LEN_ZERO]], label [[EXIT:%.*]], label [[LOOP:%.*]]
; CHECK: loop:
diff --git a/llvm/test/Transforms/InstCombine/pr38984-inseltpoison.ll b/llvm/test/Transforms/InstCombine/pr38984-inseltpoison.ll
index a9f310b17e6b6..bd5a751c717d3 100644
--- a/llvm/test/Transforms/InstCombine/pr38984-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/pr38984-inseltpoison.ll
@@ -9,7 +9,12 @@ target datalayout = "p:16:16"
define <4 x i1> @PR38984_1() {
; CHECK-LABEL: @PR38984_1(
; CHECK-NEXT: entry:
-; CHECK-NEXT: ret <4 x i1> splat (i1 true)
+; CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr @offsets, align 1
+; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i16> poison, i16 [[TMP0]], i64 3
+; CHECK-NEXT: [[TMP2:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 2)
+; CHECK-NEXT: [[TMP3:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 2)
+; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x i16> [[TMP2]], [[TMP3]]
+; CHECK-NEXT: ret <4 x i1> [[TMP4]]
;
entry:
%0 = load i16, ptr @offsets, align 1
@@ -26,8 +31,10 @@ define <4 x i1> @PR38984_2() {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr @offsets, align 2
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i16> poison, i16 [[TMP0]], i64 3
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i16, ptr getelementptr inbounds nuw (i8, ptr @a, i16 42), <4 x i16> [[TMP1]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i16, ptr null, <4 x i16> [[TMP1]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 1)
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr getelementptr inbounds nuw (i8, ptr @a, i16 42), <4 x i16> [[TMP5]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 1)
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr null, <4 x i16> [[TMP6]]
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x ptr> [[TMP2]], [[TMP3]]
; CHECK-NEXT: ret <4 x i1> [[TMP4]]
;
diff --git a/llvm/test/Transforms/InstCombine/pr38984.ll b/llvm/test/Transforms/InstCombine/pr38984.ll
index 19fa3346797fb..ebca194dc132d 100644
--- a/llvm/test/Transforms/InstCombine/pr38984.ll
+++ b/llvm/test/Transforms/InstCombine/pr38984.ll
@@ -9,7 +9,12 @@ target datalayout = "p:16:16"
define <4 x i1> @PR38984_1() {
; CHECK-LABEL: @PR38984_1(
; CHECK-NEXT: entry:
-; CHECK-NEXT: ret <4 x i1> splat (i1 true)
+; CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr @offsets, align 1
+; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i16> <i16 undef, i16 undef, i16 undef, i16 poison>, i16 [[TMP0]], i64 3
+; CHECK-NEXT: [[TMP2:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 2)
+; CHECK-NEXT: [[TMP3:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 2)
+; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x i16> [[TMP2]], [[TMP3]]
+; CHECK-NEXT: ret <4 x i1> [[TMP4]]
;
entry:
%0 = load i16, ptr @offsets, align 1
@@ -26,8 +31,10 @@ define <4 x i1> @PR38984_2() {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr @offsets, align 2
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i16> <i16 undef, i16 undef, i16 undef, i16 poison>, i16 [[TMP0]], i64 3
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i16, ptr getelementptr inbounds nuw (i8, ptr @a, i16 42), <4 x i16> [[TMP1]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i16, ptr null, <4 x i16> [[TMP1]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 1)
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr getelementptr inbounds nuw (i8, ptr @a, i16 42), <4 x i16> [[TMP5]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl <4 x i16> [[TMP1]], splat (i16 1)
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr null, <4 x i16> [[TMP6]]
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x ptr> [[TMP2]], [[TMP3]]
; CHECK-NEXT: ret <4 x i1> [[TMP4]]
;
diff --git a/llvm/test/Transforms/InstCombine/pr58901.ll b/llvm/test/Transforms/InstCombine/pr58901.ll
index aa5aad23b082d..0c4213d1f73c0 100644
--- a/llvm/test/Transforms/InstCombine/pr58901.ll
+++ b/llvm/test/Transforms/InstCombine/pr58901.ll
@@ -4,7 +4,8 @@
define ptr @f1(ptr %arg, i64 %arg1) {
; CHECK-LABEL: @f1(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[ARG:%.*]], i64 72
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[ARG1:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl i64 [[ARG1:%.*]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP1]], i64 [[TMP3]]
; CHECK-NEXT: ret ptr [[TMP2]]
;
%1 = getelementptr [6 x i32], ptr %arg, i64 3
@@ -15,8 +16,10 @@ define ptr @f1(ptr %arg, i64 %arg1) {
define ptr @f2(ptr %arg, i64 %arg1) {
; CHECK-LABEL: @f2(
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[ARG:%.*]], i64 72
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr [6 x i32], ptr [[TMP1]], i64 [[ARG1:%.*]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[DOTSPLIT]], i64 [[ARG1]]
+; CHECK-NEXT: [[TMP4:%.*]] = mul i64 [[ARG1:%.*]], 24
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr i8, ptr [[TMP1]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl i64 [[ARG1]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[DOTSPLIT]], i64 [[TMP3]]
; CHECK-NEXT: ret ptr [[TMP2]]
;
%1 = getelementptr [6 x i32], ptr %arg, i64 3
diff --git a/llvm/test/Transforms/InstCombine/ptrtoint-nullgep.ll b/llvm/test/Transforms/InstCombine/ptrtoint-nullgep.ll
index bf978801fec5d..e6ebff21bc808 100644
--- a/llvm/test/Transforms/InstCombine/ptrtoint-nullgep.ll
+++ b/llvm/test/Transforms/InstCombine/ptrtoint-nullgep.ll
@@ -333,8 +333,8 @@ define i64 @fold_ptrtoint_nullgep_i32_variable(i64 %val) {
;
; INSTCOMBINE-LABEL: define {{[^@]+}}@fold_ptrtoint_nullgep_i32_variable
; INSTCOMBINE-SAME: (i64 [[VAL:%.*]]) {
-; INSTCOMBINE-NEXT: [[PTR_IDX:%.*]] = shl i64 [[VAL]], 2
-; INSTCOMBINE-NEXT: ret i64 [[PTR_IDX]]
+; INSTCOMBINE-NEXT: [[TMP1:%.*]] = shl i64 [[VAL]], 2
+; INSTCOMBINE-NEXT: ret i64 [[TMP1]]
;
%ptr = getelementptr i32, ptr addrspace(1) null, i64 %val
%ret = ptrtoint ptr addrspace(1) %ptr to i64
@@ -500,8 +500,8 @@ define i64 @fold_ptrtoint_nullgep_array_one_var_1(i64 %x) {
;
; INSTCOMBINE-LABEL: define {{[^@]+}}@fold_ptrtoint_nullgep_array_one_var_1
; INSTCOMBINE-SAME: (i64 [[X:%.*]]) {
-; INSTCOMBINE-NEXT: [[PTR_SPLIT_IDX:%.*]] = shl i64 [[X]], 2
-; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[PTR_SPLIT_IDX]], 6
+; INSTCOMBINE-NEXT: [[TMP1:%.*]] = shl i64 [[X]], 2
+; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[TMP1]], 6
; INSTCOMBINE-NEXT: ret i64 [[RET]]
;
%ptr = getelementptr [2 x i16], ptr addrspace(1) null, i64 %x, i64 3
@@ -524,8 +524,8 @@ define i64 @fold_ptrtoint_nullgep_array_one_var_2(i64 %x) {
;
; INSTCOMBINE-LABEL: define {{[^@]+}}@fold_ptrtoint_nullgep_array_one_var_2
; INSTCOMBINE-SAME: (i64 [[X:%.*]]) {
-; INSTCOMBINE-NEXT: [[PTR_IDX:%.*]] = shl i64 [[X]], 1
-; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[PTR_IDX]], 28
+; INSTCOMBINE-NEXT: [[TMP1:%.*]] = shl i64 [[X]], 1
+; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[TMP1]], 28
; INSTCOMBINE-NEXT: ret i64 [[RET]]
;
%ptr = getelementptr [2 x i16], ptr addrspace(1) null, i64 7, i64 %x
@@ -548,9 +548,9 @@ define i64 @fold_ptrtoint_nested_array_two_vars(i64 %x, i64 %y) {
;
; INSTCOMBINE-LABEL: define {{[^@]+}}@fold_ptrtoint_nested_array_two_vars
; INSTCOMBINE-SAME: (i64 [[X:%.*]], i64 [[Y:%.*]]) {
-; INSTCOMBINE-NEXT: [[PTR_SPLIT_IDX:%.*]] = shl i64 [[X]], 2
-; INSTCOMBINE-NEXT: [[PTR_IDX:%.*]] = shl i64 [[Y]], 1
-; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[PTR_SPLIT_IDX]], [[PTR_IDX]]
+; INSTCOMBINE-NEXT: [[TMP1:%.*]] = shl i64 [[X]], 2
+; INSTCOMBINE-NEXT: [[TMP2:%.*]] = shl i64 [[Y]], 1
+; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[TMP1]], [[TMP2]]
; INSTCOMBINE-NEXT: ret i64 [[RET]]
;
@@ -574,9 +574,9 @@ define i64 @fold_ptrtoint_nested_array_two_vars_plus_zero(i64 %x, i64 %y) {
;
; INSTCOMBINE-LABEL: define {{[^@]+}}@fold_ptrtoint_nested_array_two_vars_plus_zero
; INSTCOMBINE-SAME: (i64 [[X:%.*]], i64 [[Y:%.*]]) {
-; INSTCOMBINE-NEXT: [[PTR_SPLIT_IDX:%.*]] = shl i64 [[X]], 3
-; INSTCOMBINE-NEXT: [[PTR_IDX:%.*]] = shl i64 [[Y]], 2
-; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[PTR_SPLIT_IDX]], [[PTR_IDX]]
+; INSTCOMBINE-NEXT: [[TMP1:%.*]] = shl i64 [[X]], 3
+; INSTCOMBINE-NEXT: [[TMP2:%.*]] = shl i64 [[Y]], 2
+; INSTCOMBINE-NEXT: [[RET:%.*]] = add i64 [[TMP1]], [[TMP2]]
; INSTCOMBINE-NEXT: ret i64 [[RET]]
;
%ptr = getelementptr [2 x [2 x i16]], ptr addrspace(1) null, i64 %x, i64 %y, i64 0
@@ -599,10 +599,10 @@ define i64 @fold_ptrtoint_nested_array_two_vars_plus_const(i64 %x, i64 %y) {
;
; INSTCOMBINE-LABEL: define {{[^@]+}}@fold_ptrtoint_nested_array_two_vars_plus_const
; INSTCOMBINE-SAME: (i64 [[X:%.*]], i64 [[Y:%.*]]) {
-; INSTCOMBINE-NEXT: [[PTR_SPLIT_IDX:%.*]] = shl i64 [[X]], 3
-; INSTCOMBINE-NEXT: [[PTR_SPLIT1_IDX:%.*]] = shl i64 [[Y]], 2
-; INSTCOMBINE-NEXT: [[TMP1:%.*]] = add i64 [[PTR_SPLIT_IDX]], [[PTR_SPLIT1_IDX]]
-; INSTCOMBINE-NEXT: [[RET:%.*]] = or disjoint i64 [[TMP1]], 2
+; INSTCOMBINE-NEXT: [[TMP1:%.*]] = shl i64 [[X]], 3
+; INSTCOMBINE-NEXT: [[TMP2:%.*]] = shl i64 [[Y]], 2
+; INSTCOMBINE-NEXT: [[TMP3:%.*]] = add i64 [[TMP1]], [[TMP2]]
+; INSTCOMBINE-NEXT: [[RET:%.*]] = or disjoint i64 [[TMP3]], 2
; INSTCOMBINE-NEXT: ret i64 [[RET]]
;
%ptr = getelementptr [2 x [2 x i16]], ptr addrspace(1) null, i64 %x, i64 %y, i64 1
@@ -628,8 +628,10 @@ define i64 @fold_ptrtoint_nested_nullgep_array_variable_multiple_uses(i64 %x, i6
;
; INSTCOMBINE-LABEL: define {{[^@]+}}@fold_ptrtoint_nested_nullgep_array_variable_multiple_uses
; INSTCOMBINE-SAME: (i64 [[X:%.*]], i64 [[Y:%.*]]) {
-; INSTCOMBINE-NEXT: [[PTR_SPLIT:%.*]] = getelementptr [2 x i16], ptr addrspace(1) null, i64 [[X]]
-; INSTCOMBINE-NEXT: [[PTR:%.*]] = getelementptr i16, ptr addrspace(1) [[PTR_SPLIT]], i64 [[Y]]
+; INSTCOMBINE-NEXT: [[TMP1:%.*]] = shl i64 [[X]], 2
+; INSTCOMBINE-NEXT: [[PTR_SPLIT:%.*]] = getelementptr i8, ptr addrspace(1) null, i64 [[TMP1]]
+; INSTCOMBINE-NEXT: [[TMP2:%.*]] = shl i64 [[Y]], 1
+; INSTCOMBINE-NEXT: [[PTR:%.*]] = getelementptr i8, ptr addrspace(1) [[PTR_SPLIT]], i64 [[TMP2]]
; INSTCOMBINE-NEXT: call void @use_ptr(ptr addrspace(1) [[PTR]])
; INSTCOMBINE-NEXT: [[RET:%.*]] = ptrtoint ptr addrspace(1) [[PTR]] to i64
; INSTCOMBINE-NEXT: ret i64 [[RET]]
diff --git a/llvm/test/Transforms/InstCombine/remove-loop-phi-multiply-by-zero.ll b/llvm/test/Transforms/InstCombine/remove-loop-phi-multiply-by-zero.ll
index a4a9de7fd1285..31c5a3e0265f5 100644
--- a/llvm/test/Transforms/InstCombine/remove-loop-phi-multiply-by-zero.ll
+++ b/llvm/test/Transforms/InstCombine/remove-loop-phi-multiply-by-zero.ll
@@ -67,7 +67,8 @@ define double @test_nnan_flag_enabled(ptr %arr_d) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[MUL]] = fmul nnan double [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -101,7 +102,8 @@ define double @test_ninf_flag_enabled(ptr %arr_d) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[MUL]] = fmul ninf double [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -135,7 +137,8 @@ define double @test_nsz_flag_enabled(ptr %arr_d) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[MUL]] = fmul nsz double [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -169,7 +172,8 @@ define double @test_phi_initalise_to_non_zero(ptr %arr_d) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi double [ 1.000000e+00, [[ENTRY]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[MUL]] = fmul fast double [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -205,7 +209,8 @@ define double @test_multiple_phi_operands(ptr %arr_d, i1 %entry_cond) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ 0, [[ENTRY_2]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi double [ 0.000000e+00, [[ENTRY]] ], [ 0.000000e+00, [[ENTRY_2]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[MUL]] = fmul fast double [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -244,7 +249,8 @@ define double @test_multiple_phi_operands_with_non_zero(ptr %arr_d, i1 %entry_co
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ 0, [[ENTRY_2]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi double [ 1.000000e+00, [[ENTRY]] ], [ 0.000000e+00, [[ENTRY_2]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[MUL]] = fmul fast double [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -281,7 +287,8 @@ define i32 @test_int_phi_operands(ptr %arr_d) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL]] = mul nsw i32 [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -315,7 +322,8 @@ define i32 @test_int_phi_operands_initalise_to_non_zero(ptr %arr_d) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi i32 [ 1, [[ENTRY]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL]] = mul i32 [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -351,7 +359,8 @@ define i32 @test_multiple_int_phi_operands(ptr %arr_d, i1 %entry_cond) {
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ 0, [[ENTRY_2]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ 0, [[ENTRY_2]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL]] = mul i32 [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
@@ -390,7 +399,8 @@ define i32 @test_multiple_int_phi_operands_initalise_to_non_zero(ptr %arr_d, i1
; CHECK: for.body:
; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ 0, [[ENTRY_2]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[F_PROD_01:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ 1, [[ENTRY_2]] ], [ [[MUL:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARR_D:%.*]], i64 [[I_02]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I_02]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR_D:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL]] = mul i32 [[F_PROD_01]], [[TMP0]]
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_02]], 1
diff --git a/llvm/test/Transforms/InstCombine/scalable-select.ll b/llvm/test/Transforms/InstCombine/scalable-select.ll
index 717507dbfa356..34511a463fb07 100644
--- a/llvm/test/Transforms/InstCombine/scalable-select.ll
+++ b/llvm/test/Transforms/InstCombine/scalable-select.ll
@@ -17,8 +17,10 @@ define <vscale x 1 x i32> @select_opt(<vscale x 1 x i32> %b, <vscale x 1 x i1> %
define <vscale x 2 x i64> @load_scalable_of_selected_ptrs(ptr %p0, ptr %p1, i64 %idx) {
; CHECK-LABEL: @load_scalable_of_selected_ptrs(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[ARRAYIDX0:%.*]] = getelementptr inbounds i32, ptr [[P0:%.*]], i64 [[IDX:%.*]]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[P1:%.*]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[ARRAYIDX0:%.*]] = getelementptr inbounds i8, ptr [[P0:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDX]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[P1:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[L0:%.*]] = load i32, ptr [[ARRAYIDX0]], align 4
; CHECK-NEXT: [[L1:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[L0]], [[L1]]
diff --git a/llvm/test/Transforms/InstCombine/select-gep.ll b/llvm/test/Transforms/InstCombine/select-gep.ll
index 79287fd933cc0..8bd38314d4d3c 100644
--- a/llvm/test/Transforms/InstCombine/select-gep.ll
+++ b/llvm/test/Transforms/InstCombine/select-gep.ll
@@ -60,7 +60,8 @@ define ptr @test1d(ptr %p, ptr %q) {
define ptr @test2(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2(
; CHECK-NEXT: [[SELECT_V:%.*]] = call i64 @llvm.umax.i64(i64 [[X:%.*]], i64 [[Y:%.*]])
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[SELECT_V]]
+; CHECK-NEXT: [[SELECT_V1:%.*]] = shl nsw i64 [[SELECT_V]], 2
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[SELECT_V1]]
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep1 = getelementptr inbounds i32, ptr %p, i64 %x
@@ -73,9 +74,10 @@ define ptr @test2(ptr %p, i64 %x, i64 %y) {
; PR50183
define ptr @test2a(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2a(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 [[X]], i64 0
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[SELECT_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
+; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 [[TMP1]], i64 0
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[SELECT_IDX]]
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep = getelementptr inbounds i32, ptr %p, i64 %x
@@ -86,9 +88,10 @@ define ptr @test2a(ptr %p, i64 %x, i64 %y) {
define ptr @test2a_nusw(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2a_nusw(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 [[X]], i64 0
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr nusw i32, ptr [[P:%.*]], i64 [[SELECT_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
+; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 [[TMP1]], i64 0
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr nusw i8, ptr [[P:%.*]], i64 [[SELECT_IDX]]
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep = getelementptr nusw i32, ptr %p, i64 %x
@@ -99,9 +102,10 @@ define ptr @test2a_nusw(ptr %p, i64 %x, i64 %y) {
define ptr @test2a_nuw(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2a_nuw(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 [[X]], i64 0
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr nuw i32, ptr [[P:%.*]], i64 [[SELECT_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
+; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 [[TMP1]], i64 0
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr nuw i8, ptr [[P:%.*]], i64 [[SELECT_IDX]]
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep = getelementptr nuw i32, ptr %p, i64 %x
@@ -113,9 +117,10 @@ define ptr @test2a_nuw(ptr %p, i64 %x, i64 %y) {
; PR50183
define ptr @test2b(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2b(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 0, i64 [[X]]
-; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[SELECT_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
+; CHECK-NEXT: [[SELECT_IDX:%.*]] = select i1 [[CMP]], i64 0, i64 [[TMP1]]
+; CHECK-NEXT: [[SELECT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[SELECT_IDX]]
; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep = getelementptr inbounds i32, ptr %p, i64 %x
@@ -127,7 +132,8 @@ define ptr @test2b(ptr %p, i64 %x, i64 %y) {
; PR51069
define ptr @test2c(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2c(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[ICMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
; CHECK-NEXT: [[SEL_IDX:%.*]] = select i1 [[ICMP]], i64 0, i64 24
; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP1]], i64 [[SEL_IDX]]
@@ -143,7 +149,8 @@ define ptr @test2c(ptr %p, i64 %x, i64 %y) {
; PR51069
define ptr @test2d(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test2d(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[ICMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
; CHECK-NEXT: [[SEL_IDX:%.*]] = select i1 [[ICMP]], i64 24, i64 0
; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP1]], i64 [[SEL_IDX]]
@@ -162,9 +169,11 @@ define ptr @test2d(ptr %p, i64 %x, i64 %y) {
define ptr @test3a(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test3a(
; CHECK-NEXT: [[GEP1_SPLIT:%.*]] = getelementptr inbounds nuw i8, ptr [[P:%.*]], i64 32
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[GEP1_SPLIT]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[GEP1_SPLIT]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP2_SPLIT:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 32
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i32, ptr [[GEP2_SPLIT]], i64 [[Y:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Y:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[GEP2_SPLIT]], i64 [[TMP2]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y]]
; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], ptr [[GEP1]], ptr [[GEP2]]
; CHECK-NEXT: ret ptr [[SELECT]]
@@ -178,11 +187,15 @@ define ptr @test3a(ptr %p, i64 %x, i64 %y) {
define ptr @test3b(ptr %p, ptr %q, i64 %x, i64 %y) {
; CHECK-LABEL: @test3b(
-; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X:%.*]], [[Y:%.*]]
-; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], ptr [[GEP1:%.*]], ptr [[GEP2:%.*]]
-; CHECK-NEXT: [[SELECT_V:%.*]] = getelementptr inbounds nuw i8, ptr [[SELECT]], i64 32
-; CHECK-NEXT: [[SELECT1:%.*]] = getelementptr inbounds i32, ptr [[SELECT_V]], i64 [[X]]
-; CHECK-NEXT: ret ptr [[SELECT1]]
+; CHECK-NEXT: [[GEP1_SPLIT:%.*]] = getelementptr inbounds nuw i8, ptr [[P:%.*]], i64 32
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[GEP1_SPLIT]], i64 [[TMP1]]
+; CHECK-NEXT: [[GEP2_SPLIT:%.*]] = getelementptr inbounds nuw i8, ptr [[Q:%.*]], i64 32
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[GEP2_SPLIT]], i64 [[TMP2]]
+; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
+; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], ptr [[GEP1]], ptr [[GEP2]]
+; CHECK-NEXT: ret ptr [[SELECT]]
;
%gep1 = getelementptr inbounds [4 x i32], ptr %p, i64 2, i64 %x
%gep2 = getelementptr inbounds [4 x i32], ptr %q, i64 2, i64 %x
@@ -193,9 +206,11 @@ define ptr @test3b(ptr %p, ptr %q, i64 %x, i64 %y) {
define ptr @test3c(ptr %p, ptr %q, i64 %x, i64 %y) {
; CHECK-LABEL: @test3c(
-; CHECK-NEXT: [[GEP1_SPLIT:%.*]] = getelementptr inbounds [4 x i32], ptr [[P:%.*]], i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 4
+; CHECK-NEXT: [[GEP1_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP1_SPLIT]], i64 8
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i32, ptr [[Q:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[Q:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], ptr [[GEP1]], ptr [[GEP2]]
; CHECK-NEXT: ret ptr [[SELECT]]
@@ -209,8 +224,10 @@ define ptr @test3c(ptr %p, ptr %q, i64 %x, i64 %y) {
define ptr @test3d(ptr %p, ptr %q, i64 %x, i64 %y) {
; CHECK-LABEL: @test3d(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[X:%.*]]
-; CHECK-NEXT: [[GEP2_SPLIT:%.*]] = getelementptr inbounds [4 x i32], ptr [[Q:%.*]], i64 [[X]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[X]], 4
+; CHECK-NEXT: [[GEP2_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[Q:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds nuw i8, ptr [[GEP2_SPLIT]], i64 8
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y:%.*]]
; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], ptr [[GEP1]], ptr [[GEP2]]
@@ -227,8 +244,10 @@ define ptr @test3d(ptr %p, ptr %q, i64 %x, i64 %y) {
define ptr @test4(ptr %p, ptr %q, i64 %x, i64 %y) {
; CHECK-LABEL: @test4(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[X:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i32, ptr [[Q:%.*]], i64 [[Y:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Y:%.*]], 2
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[Q:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[CMP:%.*]] = icmp ugt i64 [[X]], [[Y]]
; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[CMP]], ptr [[GEP1]], ptr [[GEP2]]
; CHECK-NEXT: ret ptr [[SELECT]]
@@ -244,8 +263,10 @@ define ptr @test4(ptr %p, ptr %q, i64 %x, i64 %y) {
define <2 x ptr> @test5(ptr %p1, ptr %p2, <2 x i64> %idx, <2 x i1> %cc) {
; CHECK-LABEL: @test5(
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i64, ptr [[P1:%.*]], <2 x i64> [[IDX:%.*]]
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i64, ptr [[P2:%.*]], <2 x i64> [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <2 x i64> [[IDX:%.*]], splat (i64 3)
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[P1:%.*]], <2 x i64> [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl <2 x i64> [[IDX]], splat (i64 3)
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i8, ptr [[P2:%.*]], <2 x i64> [[TMP2]]
; CHECK-NEXT: [[SELECT:%.*]] = select <2 x i1> [[CC:%.*]], <2 x ptr> [[GEP1]], <2 x ptr> [[GEP2]]
; CHECK-NEXT: ret <2 x ptr> [[SELECT]]
;
@@ -258,10 +279,12 @@ define <2 x ptr> @test5(ptr %p1, ptr %p2, <2 x i64> %idx, <2 x i1> %cc) {
; PR51069 - multiple uses
define ptr @test6(ptr %p, i64 %x, i64 %y) {
; CHECK-LABEL: @test6(
-; 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 [[Y]], i64 0
-; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds i32, ptr [[GEP1]], i64 [[SEL_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[Y:%.*]], 2
+; CHECK-NEXT: [[ICMP:%.*]] = icmp ugt i64 [[X]], [[Y]]
+; CHECK-NEXT: [[SEL_IDX:%.*]] = select i1 [[ICMP]], i64 [[TMP2]], i64 0
+; CHECK-NEXT: [[SEL:%.*]] = getelementptr inbounds i8, ptr [[GEP1]], i64 [[SEL_IDX]]
; CHECK-NEXT: call void @use_i32p(ptr [[GEP1]])
; CHECK-NEXT: ret ptr [[SEL]]
;
@@ -278,7 +301,8 @@ declare void @use_i32p(ptr)
define <2 x ptr> @test7(<2 x ptr> %p1, i64 %idx, <2 x i1> %cc) {
; CHECK-LABEL: @test7(
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i64, <2 x ptr> [[P1:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 3
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, <2 x ptr> [[P1:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[SELECT:%.*]] = select <2 x i1> [[CC:%.*]], <2 x ptr> [[P1]], <2 x ptr> [[GEP]]
; CHECK-NEXT: ret <2 x ptr> [[SELECT]]
;
diff --git a/llvm/test/Transforms/InstCombine/shift.ll b/llvm/test/Transforms/InstCombine/shift.ll
index 2ff3c3cd9d990..50ff9858cfaa7 100644
--- a/llvm/test/Transforms/InstCombine/shift.ll
+++ b/llvm/test/Transforms/InstCombine/shift.ll
@@ -1692,7 +1692,8 @@ define i177 @lshr_out_of_range(i177 %Y, ptr %A2, ptr %ptr) {
; CHECK-NEXT: [[B4:%.*]] = sext i1 [[TMP1]] to i177
; CHECK-NEXT: [[C8:%.*]] = icmp ugt i177 [[Y]], [[B4]]
; CHECK-NEXT: [[TMP2:%.*]] = sext i1 [[C8]] to i64
-; CHECK-NEXT: [[G18:%.*]] = getelementptr ptr, ptr [[A2:%.*]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[G18:%.*]] = getelementptr i8, ptr [[A2:%.*]], i64 [[TMP3]]
; CHECK-NEXT: store ptr [[G18]], ptr [[PTR:%.*]], align 8
; CHECK-NEXT: ret i177 0
;
@@ -1716,7 +1717,8 @@ define i177 @lshr_out_of_range2(i177 %Y, ptr %A2, ptr %ptr) {
; CHECK-LABEL: @lshr_out_of_range2(
; CHECK-NEXT: [[C8:%.*]] = icmp ne i177 [[Y:%.*]], 0
; CHECK-NEXT: [[TMP1:%.*]] = sext i1 [[C8]] to i64
-; CHECK-NEXT: [[G18:%.*]] = getelementptr ptr, ptr [[A2:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 3
+; CHECK-NEXT: [[G18:%.*]] = getelementptr i8, ptr [[A2:%.*]], i64 [[TMP2]]
; CHECK-NEXT: store ptr [[G18]], ptr [[PTR:%.*]], align 8
; CHECK-NEXT: ret i177 0
;
@@ -1742,14 +1744,14 @@ define void @ashr_out_of_range(ptr %A) "instcombine-no-verify-fixpoint" {
; CHECK-LABEL: @ashr_out_of_range(
; CHECK-NEXT: [[L:%.*]] = load i177, ptr [[A:%.*]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i177 [[L]], -1
-; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[TMP1]], i64 -1, i64 -2
-; CHECK-NEXT: [[G11:%.*]] = getelementptr i177, ptr [[A]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[TMP1]], i64 -24, i64 -48
+; CHECK-NEXT: [[G11:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: [[L7:%.*]] = load i177, ptr [[G11]], align 4
; CHECK-NEXT: [[L7_FROZEN:%.*]] = freeze i177 [[L7]]
; CHECK-NEXT: [[C171:%.*]] = icmp slt i177 [[L7_FROZEN]], 0
; CHECK-NEXT: [[C17:%.*]] = select i1 [[TMP1]], i1 [[C171]], i1 false
-; CHECK-NEXT: [[TMP3:%.*]] = sext i1 [[C17]] to i64
-; CHECK-NEXT: [[G62:%.*]] = getelementptr i177, ptr [[G11]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP3:%.*]] = select i1 [[C17]], i64 -24, i64 0
+; CHECK-NEXT: [[G62:%.*]] = getelementptr i8, ptr [[G11]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i177 [[L7_FROZEN]], -1
; CHECK-NEXT: [[B28:%.*]] = select i1 [[TMP4]], i177 0, i177 [[L7_FROZEN]]
; CHECK-NEXT: store i177 [[B28]], ptr [[G62]], align 4
@@ -1779,11 +1781,11 @@ define void @ashr_out_of_range_1(ptr %A) {
; CHECK-NEXT: [[L_FROZEN:%.*]] = freeze i177 [[L]]
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq i177 [[L_FROZEN]], -1
; CHECK-NEXT: [[TMP2:%.*]] = trunc i177 [[L_FROZEN]] to i64
-; CHECK-NEXT: [[TMP3:%.*]] = select i1 [[TMP1]], i64 0, i64 [[TMP2]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i177, ptr [[A]], i64 [[TMP3]]
-; CHECK-NEXT: [[G11:%.*]] = getelementptr i8, ptr [[TMP4]], i64 -24
-; CHECK-NEXT: [[TMP5:%.*]] = sext i1 [[TMP1]] to i64
-; CHECK-NEXT: [[G62:%.*]] = getelementptr i177, ptr [[G11]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP3:%.*]] = mul i64 [[TMP2]], 24
+; CHECK-NEXT: [[TMP4:%.*]] = select i1 [[TMP1]], i64 0, i64 [[TMP3]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP7:%.*]] = select i1 [[TMP1]], i64 -48, i64 -24
+; CHECK-NEXT: [[G62:%.*]] = getelementptr i8, ptr [[TMP5]], i64 [[TMP7]]
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i177 [[L_FROZEN]], -1
; CHECK-NEXT: [[B28:%.*]] = select i1 [[TMP6]], i177 0, i177 [[L_FROZEN]]
; CHECK-NEXT: store i177 [[B28]], ptr [[G62]], align 4
diff --git a/llvm/test/Transforms/InstCombine/sink_instruction.ll b/llvm/test/Transforms/InstCombine/sink_instruction.ll
index cb9a3069ca5fd..7380354c05a23 100644
--- a/llvm/test/Transforms/InstCombine/sink_instruction.ll
+++ b/llvm/test/Transforms/InstCombine/sink_instruction.ll
@@ -90,7 +90,8 @@ define i32 @test3(ptr nocapture readonly %P, i32 %i) {
; CHECK-NEXT: ]
; CHECK: sw.bb:
; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[I]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[I]]
; CHECK-NEXT: br label [[SW_EPILOG]]
@@ -148,7 +149,8 @@ define i32 @test5(ptr nocapture readonly %P, i32 %i, i1 %cond) {
; CHECK-LABEL: @test5(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[I:%.*]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: br i1 [[COND:%.*]], label [[DISPATCHBB:%.*]], label [[SW_EPILOG:%.*]]
; CHECK: dispatchBB:
@@ -186,7 +188,8 @@ define i32 @test6(ptr nocapture readonly %P, i32 %i, i1 %cond) {
; CHECK-NEXT: br label [[DISPATCHBB:%.*]]
; CHECK: dispatchBB:
; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[I]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[P:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: switch i32 [[I]], label [[SW_BB:%.*]] [
; CHECK-NEXT: i32 5, label [[SW_EPILOG:%.*]]
diff --git a/llvm/test/Transforms/InstCombine/store.ll b/llvm/test/Transforms/InstCombine/store.ll
index daa40da1828b5..fbbde49ae092a 100644
--- a/llvm/test/Transforms/InstCombine/store.ll
+++ b/llvm/test/Transforms/InstCombine/store.ll
@@ -49,7 +49,8 @@ define void @test2(ptr %P) {
define void @store_at_gep_off_null_inbounds(i64 %offset) {
; CHECK-LABEL: @store_at_gep_off_null_inbounds(
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr null, i64 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[OFFSET:%.*]], 2
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr null, i64 [[TMP1]]
; CHECK-NEXT: store i32 poison, ptr [[PTR]], align 4
; CHECK-NEXT: ret void
;
@@ -60,7 +61,8 @@ define void @store_at_gep_off_null_inbounds(i64 %offset) {
define void @store_at_gep_off_null_not_inbounds(i64 %offset) {
; CHECK-LABEL: @store_at_gep_off_null_not_inbounds(
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr i32, ptr null, i64 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[OFFSET:%.*]], 2
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr i8, ptr null, i64 [[TMP1]]
; CHECK-NEXT: store i32 poison, ptr [[PTR]], align 4
; CHECK-NEXT: ret void
;
@@ -71,7 +73,8 @@ define void @store_at_gep_off_null_not_inbounds(i64 %offset) {
define void @store_at_gep_off_no_null_opt(i64 %offset) #0 {
; CHECK-LABEL: @store_at_gep_off_no_null_opt(
-; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr null, i64 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[OFFSET:%.*]], 2
+; CHECK-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr null, i64 [[TMP1]]
; CHECK-NEXT: store i32 24, ptr [[PTR]], align 4
; CHECK-NEXT: ret void
;
@@ -170,7 +173,8 @@ define void @test6(i32 %n, ptr %a, ptr %gi) nounwind uwtable ssp {
; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[STOREMERGE]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: store float 0.000000e+00, ptr [[ARRAYIDX]], align 4, !tbaa [[TBAA4:![0-9]+]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[GI]], align 4, !tbaa [[TBAA0]]
; CHECK-NEXT: [[INC]] = add nsw i32 [[TMP0]], 1
diff --git a/llvm/test/Transforms/InstCombine/strcmp-3.ll b/llvm/test/Transforms/InstCombine/strcmp-3.ll
index 72da736a0a9fd..89da034f5cc5f 100644
--- a/llvm/test/Transforms/InstCombine/strcmp-3.ll
+++ b/llvm/test/Transforms/InstCombine/strcmp-3.ll
@@ -25,7 +25,8 @@ define i32 @fold_strcmp_a5i0_a5i1_to_0() {
define i32 @call_strcmp_a5i0_a5iI(i64 %I) {
; CHECK-LABEL: @call_strcmp_a5i0_a5iI(
-; CHECK-NEXT: [[Q:%.*]] = getelementptr [4 x i8], ptr @a5, i64 [[I:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[I:%.*]], 2
+; CHECK-NEXT: [[Q:%.*]] = getelementptr i8, ptr @a5, i64 [[TMP1]]
; CHECK-NEXT: [[CMP:%.*]] = call i32 @strcmp(ptr noundef nonnull dereferenceable(4) @a5, ptr noundef nonnull dereferenceable(1) [[Q]])
; CHECK-NEXT: ret i32 [[CMP]]
;
@@ -40,7 +41,8 @@ define i32 @call_strcmp_a5i0_a5iI(i64 %I) {
define i32 @call_strcmp_a5iI_a5i0(i64 %I) {
; CHECK-LABEL: @call_strcmp_a5iI_a5i0(
-; CHECK-NEXT: [[P:%.*]] = getelementptr [4 x i8], ptr @a5, i64 [[I:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[I:%.*]], 2
+; CHECK-NEXT: [[P:%.*]] = getelementptr i8, ptr @a5, i64 [[TMP1]]
; CHECK-NEXT: [[CMP:%.*]] = call i32 @strcmp(ptr noundef nonnull dereferenceable(1) [[P]], ptr noundef nonnull dereferenceable(4) @a5)
; CHECK-NEXT: ret i32 [[CMP]]
;
diff --git a/llvm/test/Transforms/InstCombine/sub-gep.ll b/llvm/test/Transforms/InstCombine/sub-gep.ll
index 01da63fa5b0af..2747521733449 100644
--- a/llvm/test/Transforms/InstCombine/sub-gep.ll
+++ b/llvm/test/Transforms/InstCombine/sub-gep.ll
@@ -329,7 +329,8 @@ define i64 @test25(ptr %P, i64 %A){
define i64 @zext_ptrtoint_sub_ptrtoint(ptr %p, i32 %offset) {
; CHECK-LABEL: @zext_ptrtoint_sub_ptrtoint(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = getelementptr bfloat, ptr @Arr, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr @Arr, i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64
; CHECK-NEXT: [[C:%.*]] = and i64 [[TMP2]], 4294967294
; CHECK-NEXT: [[D:%.*]] = sub i64 [[C]], ptrtoint (ptr @Arr to i64)
@@ -345,7 +346,8 @@ define i64 @zext_ptrtoint_sub_ptrtoint(ptr %p, i32 %offset) {
define i64 @ptrtoint_sub_zext_ptrtoint(ptr %p, i32 %offset) {
; CHECK-LABEL: @ptrtoint_sub_zext_ptrtoint(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = getelementptr bfloat, ptr @Arr, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr @Arr, i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64
; CHECK-NEXT: [[C:%.*]] = and i64 [[TMP2]], 4294967294
; CHECK-NEXT: [[D:%.*]] = sub i64 ptrtoint (ptr @Arr to i64), [[C]]
@@ -361,7 +363,8 @@ define i64 @ptrtoint_sub_zext_ptrtoint(ptr %p, i32 %offset) {
define i64 @negative_zext_ptrtoint_sub_ptrtoint(ptr %p, i32 %offset) {
; CHECK-LABEL: @negative_zext_ptrtoint_sub_ptrtoint(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = getelementptr bfloat, ptr @Arr, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr @Arr, i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64
; CHECK-NEXT: [[C:%.*]] = and i64 [[TMP2]], 65534
; CHECK-NEXT: [[D:%.*]] = sub i64 [[C]], ptrtoint (ptr @Arr to i64)
@@ -377,7 +380,8 @@ define i64 @negative_zext_ptrtoint_sub_ptrtoint(ptr %p, i32 %offset) {
define i64 @negative_ptrtoint_sub_zext_ptrtoint(ptr %p, i32 %offset) {
; CHECK-LABEL: @negative_ptrtoint_sub_zext_ptrtoint(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET:%.*]] to i64
-; CHECK-NEXT: [[A:%.*]] = getelementptr bfloat, ptr @Arr, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr i8, ptr @Arr, i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64
; CHECK-NEXT: [[C:%.*]] = and i64 [[TMP2]], 65534
; CHECK-NEXT: [[D:%.*]] = sub i64 ptrtoint (ptr @Arr to i64), [[C]]
@@ -409,7 +413,8 @@ define i16 @test25_as1(ptr addrspace(1) %P, i64 %A) {
define i64 @ptrtoint_sub_zext_ptrtoint_as2_inbounds(i32 %offset) {
; CHECK-LABEL: @ptrtoint_sub_zext_ptrtoint_as2_inbounds(
-; CHECK-NEXT: [[A:%.*]] = getelementptr inbounds bfloat, ptr addrspace(2) @Arr_as2, i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[OFFSET:%.*]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr inbounds i8, ptr addrspace(2) @Arr_as2, i32 [[TMP1]]
; CHECK-NEXT: [[B:%.*]] = ptrtoint ptr addrspace(2) [[A]] to i32
; CHECK-NEXT: [[C:%.*]] = zext i32 [[B]] to i64
; CHECK-NEXT: [[D:%.*]] = sub nsw i64 ptrtoint (ptr addrspace(2) @Arr_as2 to i64), [[C]]
@@ -491,7 +496,8 @@ define i64 @zext_ptrtoint_sub_zext_ptrtoint_as2_nuw(i32 %offset) {
define i64 @negative_zext_ptrtoint_sub_ptrtoint_as2_nuw(i32 %offset) {
; CHECK-LABEL: @negative_zext_ptrtoint_sub_ptrtoint_as2_nuw(
-; CHECK-NEXT: [[A:%.*]] = getelementptr nuw bfloat, ptr addrspace(2) @Arr_as2, i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw i32 [[OFFSET:%.*]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr nuw i8, ptr addrspace(2) @Arr_as2, i32 [[TMP2]]
; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr addrspace(2) [[A]] to i32
; CHECK-NEXT: [[B_MASK:%.*]] = and i32 [[TMP1]], 65534
; CHECK-NEXT: [[C:%.*]] = zext nneg i32 [[B_MASK]] to i64
@@ -507,7 +513,8 @@ define i64 @negative_zext_ptrtoint_sub_ptrtoint_as2_nuw(i32 %offset) {
define i64 @negative_zext_ptrtoint_sub_zext_ptrtoint_as2_nuw_truncating(i32 %offset) {
; CHECK-LABEL: @negative_zext_ptrtoint_sub_zext_ptrtoint_as2_nuw_truncating(
-; CHECK-NEXT: [[A:%.*]] = getelementptr nuw bfloat, ptr addrspace(2) @Arr_as2, i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw i32 [[OFFSET:%.*]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr nuw i8, ptr addrspace(2) @Arr_as2, i32 [[TMP1]]
; CHECK-NEXT: [[A_IDX:%.*]] = ptrtoint ptr addrspace(2) [[A]] to i32
; CHECK-NEXT: [[E:%.*]] = zext i32 [[A_IDX]] to i64
; CHECK-NEXT: [[D:%.*]] = zext i16 ptrtoint (ptr addrspace(2) @Arr_as2 to i16) to i64
@@ -524,7 +531,8 @@ define i64 @negative_zext_ptrtoint_sub_zext_ptrtoint_as2_nuw_truncating(i32 %off
define i64 @ptrtoint_sub_zext_ptrtoint_as2_inbounds_local(ptr addrspace(2) %p, i32 %offset) {
; CHECK-LABEL: @ptrtoint_sub_zext_ptrtoint_as2_inbounds_local(
-; CHECK-NEXT: [[A:%.*]] = getelementptr inbounds bfloat, ptr addrspace(2) [[P:%.*]], i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[OFFSET:%.*]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr inbounds i8, ptr addrspace(2) [[P:%.*]], i32 [[TMP2]]
; CHECK-NEXT: [[B:%.*]] = ptrtoint ptr addrspace(2) [[A]] to i32
; CHECK-NEXT: [[C:%.*]] = zext i32 [[B]] to i64
; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr addrspace(2) [[P]] to i32
@@ -614,7 +622,8 @@ define i64 @zext_ptrtoint_sub_zext_ptrtoint_as2_nuw_local(ptr addrspace(2) %p, i
define i64 @negative_zext_ptrtoint_sub_ptrtoint_as2_nuw_local(ptr addrspace(2) %p, i32 %offset) {
; CHECK-LABEL: @negative_zext_ptrtoint_sub_ptrtoint_as2_nuw_local(
-; CHECK-NEXT: [[A:%.*]] = getelementptr nuw bfloat, ptr addrspace(2) [[P:%.*]], i32 [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw i32 [[OFFSET:%.*]], 1
+; CHECK-NEXT: [[A:%.*]] = getelementptr nuw i8, ptr addrspace(2) [[P:%.*]], i32 [[TMP3]]
; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr addrspace(2) [[A]] to i32
; CHECK-NEXT: [[B_MASK:%.*]] = and i32 [[TMP1]], 65535
; CHECK-NEXT: [[C:%.*]] = zext nneg i32 [[B_MASK]] to i64
@@ -955,7 +964,8 @@ declare void @use(ptr)
define i64 @multiple_geps_two_chains_gep_base(ptr %base, i64 %base.idx, i64 %idx, i64 %idx2, i64 %idx3) {
; CHECK-LABEL: @multiple_geps_two_chains_gep_base(
-; CHECK-NEXT: [[GEP_BASE:%.*]] = getelementptr inbounds i32, ptr [[BASE:%.*]], i64 [[BASE_IDX:%.*]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[BASE_IDX:%.*]], 2
+; CHECK-NEXT: [[GEP_BASE:%.*]] = getelementptr inbounds i8, ptr [[BASE:%.*]], i64 [[TMP2]]
; CHECK-NEXT: call void @use(ptr [[GEP_BASE]])
; CHECK-NEXT: [[P2_IDX1:%.*]] = add i64 [[IDX:%.*]], [[IDX2:%.*]]
; CHECK-NEXT: [[TMP1:%.*]] = sub i64 [[P2_IDX1]], [[IDX3:%.*]]
diff --git a/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll b/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll
index 00cc06f74d544..d63be4517fd1d 100644
--- a/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/vec_demanded_elts-inseltpoison.ll
@@ -501,8 +501,9 @@ define <3 x float> @shuf_frem_const_op1(<3 x float> %x) {
define ptr @gep_vbase_w_s_idx(<2 x ptr> %base, i64 %index) {
; CHECK-LABEL: @gep_vbase_w_s_idx(
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX:%.*]], 2
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x ptr> [[BASE:%.*]], i64 1
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[INDEX:%.*]]
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[TMP1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[EE]]
;
%gep = getelementptr i32, <2 x ptr> %base, i64 %index
@@ -538,7 +539,8 @@ define ptr @gep_splat_base_w_cv_idx(ptr %base) {
define ptr @gep_splat_base_w_vidx(ptr %base, <2 x i64> %idxvec) {
; CHECK-LABEL: @gep_splat_base_w_vidx(
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x i64> [[IDXVEC:%.*]], i64 1
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[TMP1]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[EE]]
;
%basevec1 = insertelement <2 x ptr> poison, ptr %base, i32 0
@@ -553,7 +555,8 @@ define ptr @gep_splat_base_w_vidx(ptr %base, <2 x i64> %idxvec) {
define ptr @gep_cvbase_w_s_idx(<2 x ptr> %base, i64 %raw_addr) {
; CHECK-LABEL: @gep_cvbase_w_s_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr @GLOBAL, i64 [[RAW_ADDR:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[RAW_ADDR:%.*]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr @GLOBAL, i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[EE]]
;
%gep = getelementptr i32, <2 x ptr> <ptr @GLOBAL, ptr @GLOBAL>, i64 %raw_addr
@@ -583,7 +586,8 @@ define ptr @gep_sbase_w_cv_idx(ptr %base) {
define ptr @gep_sbase_w_splat_idx(ptr %base, i64 %idx) {
; CHECK-LABEL: @gep_sbase_w_splat_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[EE]]
;
%idxvec1 = insertelement <2 x i64> poison, i64 %idx, i32 0
@@ -594,7 +598,8 @@ define ptr @gep_sbase_w_splat_idx(ptr %base, i64 %idx) {
}
define ptr @gep_splat_both(ptr %base, i64 %idx) {
; CHECK-LABEL: @gep_splat_both(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[EE]]
;
%basevec1 = insertelement <2 x ptr> poison, ptr %base, i32 0
@@ -610,7 +615,8 @@ define <2 x ptr> @gep_all_lanes_undef(ptr %base, i64 %idx) {;
; CHECK-LABEL: @gep_all_lanes_undef(
; CHECK-NEXT: [[BASEVEC:%.*]] = insertelement <2 x ptr> poison, ptr [[BASE:%.*]], i64 0
; CHECK-NEXT: [[IDXVEC:%.*]] = insertelement <2 x i64> poison, i64 [[IDX:%.*]], i64 1
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, <2 x ptr> [[BASEVEC]], <2 x i64> [[IDXVEC]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <2 x i64> [[IDXVEC]], splat (i64 2)
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, <2 x ptr> [[BASEVEC]], <2 x i64> [[TMP1]]
; CHECK-NEXT: ret <2 x ptr> [[GEP]]
;
%basevec = insertelement <2 x ptr> poison, ptr %base, i32 0
@@ -652,7 +658,8 @@ define ptr @zero_sized_type_extract(<4 x i64> %arg, i64 %arg1) {
; CHECK-LABEL: @zero_sized_type_extract(
; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP0:%.*]] = extractelement <4 x i64> [[ARG:%.*]], i64 0
-; CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i32, ptr @global, i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i8, ptr @global, i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[T2]]
;
bb:
diff --git a/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll b/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
index cf71eddf7dbf5..5a38e688bef70 100644
--- a/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
+++ b/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
@@ -504,8 +504,9 @@ define <3 x float> @shuf_frem_const_op1(<3 x float> %x) {
define ptr @gep_vbase_w_s_idx(<2 x ptr> %base, i64 %index) {
; CHECK-LABEL: @gep_vbase_w_s_idx(
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX:%.*]], 2
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x ptr> [[BASE:%.*]], i64 1
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[TMP1]], i64 [[INDEX:%.*]]
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[TMP1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[EE]]
;
%gep = getelementptr i32, <2 x ptr> %base, i64 %index
@@ -541,7 +542,8 @@ define ptr @gep_splat_base_w_cv_idx(ptr %base) {
define ptr @gep_splat_base_w_vidx(ptr %base, <2 x i64> %idxvec) {
; CHECK-LABEL: @gep_splat_base_w_vidx(
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x i64> [[IDXVEC:%.*]], i64 1
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[TMP1]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[EE]]
;
%basevec1 = insertelement <2 x ptr> undef, ptr %base, i32 0
@@ -556,7 +558,8 @@ define ptr @gep_splat_base_w_vidx(ptr %base, <2 x i64> %idxvec) {
define ptr @gep_cvbase_w_s_idx(<2 x ptr> %base, i64 %raw_addr) {
; CHECK-LABEL: @gep_cvbase_w_s_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr @GLOBAL, i64 [[RAW_ADDR:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[RAW_ADDR:%.*]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr @GLOBAL, i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[EE]]
;
%gep = getelementptr i32, <2 x ptr> <ptr @GLOBAL, ptr @GLOBAL>, i64 %raw_addr
@@ -586,7 +589,8 @@ define ptr @gep_sbase_w_cv_idx(ptr %base) {
define ptr @gep_sbase_w_splat_idx(ptr %base, i64 %idx) {
; CHECK-LABEL: @gep_sbase_w_splat_idx(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[EE]]
;
%idxvec1 = insertelement <2 x i64> undef, i64 %idx, i32 0
@@ -597,7 +601,8 @@ define ptr @gep_sbase_w_splat_idx(ptr %base, i64 %idx) {
}
define ptr @gep_splat_both(ptr %base, i64 %idx) {
; CHECK-LABEL: @gep_splat_both(
-; CHECK-NEXT: [[EE:%.*]] = getelementptr i32, ptr [[BASE:%.*]], i64 [[IDX:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX:%.*]], 2
+; CHECK-NEXT: [[EE:%.*]] = getelementptr i8, ptr [[BASE:%.*]], i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[EE]]
;
%basevec1 = insertelement <2 x ptr> undef, ptr %base, i32 0
@@ -613,7 +618,8 @@ define <2 x ptr> @gep_all_lanes_undef(ptr %base, i64 %idx) {;
; CHECK-LABEL: @gep_all_lanes_undef(
; CHECK-NEXT: [[BASEVEC:%.*]] = insertelement <2 x ptr> <ptr poison, ptr undef>, ptr [[BASE:%.*]], i64 0
; CHECK-NEXT: [[IDXVEC:%.*]] = insertelement <2 x i64> <i64 undef, i64 poison>, i64 [[IDX:%.*]], i64 1
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, <2 x ptr> [[BASEVEC]], <2 x i64> [[IDXVEC]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <2 x i64> [[IDXVEC]], splat (i64 2)
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, <2 x ptr> [[BASEVEC]], <2 x i64> [[TMP1]]
; CHECK-NEXT: ret <2 x ptr> [[GEP]]
;
%basevec = insertelement <2 x ptr> undef, ptr %base, i32 0
@@ -655,7 +661,8 @@ define ptr @zero_sized_type_extract(<4 x i64> %arg, i64 %arg1) {
; CHECK-LABEL: @zero_sized_type_extract(
; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP0:%.*]] = extractelement <4 x i64> [[ARG:%.*]], i64 0
-; CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i32, ptr @global, i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i8, ptr @global, i64 [[TMP1]]
; CHECK-NEXT: ret ptr [[T2]]
;
bb:
diff --git a/llvm/test/Transforms/InstCombine/vec_phi_extract-inseltpoison.ll b/llvm/test/Transforms/InstCombine/vec_phi_extract-inseltpoison.ll
index ecd5e49bd6b28..7bb98897247db 100644
--- a/llvm/test/Transforms/InstCombine/vec_phi_extract-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/vec_phi_extract-inseltpoison.ll
@@ -11,7 +11,8 @@ define void @f(i64 %val, i32 %limit, ptr %ptr) {
; CHECK-NEXT: [[END:%.*]] = icmp ult i32 [[TMP1]], [[LIMIT:%.*]]
; CHECK-NEXT: [[TMP2:%.*]] = add i32 [[TMP1]], 10
; CHECK-NEXT: [[TMP3:%.*]] = sext i32 [[TMP1]] to i64
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP3]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP6]]
; CHECK-NEXT: store i32 [[TMP2]], ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5]] = add i32 [[TMP1]], 16
; CHECK-NEXT: br i1 [[END]], label [[LOOP]], label [[RET:%.*]]
@@ -50,7 +51,8 @@ define void @copy(i64 %val, i32 %limit, ptr %ptr) {
; CHECK-NEXT: [[END:%.*]] = icmp ult i32 [[TMP1]], [[LIMIT:%.*]]
; CHECK-NEXT: [[TMP2:%.*]] = add i32 [[TMP1]], 10
; CHECK-NEXT: [[TMP3:%.*]] = sext i32 [[TMP1]] to i64
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP3]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP6]]
; CHECK-NEXT: store i32 [[TMP2]], ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5]] = add i32 [[TMP1]], 16
; CHECK-NEXT: br i1 [[END]], label [[LOOP]], label [[RET:%.*]]
@@ -95,7 +97,8 @@ define void @nocopy(i64 %val, i32 %limit, ptr %ptr) {
; CHECK-NEXT: [[END:%.*]] = icmp ult i32 [[ELT]], [[LIMIT:%.*]]
; CHECK-NEXT: [[TMP5:%.*]] = add i32 [[ELTCOPY]], 10
; CHECK-NEXT: [[TMP6:%.*]] = sext i32 [[ELT]] to i64
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[TMP6]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[TMP6]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP8]]
; CHECK-NEXT: store i32 [[TMP5]], ptr [[TMP7]], align 4
; CHECK-NEXT: [[INC]] = add <16 x i32> [[TMP4]], splat (i32 16)
; CHECK-NEXT: br i1 [[END]], label [[LOOP]], label [[RET:%.*]]
diff --git a/llvm/test/Transforms/InstCombine/vec_phi_extract.ll b/llvm/test/Transforms/InstCombine/vec_phi_extract.ll
index ef093709a98f5..40a376f9860b5 100644
--- a/llvm/test/Transforms/InstCombine/vec_phi_extract.ll
+++ b/llvm/test/Transforms/InstCombine/vec_phi_extract.ll
@@ -11,7 +11,8 @@ define void @f(i64 %val, i32 %limit, ptr %ptr) {
; CHECK-NEXT: [[END:%.*]] = icmp ult i32 [[TMP1]], [[LIMIT:%.*]]
; CHECK-NEXT: [[TMP2:%.*]] = add i32 [[TMP1]], 10
; CHECK-NEXT: [[TMP3:%.*]] = sext i32 [[TMP1]] to i64
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP3]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP6]]
; CHECK-NEXT: store i32 [[TMP2]], ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5]] = add i32 [[TMP1]], 16
; CHECK-NEXT: br i1 [[END]], label [[LOOP]], label [[RET:%.*]]
@@ -50,7 +51,8 @@ define void @copy(i64 %val, i32 %limit, ptr %ptr) {
; CHECK-NEXT: [[END:%.*]] = icmp ult i32 [[TMP1]], [[LIMIT:%.*]]
; CHECK-NEXT: [[TMP2:%.*]] = add i32 [[TMP1]], 10
; CHECK-NEXT: [[TMP3:%.*]] = sext i32 [[TMP1]] to i64
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP3]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP6]]
; CHECK-NEXT: store i32 [[TMP2]], ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5]] = add i32 [[TMP1]], 16
; CHECK-NEXT: br i1 [[END]], label [[LOOP]], label [[RET:%.*]]
@@ -95,7 +97,8 @@ define void @nocopy(i64 %val, i32 %limit, ptr %ptr) {
; CHECK-NEXT: [[END:%.*]] = icmp ult i32 [[ELT]], [[LIMIT:%.*]]
; CHECK-NEXT: [[TMP5:%.*]] = add i32 [[ELTCOPY]], 10
; CHECK-NEXT: [[TMP6:%.*]] = sext i32 [[ELT]] to i64
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[PTR:%.*]], i64 [[TMP6]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[TMP6]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[PTR:%.*]], i64 [[TMP8]]
; CHECK-NEXT: store i32 [[TMP5]], ptr [[TMP7]], align 4
; CHECK-NEXT: [[INC]] = add <16 x i32> [[TMP4]], splat (i32 16)
; CHECK-NEXT: br i1 [[END]], label [[LOOP]], label [[RET:%.*]]
diff --git a/llvm/test/Transforms/InstCombine/vector_gep1-inseltpoison.ll b/llvm/test/Transforms/InstCombine/vector_gep1-inseltpoison.ll
index 4c7a28bac6f4d..d1f348e228d9d 100644
--- a/llvm/test/Transforms/InstCombine/vector_gep1-inseltpoison.ll
+++ b/llvm/test/Transforms/InstCombine/vector_gep1-inseltpoison.ll
@@ -55,7 +55,7 @@ define <2 x i1> @test5(<2 x ptr> %a) {
define <2 x ptr> @test7(<2 x ptr> %a) {
; CHECK-LABEL: @test7(
-; CHECK-NEXT: [[W:%.*]] = getelementptr { i32, i32 }, <2 x ptr> [[A:%.*]], <2 x i64> <i64 5, i64 9>
+; CHECK-NEXT: [[W:%.*]] = getelementptr i8, <2 x ptr> [[A:%.*]], <2 x i64> <i64 40, i64 72>
; CHECK-NEXT: ret <2 x ptr> [[W]]
;
%w = getelementptr {i32, i32}, <2 x ptr> %a, <2 x i32> <i32 5, i32 9>, <2 x i32> zeroinitializer
diff --git a/llvm/test/Transforms/InstCombine/vector_gep1.ll b/llvm/test/Transforms/InstCombine/vector_gep1.ll
index 88af1c45e89a5..593580457404b 100644
--- a/llvm/test/Transforms/InstCombine/vector_gep1.ll
+++ b/llvm/test/Transforms/InstCombine/vector_gep1.ll
@@ -55,7 +55,7 @@ define <2 x i1> @test5(<2 x ptr> %a) {
define <2 x ptr> @test7(<2 x ptr> %a) {
; CHECK-LABEL: @test7(
-; CHECK-NEXT: [[W:%.*]] = getelementptr { i32, i32 }, <2 x ptr> [[A:%.*]], <2 x i64> <i64 5, i64 9>
+; CHECK-NEXT: [[W:%.*]] = getelementptr i8, <2 x ptr> [[A:%.*]], <2 x i64> <i64 40, i64 72>
; CHECK-NEXT: ret <2 x ptr> [[W]]
;
%w = getelementptr {i32, i32}, <2 x ptr> %a, <2 x i32> <i32 5, i32 9>, <2 x i32> zeroinitializer
diff --git a/llvm/test/Transforms/InstCombine/vector_gep2.ll b/llvm/test/Transforms/InstCombine/vector_gep2.ll
index cfa2f6277d858..a8b650e801cd1 100644
--- a/llvm/test/Transforms/InstCombine/vector_gep2.ll
+++ b/llvm/test/Transforms/InstCombine/vector_gep2.ll
@@ -15,7 +15,8 @@ define <2 x ptr> @testa(<2 x ptr> %a) {
define <8 x ptr> @vgep_s_v8i64(ptr %a, <8 x i64>%i) {
; CHECK-LABEL: @vgep_s_v8i64(
-; CHECK-NEXT: [[VECTORGEP:%.*]] = getelementptr double, ptr [[A:%.*]], <8 x i64> [[I:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl <8 x i64> [[I:%.*]], splat (i64 3)
+; CHECK-NEXT: [[VECTORGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], <8 x i64> [[TMP1]]
; CHECK-NEXT: ret <8 x ptr> [[VECTORGEP]]
;
%VectorGep = getelementptr double, ptr %a, <8 x i64> %i
@@ -25,7 +26,8 @@ define <8 x ptr> @vgep_s_v8i64(ptr %a, <8 x i64>%i) {
define <8 x ptr> @vgep_s_v8i32(ptr %a, <8 x i32>%i) {
; CHECK-LABEL: @vgep_s_v8i32(
; CHECK-NEXT: [[TMP1:%.*]] = sext <8 x i32> [[I:%.*]] to <8 x i64>
-; CHECK-NEXT: [[VECTORGEP:%.*]] = getelementptr double, ptr [[A:%.*]], <8 x i64> [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw <8 x i64> [[TMP1]], splat (i64 3)
+; CHECK-NEXT: [[VECTORGEP:%.*]] = getelementptr i8, ptr [[A:%.*]], <8 x i64> [[TMP2]]
; CHECK-NEXT: ret <8 x ptr> [[VECTORGEP]]
;
%VectorGep = getelementptr double, ptr %a, <8 x i32> %i
diff --git a/llvm/test/Transforms/InstCombine/vectorgep-crash.ll b/llvm/test/Transforms/InstCombine/vectorgep-crash.ll
index d33a2d3417bfd..5457b36e42574 100644
--- a/llvm/test/Transforms/InstCombine/vectorgep-crash.ll
+++ b/llvm/test/Transforms/InstCombine/vectorgep-crash.ll
@@ -14,7 +14,8 @@ define <8 x ptr> @test_vector_gep(ptr %arg1, <8 x i64> %arg2) {
; CHECK-LABEL: define <8 x ptr> @test_vector_gep(
; CHECK-SAME: ptr [[ARG1:%.*]], <8 x i64> [[ARG2:%.*]]) {
; CHECK-NEXT: [[TOP:.*:]]
-; CHECK-NEXT: [[VECTORGEP14_SPLIT:%.*]] = getelementptr inbounds [[DUAL:%.*]], ptr [[ARG1]], <8 x i64> [[ARG2]]
+; CHECK-NEXT: [[TMP0:%.*]] = mul nsw <8 x i64> [[ARG2]], splat (i64 72)
+; CHECK-NEXT: [[VECTORGEP14_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[ARG1]], <8 x i64> [[TMP0]]
; CHECK-NEXT: [[VECTORGEP14:%.*]] = getelementptr inbounds i8, <8 x ptr> [[VECTORGEP14_SPLIT]], i64 32
; CHECK-NEXT: ret <8 x ptr> [[VECTORGEP14]]
;
@@ -32,7 +33,7 @@ top:
define <16 x ptr> @test() {
; CHECK-LABEL: define <16 x ptr> @test() {
; CHECK-NEXT: [[VECTOR_BODY:.*:]]
-; CHECK-NEXT: ret <16 x ptr> getelementptr ([65 x %struct.A], ptr @G, <16 x i64> zeroinitializer, <16 x i64> <i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16>, i32 0)
+; CHECK-NEXT: ret <16 x ptr> getelementptr ([65 x [[STRUCT_A:%.*]]], ptr @G, <16 x i64> zeroinitializer, <16 x i64> <i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16>, i32 0)
;
vector.body:
%VectorGep = getelementptr [65 x %struct.A], ptr @G, <16 x i64> zeroinitializer, <16 x i64> <i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16>, <16 x i32> zeroinitializer
@@ -42,7 +43,7 @@ vector.body:
define <16 x ptr> @test2() {
; CHECK-LABEL: define <16 x ptr> @test2() {
; CHECK-NEXT: [[VECTOR_BODY:.*:]]
-; CHECK-NEXT: ret <16 x ptr> getelementptr ([65 x %struct.A], ptr @G, <16 x i64> zeroinitializer, <16 x i64> <i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16>, i32 0)
+; CHECK-NEXT: ret <16 x ptr> getelementptr ([65 x [[STRUCT_A:%.*]]], ptr @G, <16 x i64> zeroinitializer, <16 x i64> <i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16>, i32 0)
;
vector.body:
%VectorGep = getelementptr [65 x %struct.A], ptr @G, <16 x i32> zeroinitializer, <16 x i64> <i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7, i64 8, i64 9, i64 10, i64 11, i64 12, i64 13, i64 14, i64 15, i64 16>, <16 x i32> zeroinitializer
diff --git a/llvm/test/Transforms/InstCombine/wcslen-1.ll b/llvm/test/Transforms/InstCombine/wcslen-1.ll
index 3c79c32ffe0da..be1e9add4bdd7 100644
--- a/llvm/test/Transforms/InstCombine/wcslen-1.ll
+++ b/llvm/test/Transforms/InstCombine/wcslen-1.ll
@@ -110,7 +110,9 @@ define i64 @test_simplify9(i1 %x) {
define i64 @test_simplify10(i32 %x) {
; CHECK-LABEL: @test_simplify10(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[X:%.*]] to i64
-; CHECK-NEXT: [[HELLO_L:%.*]] = sub nsw i64 5, [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @hello, i64 [[TMP2]]
+; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
%hello_p = getelementptr inbounds [6 x i32], ptr @hello, i32 0, i32 %x
@@ -120,7 +122,9 @@ define i64 @test_simplify10(i32 %x) {
define i64 @test_simplify10_gepi32(i64 %x) {
; CHECK-LABEL: @test_simplify10_gepi32(
-; CHECK-NEXT: [[HELLO_L:%.*]] = sub i64 5, [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 2
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @hello, i64 [[TMP1]]
+; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
%hello_p = getelementptr inbounds i32, ptr @hello, i64 %x
@@ -131,7 +135,8 @@ define i64 @test_simplify10_gepi32(i64 %x) {
define i64 @test_simplify10_gepi64(i32 %x) {
; CHECK-LABEL: @test_simplify10_gepi64(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[X:%.*]] to i64
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i64, ptr @hello, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 3
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @hello, i64 [[TMP2]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
@@ -142,7 +147,8 @@ define i64 @test_simplify10_gepi64(i32 %x) {
define i64 @test_simplify10_gepi16(i64 %x) {
; CHECK-LABEL: @test_simplify10_gepi16(
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i16, ptr @hello, i64 [[X:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[X:%.*]], 1
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @hello, i64 [[TMP1]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
@@ -179,10 +185,12 @@ define i64 @test_simplify10_gepi8mul4(i64 %x) {
define i64 @test_simplify11(i32 %x) {
; CHECK-LABEL: @test_simplify11(
-; CHECK-NEXT: [[AND:%.*]] = and i32 [[X:%.*]], 7
-; CHECK-NEXT: [[NARROW:%.*]] = sub nuw nsw i32 9, [[AND]]
+; CHECK-NEXT: [[AND:%.*]] = shl i32 [[X:%.*]], 2
+; CHECK-NEXT: [[NARROW:%.*]] = and i32 [[AND]], 28
; CHECK-NEXT: [[HELLO_L:%.*]] = zext nneg i32 [[NARROW]] to i64
-; CHECK-NEXT: ret i64 [[HELLO_L]]
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i8, ptr @null_hello_mid, i64 [[HELLO_L]]
+; CHECK-NEXT: [[HELLO_L1:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
+; CHECK-NEXT: ret i64 [[HELLO_L1]]
;
%and = and i32 %x, 7
%hello_p = getelementptr inbounds [13 x i32], ptr @null_hello_mid, i32 0, i32 %and
@@ -192,10 +200,12 @@ define i64 @test_simplify11(i32 %x) {
define i64 @test_simplify11_gepi32(i32 %x) {
; CHECK-LABEL: @test_simplify11_gepi32(
-; CHECK-NEXT: [[AND:%.*]] = and i32 [[X:%.*]], 7
-; CHECK-NEXT: [[NARROW:%.*]] = sub nuw nsw i32 9, [[AND]]
+; CHECK-NEXT: [[AND:%.*]] = shl i32 [[X:%.*]], 2
+; CHECK-NEXT: [[NARROW:%.*]] = and i32 [[AND]], 28
; CHECK-NEXT: [[HELLO_L:%.*]] = zext nneg i32 [[NARROW]] to i64
-; CHECK-NEXT: ret i64 [[HELLO_L]]
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i8, ptr @null_hello_mid, i64 [[HELLO_L]]
+; CHECK-NEXT: [[HELLO_L1:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
+; CHECK-NEXT: ret i64 [[HELLO_L1]]
;
%and = and i32 %x, 7
%hello_p = getelementptr inbounds i32, ptr @null_hello_mid, i32 %and
@@ -233,7 +243,8 @@ define i64 @test_no_simplify1() {
define i64 @test_no_simplify2(i32 %x) {
; CHECK-LABEL: @test_no_simplify2(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[X:%.*]] to i64
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i32, ptr @null_hello, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @null_hello, i64 [[TMP2]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
@@ -245,7 +256,8 @@ define i64 @test_no_simplify2(i32 %x) {
define i64 @test_no_simplify2_no_null_opt(i32 %x) #0 {
; CHECK-LABEL: @test_no_simplify2_no_null_opt(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[X:%.*]] to i64
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i32, ptr @null_hello, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @null_hello, i64 [[TMP2]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
@@ -258,9 +270,10 @@ define i64 @test_no_simplify2_no_null_opt(i32 %x) #0 {
define i64 @test_no_simplify3(i32 %x) {
; CHECK-LABEL: @test_no_simplify3(
-; CHECK-NEXT: [[AND:%.*]] = and i32 [[X:%.*]], 15
+; CHECK-NEXT: [[AND1:%.*]] = shl i32 [[X:%.*]], 2
+; CHECK-NEXT: [[AND:%.*]] = and i32 [[AND1]], 60
; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[AND]] to i64
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i32, ptr @null_hello_mid, i64 [[TMP1]]
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i8, ptr @null_hello_mid, i64 [[TMP1]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
@@ -272,9 +285,10 @@ define i64 @test_no_simplify3(i32 %x) {
define i64 @test_no_simplify3_no_null_opt(i32 %x) #0 {
; CHECK-LABEL: @test_no_simplify3_no_null_opt(
-; CHECK-NEXT: [[AND:%.*]] = and i32 [[X:%.*]], 15
+; CHECK-NEXT: [[AND1:%.*]] = shl i32 [[X:%.*]], 2
+; CHECK-NEXT: [[AND:%.*]] = and i32 [[AND1]], 60
; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[AND]] to i64
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i32, ptr @null_hello_mid, i64 [[TMP1]]
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i8, ptr @null_hello_mid, i64 [[TMP1]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
diff --git a/llvm/test/Transforms/InstCombine/wcslen-3.ll b/llvm/test/Transforms/InstCombine/wcslen-3.ll
index 89fc4e758c54e..229c0ae2d2a2e 100644
--- a/llvm/test/Transforms/InstCombine/wcslen-3.ll
+++ b/llvm/test/Transforms/InstCombine/wcslen-3.ll
@@ -111,7 +111,9 @@ define i64 @test_simplify9(i1 %x) {
define i64 @test_simplify10(i16 %x) {
; CHECK-LABEL: @test_simplify10(
; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[X:%.*]] to i64
-; CHECK-NEXT: [[HELLO_L:%.*]] = sub nsw i64 5, [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 1
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @hello, i64 [[TMP2]]
+; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
%hello_p = getelementptr inbounds [6 x i16], ptr @hello, i16 0, i16 %x
@@ -123,10 +125,12 @@ define i64 @test_simplify10(i16 %x) {
define i64 @test_simplify11(i16 %x) {
; CHECK-LABEL: @test_simplify11(
-; CHECK-NEXT: [[AND:%.*]] = and i16 [[X:%.*]], 7
-; CHECK-NEXT: [[NARROW:%.*]] = sub nuw nsw i16 9, [[AND]]
+; CHECK-NEXT: [[AND:%.*]] = shl i16 [[X:%.*]], 1
+; CHECK-NEXT: [[NARROW:%.*]] = and i16 [[AND]], 14
; CHECK-NEXT: [[HELLO_L:%.*]] = zext nneg i16 [[NARROW]] to i64
-; CHECK-NEXT: ret i64 [[HELLO_L]]
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i8, ptr @null_hello_mid, i64 [[HELLO_L]]
+; CHECK-NEXT: [[HELLO_L1:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
+; CHECK-NEXT: ret i64 [[HELLO_L1]]
;
%and = and i16 %x, 7
%hello_p = getelementptr inbounds [13 x i16], ptr @null_hello_mid, i16 0, i16 %and
@@ -150,7 +154,8 @@ define i64 @test_no_simplify1() {
define i64 @test_no_simplify2(i16 %x) {
; CHECK-LABEL: @test_no_simplify2(
; CHECK-NEXT: [[TMP1:%.*]] = sext i16 [[X:%.*]] to i64
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i16, ptr @null_hello, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[TMP1]], 1
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds i8, ptr @null_hello, i64 [[TMP2]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
@@ -163,9 +168,10 @@ define i64 @test_no_simplify2(i16 %x) {
define i64 @test_no_simplify3(i16 %x) {
; CHECK-LABEL: @test_no_simplify3(
-; CHECK-NEXT: [[AND:%.*]] = and i16 [[X:%.*]], 15
+; CHECK-NEXT: [[AND1:%.*]] = shl i16 [[X:%.*]], 1
+; CHECK-NEXT: [[AND:%.*]] = and i16 [[AND1]], 30
; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i16 [[AND]] to i64
-; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i16, ptr @null_hello_mid, i64 [[TMP1]]
+; CHECK-NEXT: [[HELLO_P:%.*]] = getelementptr inbounds nuw i8, ptr @null_hello_mid, i64 [[TMP1]]
; CHECK-NEXT: [[HELLO_L:%.*]] = call i64 @wcslen(ptr nonnull [[HELLO_P]])
; CHECK-NEXT: ret i64 [[HELLO_L]]
;
diff --git a/llvm/test/Transforms/InstCombine/wcslen-5.ll b/llvm/test/Transforms/InstCombine/wcslen-5.ll
index b127f2ae5ea1b..622f9d5330213 100644
--- a/llvm/test/Transforms/InstCombine/wcslen-5.ll
+++ b/llvm/test/Transforms/InstCombine/wcslen-5.ll
@@ -19,7 +19,8 @@ declare i64 @wcslen(ptr)
define dso_local i64 @fold_wcslen_s3_pi_s5(i1 zeroext %0, i64 %1) {
; CHECK-LABEL: @fold_wcslen_s3_pi_s5(
-; CHECK-NEXT: [[PS3_PI:%.*]] = getelementptr inbounds i32, ptr @ws3, i64 [[TMP1:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1:%.*]], 2
+; CHECK-NEXT: [[PS3_PI:%.*]] = getelementptr inbounds i8, ptr @ws3, i64 [[TMP3]]
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[TMP0:%.*]], ptr [[PS3_PI]], ptr @ws5
; CHECK-NEXT: [[LEN:%.*]] = tail call i64 @wcslen(ptr nonnull [[SEL]])
; CHECK-NEXT: ret i64 [[LEN]]
@@ -41,7 +42,8 @@ define dso_local i64 @fold_wcslen_s3_pi_p1_s5(i1 zeroext %0, i64 %1) {
; XFAIL-CHECK-NEXT: [[SEL:%.*]] = select i1 %0, i64 [[DIF_I]], i64 5
; XFAIL-CHECK-NEXT: ret i64 [[SEL]]
; CHECK-LABEL: @fold_wcslen_s3_pi_p1_s5(
-; CHECK-NEXT: [[PS3_PI:%.*]] = getelementptr inbounds i32, ptr @ws3, i64 [[TMP1:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1:%.*]], 2
+; CHECK-NEXT: [[PS3_PI:%.*]] = getelementptr inbounds i8, ptr @ws3, i64 [[TMP3]]
; CHECK-NEXT: [[PS3_PI_P1:%.*]] = getelementptr inbounds nuw 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]])
@@ -62,7 +64,8 @@ define dso_local i64 @fold_wcslen_s3_pi_p1_s5(i1 zeroext %0, i64 %1) {
define dso_local i64 @call_wcslen_s5_3_pi_s5(i1 zeroext %0, i64 %1) {
; CHECK-LABEL: @call_wcslen_s5_3_pi_s5(
-; CHECK-NEXT: [[PS5_3_PI:%.*]] = getelementptr inbounds i32, ptr @ws5_3, i64 [[TMP1:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1:%.*]], 2
+; CHECK-NEXT: [[PS5_3_PI:%.*]] = getelementptr inbounds i8, ptr @ws5_3, i64 [[TMP3]]
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[TMP0:%.*]], ptr [[PS5_3_PI]], ptr @ws5
; CHECK-NEXT: [[LEN:%.*]] = tail call i64 @wcslen(ptr nonnull [[SEL]])
; CHECK-NEXT: ret i64 [[LEN]]
@@ -79,7 +82,8 @@ define dso_local i64 @call_wcslen_s5_3_pi_s5(i1 zeroext %0, i64 %1) {
define dso_local i64 @call_wcslen_s5_3_s5_pj(i1 zeroext %0, i64 %1) {
; CHECK-LABEL: @call_wcslen_s5_3_s5_pj(
-; CHECK-NEXT: [[PS5:%.*]] = getelementptr inbounds i32, ptr @ws5, i64 [[TMP1:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1:%.*]], 2
+; CHECK-NEXT: [[PS5:%.*]] = getelementptr inbounds i8, ptr @ws5, i64 [[TMP3]]
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[TMP0:%.*]], ptr @ws5_3, ptr [[PS5]]
; CHECK-NEXT: [[LEN:%.*]] = tail call i64 @wcslen(ptr nonnull [[SEL]])
; CHECK-NEXT: ret i64 [[LEN]]
@@ -96,7 +100,8 @@ define dso_local i64 @call_wcslen_s5_3_s5_pj(i1 zeroext %0, i64 %1) {
define dso_local i64 @fold_wcslen_s3_s5_pj(i1 zeroext %0, i64 %1) {
; CHECK-LABEL: @fold_wcslen_s3_s5_pj(
-; CHECK-NEXT: [[PS5_PJ:%.*]] = getelementptr inbounds i32, ptr @ws5, i64 [[TMP1:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1:%.*]], 2
+; CHECK-NEXT: [[PS5_PJ:%.*]] = getelementptr inbounds i8, ptr @ws5, i64 [[TMP3]]
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[TMP0:%.*]], ptr @ws3, ptr [[PS5_PJ]]
; CHECK-NEXT: [[LEN:%.*]] = tail call i64 @wcslen(ptr nonnull [[SEL]])
; CHECK-NEXT: ret i64 [[LEN]]
@@ -115,7 +120,8 @@ define dso_local i64 @fold_wcslen_s3_s5_pj(i1 zeroext %0, i64 %1) {
define dso_local i64 @call_wcslen_s3_s5_3_pj(i1 zeroext %0, i64 %1) {
; CHECK-LABEL: @call_wcslen_s3_s5_3_pj(
-; CHECK-NEXT: [[PS5_3_PJ:%.*]] = getelementptr inbounds i32, ptr @ws5_3, i64 [[TMP1:%.*]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[TMP1:%.*]], 2
+; CHECK-NEXT: [[PS5_3_PJ:%.*]] = getelementptr inbounds i8, ptr @ws5_3, i64 [[TMP3]]
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[TMP0:%.*]], ptr @ws3, ptr [[PS5_3_PJ]]
; CHECK-NEXT: [[LEN:%.*]] = tail call i64 @wcslen(ptr nonnull [[SEL]])
; CHECK-NEXT: ret i64 [[LEN]]
@@ -132,8 +138,10 @@ define dso_local i64 @call_wcslen_s3_s5_3_pj(i1 zeroext %0, i64 %1) {
define dso_local i64 @fold_wcslen_s3_pi_s5_pj(i1 zeroext %0, i64 %1, i64 %2) {
; CHECK-LABEL: @fold_wcslen_s3_pi_s5_pj(
-; CHECK-NEXT: [[PS3_PI:%.*]] = getelementptr inbounds i32, ptr @ws3, i64 [[TMP1:%.*]]
-; CHECK-NEXT: [[PS5_PJ:%.*]] = getelementptr inbounds i32, ptr @ws5, i64 [[TMP2:%.*]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP1:%.*]], 2
+; CHECK-NEXT: [[PS3_PI:%.*]] = getelementptr inbounds i8, ptr @ws3, i64 [[TMP4]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP2:%.*]], 2
+; CHECK-NEXT: [[PS5_PJ:%.*]] = getelementptr inbounds i8, ptr @ws5, i64 [[TMP5]]
; CHECK-NEXT: [[SEL:%.*]] = select i1 [[TMP0:%.*]], ptr [[PS3_PI]], ptr [[PS5_PJ]]
; CHECK-NEXT: [[LEN:%.*]] = tail call i64 @wcslen(ptr nonnull [[SEL]])
; CHECK-NEXT: ret i64 [[LEN]]
diff --git a/llvm/test/Transforms/LoopSimplify/merge-exits.ll b/llvm/test/Transforms/LoopSimplify/merge-exits.ll
index 10e634fd47a1a..75c802cc17934 100644
--- a/llvm/test/Transforms/LoopSimplify/merge-exits.ll
+++ b/llvm/test/Transforms/LoopSimplify/merge-exits.ll
@@ -22,11 +22,13 @@ define float @test1(ptr %pTmp1, ptr %peakWeight, i32 %bandEdgeIndex) nounwind {
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[BB_LR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[BB]] ]
; CHECK-NEXT: [[DISTERBHI_04:%.*]] = phi float [ 0.000000e+00, [[BB_LR_PH]] ], [ [[T4:%.*]], [[BB]] ]
; CHECK-NEXT: [[PEAKCOUNT_02:%.*]] = phi float [ [[T0]], [[BB_LR_PH]] ], [ [[T9:%.*]], [[BB]] ]
-; CHECK-NEXT: [[T2:%.*]] = getelementptr float, ptr [[PTMP1:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[T2:%.*]] = getelementptr i8, ptr [[PTMP1:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[T3:%.*]] = load float, ptr [[T2]], align 4
; CHECK-NEXT: [[T4]] = fadd float [[T3]], [[DISTERBHI_04]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[T7:%.*]] = getelementptr float, ptr [[PEAKWEIGHT]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT]], 2
+; CHECK-NEXT: [[T7:%.*]] = getelementptr i8, ptr [[PEAKWEIGHT]], i64 [[TMP2]]
; CHECK-NEXT: [[T8:%.*]] = load float, ptr [[T7]], align 4
; CHECK-NEXT: [[T9]] = fadd float [[T8]], [[PEAKCOUNT_02]]
; CHECK-NEXT: [[T10:%.*]] = fcmp olt float [[T4]], 2.500000e+00
@@ -93,11 +95,13 @@ define float @merge_branches_profile_metadata(ptr %pTmp1, ptr %peakWeight, i32 %
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[BB_LR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[BB]] ]
; CHECK-NEXT: [[DISTERBHI_04:%.*]] = phi float [ 0.000000e+00, [[BB_LR_PH]] ], [ [[T4:%.*]], [[BB]] ]
; CHECK-NEXT: [[PEAKCOUNT_02:%.*]] = phi float [ [[T0]], [[BB_LR_PH]] ], [ [[T9:%.*]], [[BB]] ]
-; CHECK-NEXT: [[T2:%.*]] = getelementptr float, ptr [[PTMP1:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[T2:%.*]] = getelementptr i8, ptr [[PTMP1:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[T3:%.*]] = load float, ptr [[T2]], align 4
; CHECK-NEXT: [[T4]] = fadd float [[T3]], [[DISTERBHI_04]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[T7:%.*]] = getelementptr float, ptr [[PEAKWEIGHT]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT]], 2
+; CHECK-NEXT: [[T7:%.*]] = getelementptr i8, ptr [[PEAKWEIGHT]], i64 [[TMP2]]
; CHECK-NEXT: [[T8:%.*]] = load float, ptr [[T7]], align 4
; CHECK-NEXT: [[T9]] = fadd float [[T8]], [[PEAKCOUNT_02]]
; CHECK-NEXT: [[T10:%.*]] = fcmp olt float [[T4]], 2.500000e+00
diff --git a/llvm/test/Transforms/LoopUnroll/AArch64/runtime-unroll-generic.ll b/llvm/test/Transforms/LoopUnroll/AArch64/runtime-unroll-generic.ll
index 66c55f20e533f..4adf8975c7473 100644
--- a/llvm/test/Transforms/LoopUnroll/AArch64/runtime-unroll-generic.ll
+++ b/llvm/test/Transforms/LoopUnroll/AArch64/runtime-unroll-generic.ll
@@ -19,50 +19,58 @@ define void @runtime_unroll_generic(i32 %arg_0, ptr %arg_1, ptr %arg_2, ptr %arg
; CHECK-A55: for.body6:
; CHECK-A55-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[FOR_BODY6_PREHEADER_NEW]] ], [ [[INDVARS_IV_NEXT_3:%.*]], [[FOR_BODY6]] ]
; CHECK-A55-NEXT: [[NITER:%.*]] = phi i64 [ 0, [[FOR_BODY6_PREHEADER_NEW]] ], [ [[NITER_NEXT_3:%.*]], [[FOR_BODY6]] ]
-; CHECK-A55-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2:%.*]], i64 [[INDVARS_IV]]
+; CHECK-A55-NEXT: [[TMP22:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-A55-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2:%.*]], i64 [[TMP22]]
; CHECK-A55-NEXT: [[TMP1:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
; CHECK-A55-NEXT: [[CONV:%.*]] = sext i16 [[TMP1]] to i32
-; CHECK-A55-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3:%.*]], i64 [[INDVARS_IV]]
+; CHECK-A55-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3:%.*]], i64 [[TMP22]]
; CHECK-A55-NEXT: [[TMP2:%.*]] = load i16, ptr [[ARRAYIDX14]], align 2
; CHECK-A55-NEXT: [[CONV15:%.*]] = sext i16 [[TMP2]] to i32
; CHECK-A55-NEXT: [[MUL16:%.*]] = mul nsw i32 [[CONV15]], [[CONV]]
-; CHECK-A55-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1:%.*]], i64 [[INDVARS_IV]]
+; CHECK-A55-NEXT: [[TMP23:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 2
+; CHECK-A55-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1:%.*]], i64 [[TMP23]]
; CHECK-A55-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX20]], align 4
; CHECK-A55-NEXT: [[ADD21:%.*]] = add nsw i32 [[MUL16]], [[TMP3]]
; CHECK-A55-NEXT: store i32 [[ADD21]], ptr [[ARRAYIDX20]], align 4
; CHECK-A55-NEXT: [[INDVARS_IV_NEXT:%.*]] = or disjoint i64 [[INDVARS_IV]], 1
-; CHECK-A55-NEXT: [[ARRAYIDX10_1:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-A55-NEXT: [[TMP25:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT]], 1
+; CHECK-A55-NEXT: [[ARRAYIDX10_1:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2]], i64 [[TMP25]]
; CHECK-A55-NEXT: [[TMP4:%.*]] = load i16, ptr [[ARRAYIDX10_1]], align 2
; CHECK-A55-NEXT: [[CONV_1:%.*]] = sext i16 [[TMP4]] to i32
-; CHECK-A55-NEXT: [[ARRAYIDX14_1:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-A55-NEXT: [[ARRAYIDX14_1:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3]], i64 [[TMP25]]
; CHECK-A55-NEXT: [[TMP5:%.*]] = load i16, ptr [[ARRAYIDX14_1]], align 2
; CHECK-A55-NEXT: [[CONV15_1:%.*]] = sext i16 [[TMP5]] to i32
; CHECK-A55-NEXT: [[MUL16_1:%.*]] = mul nsw i32 [[CONV15_1]], [[CONV_1]]
-; CHECK-A55-NEXT: [[ARRAYIDX20_1:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-A55-NEXT: [[TMP27:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT]], 2
+; CHECK-A55-NEXT: [[ARRAYIDX20_1:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1]], i64 [[TMP27]]
; CHECK-A55-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX20_1]], align 4
; CHECK-A55-NEXT: [[ADD21_1:%.*]] = add nsw i32 [[MUL16_1]], [[TMP6]]
; CHECK-A55-NEXT: store i32 [[ADD21_1]], ptr [[ARRAYIDX20_1]], align 4
; CHECK-A55-NEXT: [[INDVARS_IV_NEXT_1:%.*]] = or disjoint i64 [[INDVARS_IV]], 2
-; CHECK-A55-NEXT: [[ARRAYIDX10_2:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2]], i64 [[INDVARS_IV_NEXT_1]]
+; CHECK-A55-NEXT: [[TMP28:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_1]], 1
+; CHECK-A55-NEXT: [[ARRAYIDX10_2:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2]], i64 [[TMP28]]
; CHECK-A55-NEXT: [[TMP7:%.*]] = load i16, ptr [[ARRAYIDX10_2]], align 2
; CHECK-A55-NEXT: [[CONV_2:%.*]] = sext i16 [[TMP7]] to i32
-; CHECK-A55-NEXT: [[ARRAYIDX14_2:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3]], i64 [[INDVARS_IV_NEXT_1]]
+; CHECK-A55-NEXT: [[ARRAYIDX14_2:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3]], i64 [[TMP28]]
; CHECK-A55-NEXT: [[TMP8:%.*]] = load i16, ptr [[ARRAYIDX14_2]], align 2
; CHECK-A55-NEXT: [[CONV15_2:%.*]] = sext i16 [[TMP8]] to i32
; CHECK-A55-NEXT: [[MUL16_2:%.*]] = mul nsw i32 [[CONV15_2]], [[CONV_2]]
-; CHECK-A55-NEXT: [[ARRAYIDX20_2:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1]], i64 [[INDVARS_IV_NEXT_1]]
+; CHECK-A55-NEXT: [[TMP30:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_1]], 2
+; CHECK-A55-NEXT: [[ARRAYIDX20_2:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1]], i64 [[TMP30]]
; CHECK-A55-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX20_2]], align 4
; CHECK-A55-NEXT: [[ADD21_2:%.*]] = add nsw i32 [[MUL16_2]], [[TMP9]]
; CHECK-A55-NEXT: store i32 [[ADD21_2]], ptr [[ARRAYIDX20_2]], align 4
; CHECK-A55-NEXT: [[INDVARS_IV_NEXT_2:%.*]] = or disjoint i64 [[INDVARS_IV]], 3
-; CHECK-A55-NEXT: [[ARRAYIDX10_3:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-A55-NEXT: [[TMP32:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_2]], 1
+; CHECK-A55-NEXT: [[ARRAYIDX10_3:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2]], i64 [[TMP32]]
; CHECK-A55-NEXT: [[TMP10:%.*]] = load i16, ptr [[ARRAYIDX10_3]], align 2
; CHECK-A55-NEXT: [[CONV_3:%.*]] = sext i16 [[TMP10]] to i32
-; CHECK-A55-NEXT: [[ARRAYIDX14_3:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-A55-NEXT: [[ARRAYIDX14_3:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3]], i64 [[TMP32]]
; CHECK-A55-NEXT: [[TMP11:%.*]] = load i16, ptr [[ARRAYIDX14_3]], align 2
; CHECK-A55-NEXT: [[CONV15_3:%.*]] = sext i16 [[TMP11]] to i32
; CHECK-A55-NEXT: [[MUL16_3:%.*]] = mul nsw i32 [[CONV15_3]], [[CONV_3]]
-; CHECK-A55-NEXT: [[ARRAYIDX20_3:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-A55-NEXT: [[TMP33:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_2]], 2
+; CHECK-A55-NEXT: [[ARRAYIDX20_3:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1]], i64 [[TMP33]]
; CHECK-A55-NEXT: [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX20_3]], align 4
; CHECK-A55-NEXT: [[ADD21_3:%.*]] = add nsw i32 [[MUL16_3]], [[TMP12]]
; CHECK-A55-NEXT: store i32 [[ADD21_3]], ptr [[ARRAYIDX20_3]], align 4
@@ -77,14 +85,16 @@ define void @runtime_unroll_generic(i32 %arg_0, ptr %arg_1, ptr %arg_2, ptr %arg
; CHECK-A55-NEXT: [[INDVARS_IV_UNR:%.*]] = phi i64 [ 0, [[FOR_BODY6_PREHEADER]] ], [ [[INDVARS_IV_NEXT_3]], [[FOR_END_LOOPEXIT_UNR_LCSSA1]] ]
; CHECK-A55-NEXT: [[LCMP_MOD5:%.*]] = icmp ne i64 [[XTRAITER]], 0
; CHECK-A55-NEXT: tail call void @llvm.assume(i1 [[LCMP_MOD5]])
-; CHECK-A55-NEXT: [[ARRAYIDX10_EPIL:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2]], i64 [[INDVARS_IV_UNR]]
+; CHECK-A55-NEXT: [[TMP35:%.*]] = shl nuw nsw i64 [[INDVARS_IV_UNR]], 1
+; CHECK-A55-NEXT: [[ARRAYIDX10_EPIL:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2]], i64 [[TMP35]]
; CHECK-A55-NEXT: [[TMP13:%.*]] = load i16, ptr [[ARRAYIDX10_EPIL]], align 2
; CHECK-A55-NEXT: [[CONV_EPIL:%.*]] = sext i16 [[TMP13]] to i32
-; CHECK-A55-NEXT: [[ARRAYIDX14_EPIL:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3]], i64 [[INDVARS_IV_UNR]]
+; CHECK-A55-NEXT: [[ARRAYIDX14_EPIL:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3]], i64 [[TMP35]]
; CHECK-A55-NEXT: [[TMP14:%.*]] = load i16, ptr [[ARRAYIDX14_EPIL]], align 2
; CHECK-A55-NEXT: [[CONV15_EPIL:%.*]] = sext i16 [[TMP14]] to i32
; CHECK-A55-NEXT: [[MUL16_EPIL:%.*]] = mul nsw i32 [[CONV15_EPIL]], [[CONV_EPIL]]
-; CHECK-A55-NEXT: [[ARRAYIDX20_EPIL:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1]], i64 [[INDVARS_IV_UNR]]
+; CHECK-A55-NEXT: [[TMP24:%.*]] = shl nuw nsw i64 [[INDVARS_IV_UNR]], 2
+; CHECK-A55-NEXT: [[ARRAYIDX20_EPIL:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1]], i64 [[TMP24]]
; CHECK-A55-NEXT: [[TMP15:%.*]] = load i32, ptr [[ARRAYIDX20_EPIL]], align 4
; CHECK-A55-NEXT: [[ADD21_EPIL:%.*]] = add nsw i32 [[MUL16_EPIL]], [[TMP15]]
; CHECK-A55-NEXT: store i32 [[ADD21_EPIL]], ptr [[ARRAYIDX20_EPIL]], align 4
@@ -92,14 +102,16 @@ define void @runtime_unroll_generic(i32 %arg_0, ptr %arg_1, ptr %arg_2, ptr %arg
; CHECK-A55-NEXT: br i1 [[EPIL_ITER_CMP_NOT]], label [[FOR_END]], label [[FOR_BODY6_EPIL_1:%.*]]
; CHECK-A55: for.body6.epil.1:
; CHECK-A55-NEXT: [[INDVARS_IV_NEXT_EPIL:%.*]] = add nuw nsw i64 [[INDVARS_IV_UNR]], 1
-; CHECK-A55-NEXT: [[ARRAYIDX10_EPIL_1:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2]], i64 [[INDVARS_IV_NEXT_EPIL]]
+; CHECK-A55-NEXT: [[TMP26:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_EPIL]], 1
+; CHECK-A55-NEXT: [[ARRAYIDX10_EPIL_1:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2]], i64 [[TMP26]]
; CHECK-A55-NEXT: [[TMP16:%.*]] = load i16, ptr [[ARRAYIDX10_EPIL_1]], align 2
; CHECK-A55-NEXT: [[CONV_EPIL_1:%.*]] = sext i16 [[TMP16]] to i32
-; CHECK-A55-NEXT: [[ARRAYIDX14_EPIL_1:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3]], i64 [[INDVARS_IV_NEXT_EPIL]]
+; CHECK-A55-NEXT: [[ARRAYIDX14_EPIL_1:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3]], i64 [[TMP26]]
; CHECK-A55-NEXT: [[TMP17:%.*]] = load i16, ptr [[ARRAYIDX14_EPIL_1]], align 2
; CHECK-A55-NEXT: [[CONV15_EPIL_1:%.*]] = sext i16 [[TMP17]] to i32
; CHECK-A55-NEXT: [[MUL16_EPIL_1:%.*]] = mul nsw i32 [[CONV15_EPIL_1]], [[CONV_EPIL_1]]
-; CHECK-A55-NEXT: [[ARRAYIDX20_EPIL_1:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1]], i64 [[INDVARS_IV_NEXT_EPIL]]
+; CHECK-A55-NEXT: [[TMP29:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_EPIL]], 2
+; CHECK-A55-NEXT: [[ARRAYIDX20_EPIL_1:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1]], i64 [[TMP29]]
; CHECK-A55-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX20_EPIL_1]], align 4
; CHECK-A55-NEXT: [[ADD21_EPIL_1:%.*]] = add nsw i32 [[MUL16_EPIL_1]], [[TMP18]]
; CHECK-A55-NEXT: store i32 [[ADD21_EPIL_1]], ptr [[ARRAYIDX20_EPIL_1]], align 4
@@ -107,14 +119,16 @@ define void @runtime_unroll_generic(i32 %arg_0, ptr %arg_1, ptr %arg_2, ptr %arg
; CHECK-A55-NEXT: br i1 [[EPIL_ITER_CMP_1_NOT]], label [[FOR_END]], label [[FOR_BODY6_EPIL_2:%.*]]
; CHECK-A55: for.body6.epil.2:
; CHECK-A55-NEXT: [[INDVARS_IV_NEXT_EPIL_1:%.*]] = add nuw nsw i64 [[INDVARS_IV_UNR]], 2
-; CHECK-A55-NEXT: [[ARRAYIDX10_EPIL_2:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2]], i64 [[INDVARS_IV_NEXT_EPIL_1]]
+; CHECK-A55-NEXT: [[TMP31:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_EPIL_1]], 1
+; CHECK-A55-NEXT: [[ARRAYIDX10_EPIL_2:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2]], i64 [[TMP31]]
; CHECK-A55-NEXT: [[TMP19:%.*]] = load i16, ptr [[ARRAYIDX10_EPIL_2]], align 2
; CHECK-A55-NEXT: [[CONV_EPIL_2:%.*]] = sext i16 [[TMP19]] to i32
-; CHECK-A55-NEXT: [[ARRAYIDX14_EPIL_2:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3]], i64 [[INDVARS_IV_NEXT_EPIL_1]]
+; CHECK-A55-NEXT: [[ARRAYIDX14_EPIL_2:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3]], i64 [[TMP31]]
; CHECK-A55-NEXT: [[TMP20:%.*]] = load i16, ptr [[ARRAYIDX14_EPIL_2]], align 2
; CHECK-A55-NEXT: [[CONV15_EPIL_2:%.*]] = sext i16 [[TMP20]] to i32
; CHECK-A55-NEXT: [[MUL16_EPIL_2:%.*]] = mul nsw i32 [[CONV15_EPIL_2]], [[CONV_EPIL_2]]
-; CHECK-A55-NEXT: [[ARRAYIDX20_EPIL_2:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1]], i64 [[INDVARS_IV_NEXT_EPIL_1]]
+; CHECK-A55-NEXT: [[TMP34:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_EPIL_1]], 2
+; CHECK-A55-NEXT: [[ARRAYIDX20_EPIL_2:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1]], i64 [[TMP34]]
; CHECK-A55-NEXT: [[TMP21:%.*]] = load i32, ptr [[ARRAYIDX20_EPIL_2]], align 4
; CHECK-A55-NEXT: [[ADD21_EPIL_2:%.*]] = add nsw i32 [[MUL16_EPIL_2]], [[TMP21]]
; CHECK-A55-NEXT: store i32 [[ADD21_EPIL_2]], ptr [[ARRAYIDX20_EPIL_2]], align 4
@@ -131,14 +145,16 @@ define void @runtime_unroll_generic(i32 %arg_0, ptr %arg_1, ptr %arg_2, ptr %arg
; CHECK-GENERIC-NEXT: br label [[FOR_BODY6:%.*]]
; CHECK-GENERIC: for.body6:
; CHECK-GENERIC-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[FOR_BODY6_PREHEADER]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY6]] ]
-; CHECK-GENERIC-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_2:%.*]], i64 [[INDVARS_IV]]
+; CHECK-GENERIC-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-GENERIC-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_2:%.*]], i64 [[TMP4]]
; CHECK-GENERIC-NEXT: [[TMP0:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
; CHECK-GENERIC-NEXT: [[CONV:%.*]] = sext i16 [[TMP0]] to i32
-; CHECK-GENERIC-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG_3:%.*]], i64 [[INDVARS_IV]]
+; CHECK-GENERIC-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_3:%.*]], i64 [[TMP4]]
; CHECK-GENERIC-NEXT: [[TMP1:%.*]] = load i16, ptr [[ARRAYIDX14]], align 2
; CHECK-GENERIC-NEXT: [[CONV15:%.*]] = sext i16 [[TMP1]] to i32
; CHECK-GENERIC-NEXT: [[MUL16:%.*]] = mul nsw i32 [[CONV15]], [[CONV]]
-; CHECK-GENERIC-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds nuw i32, ptr [[ARG_1:%.*]], i64 [[INDVARS_IV]]
+; CHECK-GENERIC-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 2
+; CHECK-GENERIC-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG_1:%.*]], i64 [[TMP3]]
; CHECK-GENERIC-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX20]], align 4
; CHECK-GENERIC-NEXT: [[ADD21:%.*]] = add nsw i32 [[MUL16]], [[TMP2]]
; CHECK-GENERIC-NEXT: store i32 [[ADD21]], ptr [[ARRAYIDX20]], align 4
diff --git a/llvm/test/Transforms/LoopUnroll/WebAssembly/basic-unrolling.ll b/llvm/test/Transforms/LoopUnroll/WebAssembly/basic-unrolling.ll
index 5e0fc75a6bbb2..d7aab851ae42e 100644
--- a/llvm/test/Transforms/LoopUnroll/WebAssembly/basic-unrolling.ll
+++ b/llvm/test/Transforms/LoopUnroll/WebAssembly/basic-unrolling.ll
@@ -71,28 +71,36 @@ define hidden void @compile_time_partial(ptr nocapture %a, ptr nocapture readonl
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[I_07:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INC_3:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[B:%.*]], i32 [[I_07]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nuw nsw i32 [[I_07]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i32 [[TMP0]]
; CHECK-NEXT: [[I:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT: [[ADD:%.*]] = add i16 [[I]], 1
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i16, ptr [[A:%.*]], i32 [[I_07]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[I_07]], 1
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i32 [[TMP1]]
; CHECK-NEXT: store i16 [[ADD]], ptr [[ARRAYIDX2]], align 2
; CHECK-NEXT: [[INC:%.*]] = or disjoint i32 [[I_07]], 1
-; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds nuw i16, ptr [[B]], i32 [[INC]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i32 [[INC]], 1
+; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i32 [[TMP2]]
; CHECK-NEXT: [[I_1:%.*]] = load i16, ptr [[ARRAYIDX_1]], align 2
; CHECK-NEXT: [[ADD_1:%.*]] = add i16 [[I_1]], 1
-; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds nuw i16, ptr [[A]], i32 [[INC]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i32 [[INC]], 1
+; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i32 [[TMP3]]
; CHECK-NEXT: store i16 [[ADD_1]], ptr [[ARRAYIDX2_1]], align 2
; CHECK-NEXT: [[INC_1:%.*]] = or disjoint i32 [[I_07]], 2
-; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds nuw i16, ptr [[B]], i32 [[INC_1]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i32 [[INC_1]], 1
+; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i32 [[TMP4]]
; CHECK-NEXT: [[I_2:%.*]] = load i16, ptr [[ARRAYIDX_2]], align 2
; CHECK-NEXT: [[ADD_2:%.*]] = add i16 [[I_2]], 1
-; CHECK-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds nuw i16, ptr [[A]], i32 [[INC_1]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i32 [[INC_1]], 1
+; CHECK-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i32 [[TMP5]]
; CHECK-NEXT: store i16 [[ADD_2]], ptr [[ARRAYIDX2_2]], align 2
; CHECK-NEXT: [[INC_2:%.*]] = or disjoint i32 [[I_07]], 3
-; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds nuw i16, ptr [[B]], i32 [[INC_2]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i32 [[INC_2]], 1
+; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i32 [[TMP6]]
; CHECK-NEXT: [[I_3:%.*]] = load i16, ptr [[ARRAYIDX_3]], align 2
; CHECK-NEXT: [[ADD_3:%.*]] = add i16 [[I_3]], 1
-; CHECK-NEXT: [[ARRAYIDX2_3:%.*]] = getelementptr inbounds nuw i16, ptr [[A]], i32 [[INC_2]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i32 [[INC_2]], 1
+; CHECK-NEXT: [[ARRAYIDX2_3:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i32 [[TMP7]]
; CHECK-NEXT: store i16 [[ADD_3]], ptr [[ARRAYIDX2_3]], align 2
; CHECK-NEXT: [[INC_3]] = add nuw nsw i32 [[I_07]], 4
; CHECK-NEXT: [[EXITCOND_NOT_3:%.*]] = icmp eq i32 [[INC_3]], 1000
@@ -135,12 +143,15 @@ define hidden void @runtime(ptr nocapture %a, ptr nocapture readonly %b, ptr noc
; CHECK-NEXT: [[I_09_UNR:%.*]] = phi i32 [ 0, [[FOR_BODY_PREHEADER]] ], [ [[INC_1:%.*]], [[FOR_COND_CLEANUP_LOOPEXIT_UNR_LCSSA:%.*]] ]
; CHECK-NEXT: [[LCMP_MOD1:%.*]] = trunc i32 [[N]] to i1
; CHECK-NEXT: call void @llvm.assume(i1 [[LCMP_MOD1]])
-; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[I_09_UNR]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[I_09_UNR]], 2
+; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP1]]
; CHECK-NEXT: [[I_EPIL:%.*]] = load i32, ptr [[ARRAYIDX_EPIL]], align 4
-; CHECK-NEXT: [[ARRAYIDX1_EPIL:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i32 [[I_09_UNR]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[I_09_UNR]], 2
+; CHECK-NEXT: [[ARRAYIDX1_EPIL:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i32 [[TMP2]]
; CHECK-NEXT: [[I1_EPIL:%.*]] = load i32, ptr [[ARRAYIDX1_EPIL]], align 4
; CHECK-NEXT: [[MUL_EPIL:%.*]] = mul nsw i32 [[I1_EPIL]], [[I_EPIL]]
-; CHECK-NEXT: [[ARRAYIDX2_EPIL:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[I_09_UNR]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i32 [[I_09_UNR]], 2
+; CHECK-NEXT: [[ARRAYIDX2_EPIL:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP3]]
; CHECK-NEXT: store i32 [[MUL_EPIL]], ptr [[ARRAYIDX2_EPIL]], align 4
; CHECK-NEXT: br label [[FOR_COND_CLEANUP]]
; CHECK: for.cond.cleanup:
@@ -148,20 +159,26 @@ define hidden void @runtime(ptr nocapture %a, ptr nocapture readonly %b, ptr noc
; CHECK: for.body:
; CHECK-NEXT: [[I_09:%.*]] = phi i32 [ 0, [[FOR_BODY_PREHEADER_NEW]] ], [ [[INC_1]], [[FOR_BODY]] ]
; CHECK-NEXT: [[NITER:%.*]] = phi i32 [ 0, [[FOR_BODY_PREHEADER_NEW]] ], [ [[NITER_NEXT_1:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[I_09]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i32 [[I_09]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP4]]
; CHECK-NEXT: [[I:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[C]], i32 [[I_09]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i32 [[I_09]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[C]], i32 [[TMP5]]
; CHECK-NEXT: [[I1:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[I1]], [[I]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[I_09]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[I_09]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP6]]
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INC:%.*]] = or disjoint i32 [[I_09]], 1
-; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[INC]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INC]], 2
+; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP7]]
; CHECK-NEXT: [[I_1:%.*]] = load i32, ptr [[ARRAYIDX_1]], align 4
-; CHECK-NEXT: [[ARRAYIDX1_1:%.*]] = getelementptr inbounds i32, ptr [[C]], i32 [[INC]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i32 [[INC]], 2
+; CHECK-NEXT: [[ARRAYIDX1_1:%.*]] = getelementptr inbounds i8, ptr [[C]], i32 [[TMP8]]
; CHECK-NEXT: [[I1_1:%.*]] = load i32, ptr [[ARRAYIDX1_1]], align 4
; CHECK-NEXT: [[MUL_1:%.*]] = mul nsw i32 [[I1_1]], [[I_1]]
-; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[INC]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i32 [[INC]], 2
+; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP9]]
; CHECK-NEXT: store i32 [[MUL_1]], ptr [[ARRAYIDX2_1]], align 4
; CHECK-NEXT: [[INC_1]] = add nuw i32 [[I_09]], 2
; CHECK-NEXT: [[NITER_NEXT_1]] = add i32 [[NITER]], 2
@@ -198,12 +215,15 @@ define hidden void @dont_unroll_call(ptr nocapture %a, ptr nocapture readonly %b
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[I_013:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[I_013]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i32 [[I_013]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP3]]
; CHECK-NEXT: [[I:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i32 [[I_013]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[I_013]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i32 [[TMP1]]
; CHECK-NEXT: [[I1:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[I1]], [[I]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[I_013]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[I_013]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP2]]
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = call i32 (ptr, ...) @printf(ptr noundef nonnull dereferenceable(1) @.str, i32 [[I_013]], i32 [[MUL]])
; CHECK-NEXT: [[INC]] = add nuw i32 [[I_013]], 1
diff --git a/llvm/test/Transforms/LoopUnroll/runtime-multiexit-heuristic.ll b/llvm/test/Transforms/LoopUnroll/runtime-multiexit-heuristic.ll
index 65ef3e40ec229..8cb6e4ce86a04 100644
--- a/llvm/test/Transforms/LoopUnroll/runtime-multiexit-heuristic.ll
+++ b/llvm/test/Transforms/LoopUnroll/runtime-multiexit-heuristic.ll
@@ -32,14 +32,16 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[TMP0]], 42
; CHECK-NEXT: br i1 [[CMP]], label [[OTHEREXIT_LOOPEXIT:%.*]], label [[LATCH:%.*]]
; CHECK: latch:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP3]], [[SUM_02]]
; CHECK-NEXT: br label [[FOR_EXITING_BLOCK_1:%.*]]
; CHECK: for.exiting_block.1:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_1:%.*]]
; CHECK: latch.1:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT]]
; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr i8, ptr [[TMP13]], i64 4
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX_1]], align 4
; CHECK-NEXT: [[ADD_1:%.*]] = add nsw i32 [[TMP4]], [[ADD]]
@@ -47,7 +49,8 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.2:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_2:%.*]]
; CHECK: latch.2:
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_1:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_1]]
; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr i8, ptr [[TMP15]], i64 8
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX_2]], align 4
; CHECK-NEXT: [[ADD_2:%.*]] = add nsw i32 [[TMP5]], [[ADD_1]]
@@ -55,7 +58,8 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.3:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_3:%.*]]
; CHECK: latch.3:
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_2:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_2]]
; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr i8, ptr [[TMP17]], i64 12
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX_3]], align 4
; CHECK-NEXT: [[ADD_3:%.*]] = add nsw i32 [[TMP6]], [[ADD_2]]
@@ -63,7 +67,8 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.4:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_4:%.*]]
; CHECK: latch.4:
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_3:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_3]]
; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr i8, ptr [[TMP18]], i64 16
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX_4]], align 4
; CHECK-NEXT: [[ADD_4:%.*]] = add nsw i32 [[TMP7]], [[ADD_3]]
@@ -71,7 +76,8 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.5:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_5:%.*]]
; CHECK: latch.5:
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_4:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_4]]
; CHECK-NEXT: [[ARRAYIDX_5:%.*]] = getelementptr i8, ptr [[TMP12]], i64 20
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX_5]], align 4
; CHECK-NEXT: [[ADD_5:%.*]] = add nsw i32 [[TMP8]], [[ADD_4]]
@@ -79,7 +85,8 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.6:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_6:%.*]]
; CHECK: latch.6:
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_5:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_5]]
; CHECK-NEXT: [[ARRAYIDX_6:%.*]] = getelementptr i8, ptr [[TMP14]], i64 24
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX_6]], align 4
; CHECK-NEXT: [[ADD_6:%.*]] = add nsw i32 [[TMP9]], [[ADD_5]]
@@ -87,7 +94,8 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.7:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_7]]
; CHECK: latch.7:
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_6:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_6]]
; CHECK-NEXT: [[ARRAYIDX_7:%.*]] = getelementptr i8, ptr [[TMP16]], i64 28
; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX_7]], align 4
; CHECK-NEXT: [[ADD_7]] = add nsw i32 [[TMP10]], [[ADD_6]]
@@ -113,7 +121,8 @@ define i32 @test1(ptr nocapture %a, i64 %n) {
; CHECK-NEXT: [[CMP_EPIL:%.*]] = icmp eq i64 [[TMP0]], 42
; CHECK-NEXT: br i1 [[CMP_EPIL]], label [[OTHEREXIT_LOOPEXIT3:%.*]], label [[LATCH_EPIL]]
; CHECK: latch.epil:
-; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV_EPIL]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDVARS_IV_EPIL]], 2
+; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[ARRAYIDX_EPIL]], align 4
; CHECK-NEXT: [[ADD_EPIL]] = add nsw i32 [[TMP11]], [[SUM_02_EPIL]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_EPIL]] = add i64 [[INDVARS_IV_EPIL]], 1
@@ -337,7 +346,8 @@ define i32 @test2(ptr nocapture %a, i64 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[N:%.*]], 42
; CHECK-NEXT: br i1 [[CMP]], label [[OTHEREXIT:%.*]], label [[LATCH]]
; CHECK: latch:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD]] = add nsw i32 [[TMP0]], [[SUM_02]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -550,7 +560,8 @@ define i32 @test3(ptr nocapture %a, i64 %n) !prof !{!"function_entry_count", i64
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[N:%.*]], 42
; CHECK-NEXT: br i1 [[CMP]], label [[OTHEREXIT:%.*]], label [[LATCH]]
; CHECK: latch:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD]] = add nsw i32 [[TMP0]], [[SUM_02]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -650,7 +661,8 @@ define i32 @test4(ptr nocapture %a, i64 %n) !prof !{!"function_entry_count", i64
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[INDVARS_IV]], 4096
; CHECK-NEXT: br i1 [[CMP]], label [[OTHEREXIT:%.*]], label [[LATCH]]
; CHECK: latch:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD]] = add nsw i32 [[TMP0]], [[SUM_02]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -757,14 +769,16 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i64 [[TMP0]], 42
; CHECK-NEXT: br i1 [[CMP]], label [[OTHEREXIT_LOOPEXIT:%.*]], label [[LATCH:%.*]]
; CHECK: latch:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP3]], [[SUM_02]]
; CHECK-NEXT: br label [[FOR_EXITING_BLOCK_1:%.*]]
; CHECK: for.exiting_block.1:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_1:%.*]]
; CHECK: latch.1:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT]]
; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr i8, ptr [[TMP13]], i64 4
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX_1]], align 4
; CHECK-NEXT: [[ADD_1:%.*]] = add nsw i32 [[TMP4]], [[ADD]]
@@ -772,7 +786,8 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.2:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_2:%.*]]
; CHECK: latch.2:
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_1:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_1]]
; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr i8, ptr [[TMP15]], i64 8
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX_2]], align 4
; CHECK-NEXT: [[ADD_2:%.*]] = add nsw i32 [[TMP5]], [[ADD_1]]
@@ -780,7 +795,8 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.3:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_3:%.*]]
; CHECK: latch.3:
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_2:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_2]]
; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr i8, ptr [[TMP17]], i64 12
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX_3]], align 4
; CHECK-NEXT: [[ADD_3:%.*]] = add nsw i32 [[TMP6]], [[ADD_2]]
@@ -788,7 +804,8 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.4:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_4:%.*]]
; CHECK: latch.4:
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_3:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_3]]
; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr i8, ptr [[TMP18]], i64 16
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX_4]], align 4
; CHECK-NEXT: [[ADD_4:%.*]] = add nsw i32 [[TMP7]], [[ADD_3]]
@@ -796,7 +813,8 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.5:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_5:%.*]]
; CHECK: latch.5:
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_4:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_4]]
; CHECK-NEXT: [[ARRAYIDX_5:%.*]] = getelementptr i8, ptr [[TMP12]], i64 20
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX_5]], align 4
; CHECK-NEXT: [[ADD_5:%.*]] = add nsw i32 [[TMP8]], [[ADD_4]]
@@ -804,7 +822,8 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.6:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_6:%.*]]
; CHECK: latch.6:
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_5:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_5]]
; CHECK-NEXT: [[ARRAYIDX_6:%.*]] = getelementptr i8, ptr [[TMP14]], i64 24
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX_6]], align 4
; CHECK-NEXT: [[ADD_6:%.*]] = add nsw i32 [[TMP9]], [[ADD_5]]
@@ -812,7 +831,8 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK: for.exiting_block.7:
; CHECK-NEXT: br i1 false, label [[OTHEREXIT_LOOPEXIT]], label [[LATCH_7]]
; CHECK: latch.7:
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_6:%.*]] = shl i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDVARS_IV_NEXT_6]]
; CHECK-NEXT: [[ARRAYIDX_7:%.*]] = getelementptr i8, ptr [[TMP16]], i64 28
; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX_7]], align 4
; CHECK-NEXT: [[ADD_7]] = add nsw i32 [[TMP10]], [[ADD_6]]
@@ -838,7 +858,8 @@ define i32 @test5(ptr nocapture %a, i64 %n) {
; CHECK-NEXT: [[CMP_EPIL:%.*]] = icmp eq i64 [[TMP0]], 42
; CHECK-NEXT: br i1 [[CMP_EPIL]], label [[OTHEREXIT_LOOPEXIT3:%.*]], label [[LATCH_EPIL]]
; CHECK: latch.epil:
-; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV_EPIL]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDVARS_IV_EPIL]], 2
+; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[ARRAYIDX_EPIL]], align 4
; CHECK-NEXT: [[ADD_EPIL]] = add nsw i32 [[TMP11]], [[SUM_02_EPIL]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_EPIL]] = add i64 [[INDVARS_IV_EPIL]], 1
diff --git a/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll b/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll
index 5f4bbf15f01af..5106ac37eb541 100644
--- a/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll
+++ b/llvm/test/Transforms/LoopUnroll/runtime-unroll-remainder.ll
@@ -25,9 +25,11 @@ define i32 @unroll(ptr nocapture readonly %a, ptr nocapture readonly %b, i32 %N)
; CHECK-NEXT: call void @llvm.assume(i1 [[LCMP_MOD2]])
; CHECK-NEXT: br label [[FOR_BODY_EPIL:%.*]]
; CHECK: for.body.epil:
-; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV_EPIL_INIT]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[INDVARS_IV_EPIL_INIT]], 2
+; CHECK-NEXT: [[ARRAYIDX_EPIL:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP16]]
; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX_EPIL]], align 4
-; CHECK-NEXT: [[ARRAYIDX2_EPIL:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDVARS_IV_EPIL_INIT]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDVARS_IV_EPIL_INIT]], 2
+; CHECK-NEXT: [[ARRAYIDX2_EPIL:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP18]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX2_EPIL]], align 4
; CHECK-NEXT: [[MUL_EPIL:%.*]] = mul nsw i32 [[TMP2]], [[TMP1]]
; CHECK-NEXT: [[ADD_EPIL:%.*]] = add nsw i32 [[MUL_EPIL]], [[C_010_EPIL_INIT]]
@@ -35,9 +37,11 @@ define i32 @unroll(ptr nocapture readonly %a, ptr nocapture readonly %b, i32 %N)
; CHECK-NEXT: br i1 [[EPIL_ITER_CMP_NOT]], label [[FOR_COND_CLEANUP_LOOPEXIT_EPILOG_LCSSA:%.*]], label [[FOR_BODY_EPIL_1:%.*]]
; CHECK: for.body.epil.1:
; CHECK-NEXT: [[INDVARS_IV_NEXT_EPIL:%.*]] = add nuw nsw i64 [[INDVARS_IV_EPIL_INIT]], 1
-; CHECK-NEXT: [[ARRAYIDX_EPIL_1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV_NEXT_EPIL]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_EPIL]], 2
+; CHECK-NEXT: [[ARRAYIDX_EPIL_1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX_EPIL_1]], align 4
-; CHECK-NEXT: [[ARRAYIDX2_EPIL_1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV_NEXT_EPIL]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_EPIL]], 2
+; CHECK-NEXT: [[ARRAYIDX2_EPIL_1:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP22]]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX2_EPIL_1]], align 4
; CHECK-NEXT: [[MUL_EPIL_1:%.*]] = mul nsw i32 [[TMP4]], [[TMP3]]
; CHECK-NEXT: [[ADD_EPIL_1:%.*]] = add nsw i32 [[MUL_EPIL_1]], [[ADD_EPIL]]
@@ -45,9 +49,11 @@ define i32 @unroll(ptr nocapture readonly %a, ptr nocapture readonly %b, i32 %N)
; CHECK-NEXT: br i1 [[EPIL_ITER_CMP_1_NOT]], label [[FOR_COND_CLEANUP_LOOPEXIT_EPILOG_LCSSA]], label [[FOR_BODY_EPIL_2:%.*]]
; CHECK: for.body.epil.2:
; CHECK-NEXT: [[INDVARS_IV_NEXT_EPIL_1:%.*]] = add nuw nsw i64 [[INDVARS_IV_EPIL_INIT]], 2
-; CHECK-NEXT: [[ARRAYIDX_EPIL_2:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV_NEXT_EPIL_1]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_EPIL_1]], 2
+; CHECK-NEXT: [[ARRAYIDX_EPIL_2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP24]]
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX_EPIL_2]], align 4
-; CHECK-NEXT: [[ARRAYIDX2_EPIL_2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV_NEXT_EPIL_1]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_EPIL_1]], 2
+; CHECK-NEXT: [[ARRAYIDX2_EPIL_2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP26]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX2_EPIL_2]], align 4
; CHECK-NEXT: [[MUL_EPIL_2:%.*]] = mul nsw i32 [[TMP6]], [[TMP5]]
; CHECK-NEXT: [[ADD_EPIL_2:%.*]] = add nsw i32 [[MUL_EPIL_2]], [[ADD_EPIL_1]]
@@ -65,30 +71,38 @@ define i32 @unroll(ptr nocapture readonly %a, ptr nocapture readonly %b, i32 %N)
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[FOR_BODY_LR_PH_NEW]] ], [ [[INDVARS_IV_NEXT_3]], [[FOR_BODY]] ]
; CHECK-NEXT: [[C_010:%.*]] = phi i32 [ 0, [[FOR_BODY_LR_PH_NEW]] ], [ [[ADD_3]], [[FOR_BODY]] ]
; CHECK-NEXT: [[NITER:%.*]] = phi i64 [ 0, [[FOR_BODY_LR_PH_NEW]] ], [ [[NITER_NEXT_3:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP28]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP15]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], [[TMP7]]
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[MUL]], [[C_010]]
; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = or disjoint i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT]], 2
+; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP17]]
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX_1]], align 4
-; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT]], 2
+; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[ARRAYIDX2_1]], align 4
; CHECK-NEXT: [[MUL_1:%.*]] = mul nsw i32 [[TMP10]], [[TMP9]]
; CHECK-NEXT: [[ADD_1:%.*]] = add nsw i32 [[MUL_1]], [[ADD]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_1:%.*]] = or disjoint i64 [[INDVARS_IV]], 2
-; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[INDVARS_IV_NEXT_1]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_1]], 2
+; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[ARRAYIDX_2]], align 4
-; CHECK-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[INDVARS_IV_NEXT_1]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_1]], 2
+; CHECK-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP23]]
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX2_2]], align 4
; CHECK-NEXT: [[MUL_2:%.*]] = mul nsw i32 [[TMP12]], [[TMP11]]
; CHECK-NEXT: [[ADD_2:%.*]] = add nsw i32 [[MUL_2]], [[ADD_1]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_2:%.*]] = or disjoint i64 [[INDVARS_IV]], 3
-; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_2]], 2
+; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX_3]], align 4
-; CHECK-NEXT: [[ARRAYIDX2_3:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_2]], 2
+; CHECK-NEXT: [[ARRAYIDX2_3:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP27]]
; CHECK-NEXT: [[TMP14:%.*]] = load i32, ptr [[ARRAYIDX2_3]], align 4
; CHECK-NEXT: [[MUL_3:%.*]] = mul nsw i32 [[TMP14]], [[TMP13]]
; CHECK-NEXT: [[ADD_3]] = add nsw i32 [[MUL_3]], [[ADD_2]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/aarch64-predication.ll b/llvm/test/Transforms/LoopVectorize/AArch64/aarch64-predication.ll
index ddfdb257ed49a..3d9cdc7ce0057 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/aarch64-predication.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/aarch64-predication.ll
@@ -23,7 +23,8 @@ define i64 @predicated_udiv_scalarized_operand(ptr %a, i64 %x) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_UDIV_CONTINUE2:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x i64> [ zeroinitializer, [[ENTRY]] ], [ [[TMP17:%.*]], [[PRED_UDIV_CONTINUE2]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt <2 x i64> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <2 x i1> [[TMP2]], i64 0
@@ -84,3 +85,5 @@ for.end:
%var7 = phi i64 [ %var6, %for.inc ]
ret i64 %var7
}
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; COST: {{.*}}
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 3f32bbe1f0102..a730f7793a2d7 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-cond-inv-loads.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-cond-inv-loads.ll
@@ -19,12 +19,14 @@ define void @cond_inv_load_i32i32i16(ptr noalias nocapture %a, ptr noalias nocap
; 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 [[COND:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[COND:%.*]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp ne <vscale x 4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> align 2 [[BROADCAST_SPLAT]], <vscale x 4 x i1> [[TMP6]], <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:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP11]]
; CHECK-NEXT: call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP7]], ptr align 4 [[TMP8]], <vscale x 4 x i1> [[TMP6]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP4]]
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -77,11 +79,13 @@ define void @cond_inv_load_f64f64f64(ptr noalias nocapture %a, ptr noalias nocap
; 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 double, ptr [[COND:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[COND:%.*]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x double>, ptr [[TMP5]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = fcmp ogt <vscale x 4 x double> [[WIDE_LOAD]], splat (double 4.000000e-01)
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x double> @llvm.masked.gather.nxv4f64.nxv4p0(<vscale x 4 x ptr> align 8 [[BROADCAST_SPLAT]], <vscale x 4 x i1> [[TMP6]], <vscale x 4 x double> poison)
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr double, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP10]]
; CHECK-NEXT: call void @llvm.masked.store.nxv4f64.p0(<vscale x 4 x double> [[WIDE_MASKED_GATHER]], ptr align 8 [[TMP7]], <vscale x 4 x i1> [[TMP6]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP4]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -134,14 +138,17 @@ 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: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[COND:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[COND:%.*]], i64 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = icmp ne <vscale x 4 x i32> [[WIDE_LOAD]], zeroinitializer
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP13]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP8]], <vscale x 4 x i1> [[TMP7]], <vscale x 4 x i32> poison)
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[BROADCAST_SPLAT]], <vscale x 4 x i1> [[TMP7]], <vscale x 4 x i32> poison)
; CHECK-NEXT: [[TMP9:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_MASKED_GATHER]], [[WIDE_MASKED_LOAD]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP14]]
; CHECK-NEXT: call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP9]], ptr align 4 [[TMP10]], <vscale x 4 x i1> [[TMP7]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP4]]
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-inductions.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-inductions.ll
index 337c097c85712..53af5a610da84 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-inductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-inductions.ll
@@ -31,9 +31,11 @@ define void @cond_ind64(ptr noalias nocapture %a, ptr noalias nocapture readonly
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP6]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP9:%.*]] = trunc <vscale x 4 x i64> [[VEC_IND]] to <vscale x 4 x i1>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP10]], <vscale x 4 x i1> [[TMP9]], <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP8]]
; CHECK-NEXT: call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[WIDE_MASKED_LOAD]], ptr align 4 [[TMP11]], <vscale x 4 x i1> [[TMP9]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], [[DOTSPLAT]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll
index 051db220996c5..c3df04f51024f 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll
@@ -35,15 +35,16 @@ define void @test_array_load2_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: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr @AB, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr @AB, i64 [[TMP5]]
; 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.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:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 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 i32, ptr @CD, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr @CD, i64 [[DOTIDX]]
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.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 [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
@@ -114,10 +115,12 @@ define void @test_array_load2_i16_store2(i32 %C, i32 %D) #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> [ [[TMP3]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i16, ptr @AB_i16, <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr @AB_i16, <vscale x 4 x i64> [[TMP5]]
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> align 2 [[TMP6]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i16> poison)
-; CHECK-NEXT: [[TMP7:%.*]] = or disjoint <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i16, ptr @AB_i16, <vscale x 4 x i64> [[TMP7]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP13:%.*]] = or disjoint <vscale x 4 x i64> [[TMP7]], splat (i64 2)
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr @AB_i16, <vscale x 4 x i64> [[TMP13]]
; CHECK-NEXT: [[WIDE_MASKED_GATHER1:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> align 2 [[TMP8]], <vscale x 4 x i1> splat (i1 true), <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]]
@@ -204,14 +207,16 @@ define void @test_array_load2_store2_i16(i32 noundef %C, i32 noundef %D) #1 {
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
; 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]], splat (i64 1)
; 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 i16, ptr @CD_i16, <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nuw nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr @CD_i16, <vscale x 4 x i64> [[TMP17]]
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i16.nxv4p0(<vscale x 4 x i16> [[TMP11]], <vscale x 4 x ptr> align 2 [[TMP12]], <vscale x 4 x i1> splat (i1 true))
; 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 i16, ptr @CD_i16, <vscale x 4 x i64> [[TMP9]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nuw <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP19:%.*]] = or disjoint <vscale x 4 x i64> [[TMP18]], splat (i64 2)
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr @CD_i16, <vscale x 4 x i64> [[TMP19]]
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i16.nxv4p0(<vscale x 4 x i16> [[TMP14]], <vscale x 4 x ptr> align 2 [[TMP15]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], [[BROADCAST_SPLAT4]]
@@ -267,21 +272,27 @@ define i32 @test_struct_load6(ptr %S) #1 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]]
+; CHECK-NEXT: [[TMP3:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 24)
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], <vscale x 4 x i64> [[TMP3]]
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP5]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP18:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 24)
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[S]], <vscale x 4 x i64> [[TMP18]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT]], i64 4
; CHECK-NEXT: [[WIDE_MASKED_GATHER1:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP6]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP19:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 24)
+; CHECK-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds i8, ptr [[S]], <vscale x 4 x i64> [[TMP19]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT6]], i64 8
; CHECK-NEXT: [[WIDE_MASKED_GATHER2:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP7]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[DOTSPLIT7:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP20:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 24)
+; CHECK-NEXT: [[DOTSPLIT7:%.*]] = getelementptr inbounds i8, ptr [[S]], <vscale x 4 x i64> [[TMP20]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT7]], i64 12
; CHECK-NEXT: [[WIDE_MASKED_GATHER3:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP8]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[DOTSPLIT8:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP21:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 24)
+; CHECK-NEXT: [[DOTSPLIT8:%.*]] = getelementptr inbounds i8, ptr [[S]], <vscale x 4 x i64> [[TMP21]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT8]], i64 16
; CHECK-NEXT: [[WIDE_MASKED_GATHER4:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP9]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[DOTSPLIT9:%.*]] = getelementptr inbounds [[STRUCT_ST6]], ptr [[S]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP22:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 24)
+; CHECK-NEXT: [[DOTSPLIT9:%.*]] = getelementptr inbounds i8, ptr [[S]], <vscale x 4 x i64> [[TMP22]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT9]], i64 20
; CHECK-NEXT: [[WIDE_MASKED_GATHER5:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP10]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
; CHECK-NEXT: [[TMP11:%.*]] = add <vscale x 4 x i32> [[WIDE_MASKED_GATHER]], [[VEC_PHI]]
@@ -370,10 +381,12 @@ define void @test_reversed_load2_store2(ptr noalias nocapture readonly %A, ptr n
; 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: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_ST2:%.*]], ptr [[A:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[OFFSET_IDX]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP0]], 3
; CHECK-NEXT: [[TMP8:%.*]] = sub nsw i64 2, [[TMP6]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[TMP8]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i64 [[TMP14]]
; 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.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
; CHECK-NEXT: [[TMP10:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
@@ -382,8 +395,10 @@ define void @test_reversed_load2_store2(ptr noalias nocapture readonly %A, ptr n
; CHECK-NEXT: [[REVERSE1:%.*]] = call <vscale x 4 x i32> @llvm.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]]
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[OFFSET_IDX]], 3
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP8]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[TMP16]], i64 [[TMP17]]
; CHECK-NEXT: [[REVERSE2:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP12]])
; CHECK-NEXT: [[REVERSE3:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP13]])
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[REVERSE2]], <vscale x 4 x i32> [[REVERSE3]])
@@ -449,8 +464,8 @@ define void @even_load_static_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.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:%.*]] = shl nsw <vscale x 4 x i32> [[TMP5]], splat (i32 1)
-; CHECK-NEXT: [[TMP7:%.*]] = and i64 [[INDEX]], 9223372036854775804
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[TMP7]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP6]], ptr [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -518,8 +533,8 @@ define void @even_load_dynamic_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.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: [[TMP14:%.*]] = shl nsw <vscale x 4 x i32> [[TMP13]], splat (i32 1)
-; CHECK-NEXT: [[TMP15:%.*]] = and i64 [[INDEX]], 9223372036854775804
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[TMP15]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP14]], ptr [[TMP16]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -585,8 +600,10 @@ define void @load_gap_reverse(ptr noalias nocapture readonly %P1, ptr noalias no
; 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: [[TMP4:%.*]] = add nsw <vscale x 4 x i64> [[BROADCAST_SPLAT1]], [[VEC_IND]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[PAIR:%.*]], ptr [[P1:%.*]], <vscale x 4 x i64> [[VEC_IND]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P2:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[P1:%.*]], <vscale x 4 x i64> [[TMP9]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P2:%.*]], <vscale x 4 x i64> [[TMP10]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT]], i64 8
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i64> @llvm.masked.gather.nxv4i64.nxv4p0(<vscale x 4 x ptr> align 8 [[TMP6]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i64> poison)
; CHECK-NEXT: [[TMP7:%.*]] = sub nsw <vscale x 4 x i64> [[WIDE_MASKED_GATHER]], [[VEC_IND]]
@@ -640,15 +657,17 @@ define void @mixed_load2_store2(ptr noalias nocapture readonly %A, ptr noalias n
; 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: [[TMP5:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; 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.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:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 1
; CHECK-NEXT: [[TMP6:%.*]] = mul nsw <vscale x 4 x i32> [[TMP4]], [[TMP3]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
+; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[DOTIDX]]
+; CHECK-NEXT: [[WIDE_VEC1:%.*]] = load <vscale x 8 x i32>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC1]])
; 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]]
@@ -728,7 +747,8 @@ define void @int_float_struct(ptr nocapture readonly %p) #0 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <vscale x 4 x float> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <vscale x 4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP6:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[STRUCT_INTFLOAT:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP9]]
; 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.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
@@ -809,9 +829,11 @@ define void @PR27626_0(ptr %p, 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> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 3)
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], <vscale x 4 x i64> [[TMP8]]
; CHECK-NEXT: [[TMP14:%.*]] = extractelement <vscale x 4 x ptr> [[TMP12]], i64 0
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 3)
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], <vscale x 4 x i64> [[TMP11]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT]], i64 4
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> align 4 [[TMP12]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP14]], align 4
@@ -879,8 +901,10 @@ define i32 @PR27626_1(ptr %p, i64 %n) #1 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 3)
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], <vscale x 4 x i64> [[TMP10]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT]], i64 4
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <vscale x 4 x ptr> [[TMP13]], i64 0
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP12]], align 4
@@ -956,9 +980,11 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) #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> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 3)
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], <vscale x 4 x i64> [[TMP8]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 -8
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 3)
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], <vscale x 4 x i64> [[TMP11]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT]], i64 4
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> align 4 [[TMP12]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 8 x i32>, ptr [[TMP13]], align 4
@@ -1027,11 +1053,14 @@ define i32 @PR27626_3(ptr %p, i64 %n, i32 %z) #1 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], splat (i64 1)
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP10]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], <vscale x 4 x i64> [[TMP12]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl <vscale x 4 x i64> [[VEC_IND]], splat (i64 3)
+; CHECK-NEXT: [[TMP21:%.*]] = add <vscale x 4 x i64> [[TMP12]], splat (i64 8)
+; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds i8, ptr [[P]], <vscale x 4 x i64> [[TMP21]]
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, <vscale x 4 x ptr> [[DOTSPLIT3]], i64 4
; 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.vector.deinterleave2.nxv8i32(<vscale x 8 x i32> [[WIDE_VEC]])
@@ -1117,7 +1146,8 @@ define void @PR27626_4(ptr %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> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 2)
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], <vscale x 4 x i64> [[TMP12]]
; CHECK-NEXT: [[P:%.*]] = extractelement <vscale x 4 x ptr> [[TMP13]], i64 0
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> align 4 [[TMP13]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = call <vscale x 8 x i32> @llvm.vector.interleave2.nxv8i32(<vscale x 4 x i32> [[BROADCAST_SPLAT2]], <vscale x 4 x i32> [[BROADCAST_SPLAT4]])
@@ -1195,11 +1225,14 @@ define void @PR27626_5(ptr %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: [[TMP13:%.*]] = add nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 -1)
-; CHECK-NEXT: [[TMP14:%.*]] = add nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 -3)
-; 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: [[TMP13:%.*]] = shl nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 2)
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], <vscale x 4 x i64> [[TMP13]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl <vscale x 4 x i64> [[VEC_IND]], splat (i64 2)
+; CHECK-NEXT: [[TMP20:%.*]] = add <vscale x 4 x i64> [[TMP14]], splat (i64 -4)
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[A]], <vscale x 4 x i64> [[TMP20]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl <vscale x 4 x i64> [[VEC_IND]], splat (i64 2)
+; CHECK-NEXT: [[TMP22:%.*]] = add <vscale x 4 x i64> [[TMP21]], splat (i64 -12)
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[A]], <vscale x 4 x i64> [[TMP22]]
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> align 4 [[TMP16]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT2]], <vscale x 4 x ptr> align 4 [[TMP17]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT4]], <vscale x 4 x ptr> align 4 [[TMP15]], <vscale x 4 x i1> splat (i1 true))
@@ -1284,20 +1317,23 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) #1 {
; 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> [ [[TMP15]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP18:%.*]] = or disjoint <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
-; CHECK-NEXT: [[TMP19:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 2)
-; 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> align 4 [[TMP20]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i16> poison), !alias.scope [[META34:![0-9]+]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nuw <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP18:%.*]] = or disjoint <vscale x 4 x i64> [[TMP17]], splat (i64 2)
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, 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> align 4 [[TMP19]], <vscale x 4 x i1> splat (i1 true), <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> align 4 [[TMP22]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i16> poison), !alias.scope [[META34]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl nuw <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP22:%.*]] = add <vscale x 4 x i64> [[TMP28]], splat (i64 4)
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[A]], <vscale x 4 x i64> [[TMP22]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER4]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP35]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i16> poison), !alias.scope [[META34]]
; CHECK-NEXT: [[TMP23:%.*]] = call <vscale x 4 x i16> @llvm.vector.splice.right.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: [[TMP36:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP36]]
+; CHECK-NEXT: store <vscale x 4 x i32> [[TMP27]], ptr [[TMP37]], align 4, !alias.scope [[META37:![0-9]+]], !noalias [[META34]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP10]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -1370,12 +1406,15 @@ define void @interleave_deinterleave_factor3(ptr writeonly noalias %dst, ptr rea
; 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> [ [[TMP6]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT_XYZ:%.*]], ptr [[A:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP4:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 12)
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], <vscale x 4 x i64> [[TMP4]]
; CHECK-NEXT: [[TMP7:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP19]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT_XYZ]], ptr [[B:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP26:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 12)
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], <vscale x 4 x i64> [[TMP26]]
; CHECK-NEXT: [[TMP11:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP20]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
; CHECK-NEXT: [[TMP14:%.*]] = add nsw <vscale x 4 x i32> [[TMP11]], [[TMP7]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_XYZ]], ptr [[DST:%.*]], <vscale x 4 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP27:%.*]] = mul nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 12)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], <vscale x 4 x i64> [[TMP27]]
; CHECK-NEXT: call void @llvm.masked.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP14]], <vscale x 4 x ptr> align 4 [[TMP10]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw i8, <vscale x 4 x ptr> [[TMP19]], i64 4
; CHECK-NEXT: [[TMP8:%.*]] = call <vscale x 4 x i32> @llvm.masked.gather.nxv4i32.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP21]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i32> poison)
@@ -1458,14 +1497,16 @@ define void @interleave_deinterleave(ptr writeonly noalias %dst, ptr readonly %a
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_XYZT:%.*]], ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 16 x i32>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave4.nxv16i32(<vscale x 16 x i32> [[WIDE_VEC]])
; CHECK-NEXT: [[TMP9:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <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>, <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>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 2
; CHECK-NEXT: [[TMP12:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 3
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_XYZT]], ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP14]]
; CHECK-NEXT: [[WIDE_VEC8:%.*]] = load <vscale x 16 x i32>, ptr [[TMP13]], align 4
; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave4.nxv16i32(<vscale x 16 x i32> [[WIDE_VEC8]])
; CHECK-NEXT: [[TMP16:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 0
@@ -1473,7 +1514,8 @@ define void @interleave_deinterleave(ptr writeonly noalias %dst, ptr readonly %a
; CHECK-NEXT: [[TMP18:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 2
; CHECK-NEXT: [[TMP19:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC2]], 3
; CHECK-NEXT: [[TMP20:%.*]] = add nsw <vscale x 4 x i32> [[TMP16]], [[TMP9]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_XYZT]], ptr [[DST:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], i64 [[TMP26]]
; CHECK-NEXT: [[TMP22:%.*]] = sub nsw <vscale x 4 x i32> [[TMP10]], [[TMP17]]
; CHECK-NEXT: [[TMP23:%.*]] = shl <vscale x 4 x i32> [[TMP11]], [[TMP18]]
; CHECK-NEXT: [[TMP24:%.*]] = ashr <vscale x 4 x i32> [[TMP12]], [[TMP19]]
@@ -1559,10 +1601,12 @@ define void @interleave_deinterleave_reverse(ptr noalias nocapture readonly %A,
; 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: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_XYZT:%.*]], ptr [[A:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[OFFSET_IDX]], 4
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP0]], 4
; CHECK-NEXT: [[TMP9:%.*]] = sub nsw i64 4, [[TMP6]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP5]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[TMP9]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i64 [[TMP11]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <vscale x 16 x i32>, ptr [[TMP10]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.vector.deinterleave4.nxv16i32(<vscale x 16 x i32> [[WIDE_VEC]])
; CHECK-NEXT: [[TMP13:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[STRIDED_VEC]], 0
@@ -1577,8 +1621,10 @@ define void @interleave_deinterleave_reverse(ptr noalias nocapture readonly %A,
; CHECK-NEXT: [[TMP18:%.*]] = sub nsw <vscale x 4 x i32> [[REVERSE3]], [[VEC_IND]]
; CHECK-NEXT: [[TMP19:%.*]] = mul nsw <vscale x 4 x i32> [[REVERSE4]], [[VEC_IND]]
; CHECK-NEXT: [[TMP20:%.*]] = shl nuw nsw <vscale x 4 x i32> [[REVERSE5]], [[VEC_IND]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_XYZT]], ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[OFFSET_IDX]], 4
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[TMP9]], 2
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP23]], i64 [[TMP21]]
; CHECK-NEXT: [[REVERSE6:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP17]])
; CHECK-NEXT: [[REVERSE7:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP18]])
; CHECK-NEXT: [[REVERSE8:%.*]] = call <vscale x 4 x i32> @llvm.vector.reverse.nxv4i32(<vscale x 4 x i32> [[TMP19]])
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll
index 1218f68ac369a..69abc41c3bc64 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-overflow-checks.ll
@@ -22,10 +22,12 @@ define void @cannot_overflow_i32_induction_var(ptr noalias %dst, ptr readonly %s
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[SRC:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP0]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
; CHECK-NEXT: [[TMP1:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], splat (i32 42)
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[DST:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], i64 [[TMP7]]
; CHECK-NEXT: call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP1]], ptr align 4 [[TMP2]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX_NEXT]], i64 [[WIDE_TRIP_COUNT]])
@@ -79,10 +81,12 @@ define void @can_overflow_i64_induction_var(ptr noalias %dst, ptr readonly %src,
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[SRC:%.*]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP3]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
; CHECK-NEXT: [[TMP4:%.*]] = add nsw <vscale x 4 x i32> [[WIDE_MASKED_LOAD]], splat (i32 42)
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[DST:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[DST:%.*]], i64 [[TMP10]]
; CHECK-NEXT: call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP4]], ptr align 4 [[TMP5]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX_NEXT]], i64 [[N]])
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-vector-reverse.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-vector-reverse.ll
index a2367f5d89da8..56735a942e515 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-vector-reverse.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-vector-reverse.ll
@@ -29,18 +29,24 @@ define void @vector_reverse_f64(i64 %N, ptr noalias %a, ptr noalias %b) #0{
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP7:%.*]] = xor i64 [[INDEX]], -1
; CHECK-NEXT: [[TMP8:%.*]] = add i64 [[N]], [[TMP7]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds double, ptr [[B:%.*]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[TMP8]], 3
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP9]]
; CHECK-NEXT: [[TMP12:%.*]] = sub i64 1, [[TMP5]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds double, ptr [[TMP9]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[TMP12]], 3
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 [[TMP10]]
; CHECK-NEXT: [[TMP22:%.*]] = sub i64 [[TMP12]], [[TMP5]]
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds double, ptr [[TMP9]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[TMP22]], 3
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i64 [[TMP13]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x double>, ptr [[TMP14]], align 8
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 8 x double>, ptr [[TMP15]], align 8
; CHECK-NEXT: [[TMP16:%.*]] = fadd <vscale x 8 x double> [[WIDE_LOAD]], splat (double 1.000000e+00)
; CHECK-NEXT: [[TMP17:%.*]] = fadd <vscale x 8 x double> [[WIDE_LOAD1]], splat (double 1.000000e+00)
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds double, ptr [[A:%.*]], i64 [[TMP8]]
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds double, ptr [[TMP23]], i64 [[TMP12]]
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds double, ptr [[TMP23]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[TMP8]], 3
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP23]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[TMP12]], 3
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[TMP18]], i64 [[TMP19]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[TMP22]], 3
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[TMP18]], i64 [[TMP21]]
; CHECK-NEXT: store <vscale x 8 x double> [[TMP16]], ptr [[TMP20]], align 8
; CHECK-NEXT: store <vscale x 8 x double> [[TMP17]], ptr [[TMP24]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
@@ -100,18 +106,24 @@ define void @vector_reverse_i64(i64 %N, ptr %a, ptr %b) #0 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP10:%.*]] = xor i64 [[INDEX]], -1
; CHECK-NEXT: [[TMP11:%.*]] = add i64 [[N]], [[TMP10]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[TMP11]], 3
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP12]]
; CHECK-NEXT: [[TMP15:%.*]] = sub i64 1, [[TMP8]]
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i64, ptr [[TMP12]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[TMP15]], 3
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[TMP14]], i64 [[TMP13]]
; CHECK-NEXT: [[TMP25:%.*]] = sub i64 [[TMP15]], [[TMP8]]
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i64, ptr [[TMP12]], i64 [[TMP25]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[TMP25]], 3
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP14]], i64 [[TMP16]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 8 x i64>, ptr [[TMP17]], align 8
; CHECK-NEXT: [[WIDE_LOAD3:%.*]] = load <vscale x 8 x i64>, ptr [[TMP18]], align 8
; CHECK-NEXT: [[TMP19:%.*]] = add <vscale x 8 x i64> [[WIDE_LOAD]], splat (i64 1)
; CHECK-NEXT: [[TMP20:%.*]] = add <vscale x 8 x i64> [[WIDE_LOAD3]], splat (i64 1)
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i64, ptr [[TMP26]], i64 [[TMP15]]
-; CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[TMP26]], i64 [[TMP25]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl nsw i64 [[TMP11]], 3
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP26]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[TMP15]], 3
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[TMP21]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl nsw i64 [[TMP25]], 3
+; CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[TMP21]], i64 [[TMP24]]
; CHECK-NEXT: store <vscale x 8 x i64> [[TMP19]], ptr [[TMP23]], align 8
; CHECK-NEXT: store <vscale x 8 x i64> [[TMP20]], ptr [[TMP27]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP9]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-vfabi.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-vfabi.ll
index 8b6c7fe8f0fa5..1fe73feef10f0 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-vfabi.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-vfabi.ll
@@ -13,17 +13,19 @@ define void @test_big_little_params(ptr readonly %a, ptr readonly %b, ptr noalia
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ splat (i1 true), [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP2]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8.p0(ptr align 1 [[TMP3]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i8> poison)
-; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 4 x i32> @foo_vector(<vscale x 4 x i32> [[WIDE_MASKED_LOAD]], <vscale x 4 x i8> [[WIDE_MASKED_LOAD1]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[INDEX]]
-; CHECK-NEXT: call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP4]], ptr align 4 [[TMP5]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP3]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <vscale x 4 x i8> @llvm.masked.load.nxv4i8.p0(ptr align 1 [[TMP4]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i8> poison)
+; CHECK-NEXT: [[TMP5:%.*]] = call <vscale x 4 x i32> @foo_vector(<vscale x 4 x i32> [[WIDE_MASKED_LOAD]], <vscale x 4 x i8> [[WIDE_MASKED_LOAD1]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[C]], i64 [[TMP6]]
+; CHECK-NEXT: call void @llvm.masked.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP5]], ptr align 4 [[TMP7]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX_NEXT]], i64 1025)
-; CHECK-NEXT: [[TMP6:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
-; CHECK-NEXT: br i1 [[TMP6]], label [[VECTOR_BODY]], label [[EXIT:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK-NEXT: [[TMP8:%.*]] = extractelement <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT: br i1 [[TMP8]], label [[VECTOR_BODY]], label [[EXIT:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
@@ -57,17 +59,20 @@ define void @test_little_big_params(ptr readonly %a, ptr readonly %b, ptr noalia
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 2 x i1> [ splat (i1 true), [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x float> @llvm.masked.load.nxv2f32.p0(ptr align 4 [[TMP2]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x float> poison)
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr double, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP3]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
-; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x double> @bar_vector(<vscale x 2 x float> [[WIDE_MASKED_LOAD]], <vscale x 2 x double> [[WIDE_MASKED_LOAD1]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds double, ptr [[C]], i64 [[INDEX]]
-; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP4]], ptr align 8 [[TMP5]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP2]]
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x float> @llvm.masked.load.nxv2f32.p0(ptr align 4 [[TMP3]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x float> poison)
+; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP4]]
+; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP5]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
+; CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 2 x double> @bar_vector(<vscale x 2 x float> [[WIDE_MASKED_LOAD]], <vscale x 2 x double> [[WIDE_MASKED_LOAD1]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[C]], i64 [[TMP7]]
+; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP6]], ptr align 8 [[TMP8]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX_NEXT]], i64 1025)
-; CHECK-NEXT: [[TMP6:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
-; CHECK-NEXT: br i1 [[TMP6]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK-NEXT: [[TMP9:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT: br i1 [[TMP9]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll
index dca0065eb854c..f89d68fe6275a 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll
@@ -23,7 +23,6 @@ define void @widen_ptr_phi_unrolled(ptr noalias nocapture %a, ptr noalias nocapt
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[TMP4]], 2
; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i64 [[TMP4]], 3
; CHECK-NEXT: [[DOTNOT:%.*]] = sub nsw i64 0, [[TMP5]]
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[N]], [[DOTNOT]]
@@ -47,14 +46,18 @@ define void @widen_ptr_phi_unrolled(ptr noalias nocapture %a, ptr noalias nocapt
; 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]], splat (i32 1)
; CHECK-NEXT: [[TMP14:%.*]] = add nsw <vscale x 4 x i32> [[TMP11]], splat (i32 1)
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP15]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nuw nsw i64 [[TMP4]], 4
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP15]], i64 [[TMP16]]
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP13]], ptr [[TMP15]], align 4
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP14]], ptr [[TMP17]], align 4
; CHECK-NEXT: [[TMP18:%.*]] = add nsw <vscale x 4 x i32> [[TMP10]], splat (i32 1)
; CHECK-NEXT: [[TMP19:%.*]] = add nsw <vscale x 4 x i32> [[TMP12]], splat (i32 1)
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP20]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP28]]
+; CHECK-NEXT: [[DOTIDX:%.*]] = shl nuw nsw i64 [[TMP4]], 4
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP20]], i64 [[DOTIDX]]
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP18]], ptr [[TMP20]], align 4
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP19]], ptr [[TMP22]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
@@ -75,10 +78,12 @@ define void @widen_ptr_phi_unrolled(ptr noalias nocapture %a, ptr noalias nocapt
; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds nuw i8, ptr [[PTR_014]], i64 8
; CHECK-NEXT: [[TMP25:%.*]] = load i32, ptr [[INCDEC_PTR]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP24]], 1
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I_013]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl nsw i64 [[I_013]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP29]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP25]], 1
-; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[I_013]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i64 [[I_013]], 2
+; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP27]]
; CHECK-NEXT: store i32 [[ADD2]], ptr [[ARRAYIDX3]], align 4
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_013]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[N]]
@@ -131,7 +136,6 @@ define void @widen_2ptrs_phi_unrolled(ptr noalias nocapture %dst, ptr noalias no
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
; CHECK-NEXT: [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i64 [[TMP5]], 2
; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[TMP5]], 3
; CHECK-NEXT: [[DOTNOT:%.*]] = sub nsw i64 0, [[TMP6]]
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[N]], [[DOTNOT]]
@@ -146,12 +150,14 @@ define void @widen_2ptrs_phi_unrolled(ptr noalias nocapture %dst, ptr noalias no
; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[OFFSET_IDX4:%.*]] = shl i64 [[INDEX]], 2
; CHECK-NEXT: [[NEXT_GEP5:%.*]] = getelementptr i8, ptr [[DST]], i64 [[OFFSET_IDX4]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP5]], 4
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[NEXT_GEP]], i64 [[TMP11]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[NEXT_GEP]], align 4
; CHECK-NEXT: [[WIDE_LOAD6:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = shl nsw <vscale x 4 x i32> [[WIDE_LOAD]], splat (i32 1)
; CHECK-NEXT: [[TMP10:%.*]] = shl nsw <vscale x 4 x i32> [[WIDE_LOAD6]], splat (i32 1)
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[NEXT_GEP5]], i64 [[TMP7]]
+; CHECK-NEXT: [[DOTIDX:%.*]] = shl nuw nsw i64 [[TMP5]], 4
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[NEXT_GEP5]], i64 [[DOTIDX]]
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP9]], ptr [[NEXT_GEP5]], align 4
; CHECK-NEXT: store <vscale x 4 x i32> [[TMP10]], ptr [[TMP12]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP6]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-epilogue.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-epilogue.ll
index 209fa60b260aa..ec89a405dda83 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-epilogue.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-epilogue.ll
@@ -25,10 +25,12 @@ define void @simple_histogram(ptr noalias %buckets, ptr readonly %indices, i64 %
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH1]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD1]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP14]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP14]], splat (i64 2)
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP15]], i32 1, <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP7]]
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -38,7 +40,7 @@ define void @simple_histogram(ptr noalias %buckets, ptr readonly %indices, i64 %
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
; CHECK: vec.epilog.iter.check:
; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_MOD_VF]], [[TMP1]]
-; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[SCALAR_PH]], label [[VEC_EPILOG_PH]]
+; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[SCALAR_PH]], label [[VEC_EPILOG_PH]], !prof [[PROF3:![0-9]+]]
; CHECK: vec.epilog.ph:
; CHECK-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_PH]] ]
; CHECK-NEXT: [[TMP24:%.*]] = call i64 @llvm.vscale.i64()
@@ -48,14 +50,16 @@ define void @simple_histogram(ptr noalias %buckets, ptr readonly %indices, i64 %
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: vec.epilog.vector.body:
; CHECK-NEXT: [[INDEX4:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT6:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX4]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[INDEX4]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP16]]
; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <vscale x 2 x i32>, ptr [[TMP18]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = zext <vscale x 2 x i32> [[WIDE_LOAD5]] to <vscale x 2 x i64>
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 2 x i64> [[TMP19]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nuw nsw <vscale x 2 x i64> [[TMP19]], splat (i64 2)
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 2 x i64> [[TMP17]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv2p0.i32(<vscale x 2 x ptr> [[TMP20]], i32 1, <vscale x 2 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT6]] = add nuw i64 [[INDEX4]], [[TMP25]]
; CHECK-NEXT: [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT6]], [[N_VEC3]]
-; CHECK-NEXT: br i1 [[TMP21]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP21]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: vec.epilog.middle.block:
; CHECK-NEXT: [[CMP_N7:%.*]] = icmp eq i64 [[N_MOD_VF2]], 0
; CHECK-NEXT: br i1 [[CMP_N7]], label [[FOR_EXIT]], label [[SCALAR_PH]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-no-scalar-interleave.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-no-scalar-interleave.ll
index 013c218da2e16..b541898df3166 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-no-scalar-interleave.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-no-scalar-interleave.ll
@@ -14,10 +14,12 @@ define void @simple_histogram_forced_scalar_interleave(ptr noalias %buckets, ptr
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[INDICES]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[IDXPROM1:%.*]] = zext i32 [[TMP0]] to i64
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[BUCKETS]], i64 [[IDXPROM1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[IDXPROM1]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1
; CHECK-NEXT: store i32 [[INC]], ptr [[ARRAYIDX2]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-too-many-deps.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-too-many-deps.ll
index baf050c7facee..9731f2a59c06f 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-too-many-deps.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt-too-many-deps.ll
@@ -20,17 +20,21 @@ define void @many_deps(ptr noalias %buckets, ptr %array, ptr %indices, ptr %othe
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[GEP_INDICES:%.*]] = getelementptr inbounds nuw i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[GEP_INDICES:%.*]] = getelementptr inbounds nuw i8, ptr [[INDICES]], i64 [[TMP0]]
; CHECK-NEXT: [[L_IDX:%.*]] = load i32, ptr [[GEP_INDICES]], align 4
; CHECK-NEXT: [[IDXPROM1:%.*]] = zext i32 [[L_IDX]] to i64
-; CHECK-NEXT: [[GEP_BUCKET:%.*]] = getelementptr inbounds nuw i32, ptr [[BUCKETS]], i64 [[IDXPROM1]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDXPROM1]], 2
+; CHECK-NEXT: [[GEP_BUCKET:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP1]]
; CHECK-NEXT: [[L_BUCKET:%.*]] = load i32, ptr [[GEP_BUCKET]], align 4
; CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[L_BUCKET]], 1
; CHECK-NEXT: store i32 [[INC]], ptr [[GEP_BUCKET]], align 4
-; CHECK-NEXT: [[IDX_ADDR:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[IV]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[IDX_ADDR:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAY]], i64 [[TMP2]]
; CHECK-NEXT: [[IV_TRUNC:%.*]] = trunc i64 [[IV]] to i32
; CHECK-NEXT: store i32 [[IV_TRUNC]], ptr [[IDX_ADDR]], align 4
-; CHECK-NEXT: [[GEP_OTHER:%.*]] = getelementptr inbounds nuw i32, ptr [[OTHER]], i64 [[IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[GEP_OTHER:%.*]] = getelementptr inbounds nuw i8, ptr [[OTHER]], i64 [[TMP3]]
; CHECK-NEXT: [[L_OTHER:%.*]] = load i32, ptr [[GEP_OTHER]], align 4
; CHECK-NEXT: [[ADD_OTHER:%.*]] = add i32 [[L_OTHER]], [[IV_TRUNC]]
; CHECK-NEXT: store i32 [[ADD_OTHER]], ptr [[GEP_OTHER]], align 4
@@ -78,17 +82,21 @@ define void @many_deps(ptr noalias %buckets, ptr %array, ptr %indices, ptr %othe
; NORMAL_DEP_LIMIT: vector.body:
; NORMAL_DEP_LIMIT-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
; NORMAL_DEP_LIMIT-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i32> [ [[TMP7]], [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[FOR_BODY]] ]
-; NORMAL_DEP_LIMIT-NEXT: [[GEP_INDICES:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[IV]]
-; NORMAL_DEP_LIMIT-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[GEP_INDICES]], align 4, !alias.scope [[META0:![0-9]+]]
+; NORMAL_DEP_LIMIT-NEXT: [[TMP17:%.*]] = shl nsw i64 [[IV]], 2
+; NORMAL_DEP_LIMIT-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP17]]
+; NORMAL_DEP_LIMIT-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP18]], align 4, !alias.scope [[META0:![0-9]+]]
; NORMAL_DEP_LIMIT-NEXT: [[TMP11:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
-; NORMAL_DEP_LIMIT-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP11]]
+; NORMAL_DEP_LIMIT-NEXT: [[TMP10:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP11]], splat (i64 2)
+; NORMAL_DEP_LIMIT-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP10]]
; NORMAL_DEP_LIMIT-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP12]], i32 1, <vscale x 4 x i1> splat (i1 true))
-; NORMAL_DEP_LIMIT-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[ARRAY]], i64 [[IV]]
+; NORMAL_DEP_LIMIT-NEXT: [[TMP19:%.*]] = shl nsw i64 [[IV]], 2
+; NORMAL_DEP_LIMIT-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[ARRAY]], i64 [[TMP19]]
; NORMAL_DEP_LIMIT-NEXT: store <vscale x 4 x i32> [[VEC_IND]], ptr [[TMP13]], align 4, !alias.scope [[META3:![0-9]+]], !noalias [[META5:![0-9]+]]
-; NORMAL_DEP_LIMIT-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[OTHER]], i64 [[IV]]
-; NORMAL_DEP_LIMIT-NEXT: [[WIDE_LOAD10:%.*]] = load <vscale x 4 x i32>, ptr [[TMP14]], align 4, !alias.scope [[META7:![0-9]+]], !noalias [[META0]]
+; NORMAL_DEP_LIMIT-NEXT: [[TMP14:%.*]] = shl nsw i64 [[IV]], 2
+; NORMAL_DEP_LIMIT-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[OTHER]], i64 [[TMP14]]
+; NORMAL_DEP_LIMIT-NEXT: [[WIDE_LOAD10:%.*]] = load <vscale x 4 x i32>, ptr [[TMP20]], align 4, !alias.scope [[META7:![0-9]+]], !noalias [[META0]]
; NORMAL_DEP_LIMIT-NEXT: [[TMP15:%.*]] = add <vscale x 4 x i32> [[WIDE_LOAD10]], [[VEC_IND]]
-; NORMAL_DEP_LIMIT-NEXT: store <vscale x 4 x i32> [[TMP15]], ptr [[TMP14]], align 4, !alias.scope [[META7]], !noalias [[META0]]
+; NORMAL_DEP_LIMIT-NEXT: store <vscale x 4 x i32> [[TMP15]], ptr [[TMP20]], align 4, !alias.scope [[META7]], !noalias [[META0]]
; NORMAL_DEP_LIMIT-NEXT: [[IV_NEXT]] = add nuw i64 [[IV]], [[TMP8]]
; NORMAL_DEP_LIMIT-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i32> [[VEC_IND]], [[DOTSPLAT]]
; NORMAL_DEP_LIMIT-NEXT: [[TMP16:%.*]] = icmp eq i64 [[IV_NEXT]], [[N_VEC]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll
index 711c4432659a7..350a9010585fa 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve2-histcnt.ll
@@ -43,10 +43,12 @@ define void @simple_histogram(ptr noalias %buckets, ptr readonly %indices, i64 %
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP9]], splat (i64 2)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP7]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP10]], i32 1, <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -92,10 +94,12 @@ define void @simple_histogram_inc_param(ptr noalias %buckets, ptr readonly %indi
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP9]], splat (i64 2)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP7]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP10]], i32 [[INCVAL]], <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -141,10 +145,12 @@ define void @simple_histogram_sub(ptr noalias %buckets, ptr readonly %indices, i
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = sext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw <vscale x 4 x i64> [[TMP9]], splat (i64 2)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP7]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP10]], i32 -1, <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -190,11 +196,14 @@ define void @conditional_histogram(ptr noalias %buckets, ptr readonly %indices,
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP11]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[CONDS]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP9]], splat (i64 2)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP7]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[CONDS]], i64 [[TMP15]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 4 x i32>, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = icmp sgt <vscale x 4 x i32> [[WIDE_LOAD1]], splat (i32 5100)
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP10]], i32 1, <vscale x 4 x i1> [[TMP13]])
@@ -257,7 +266,8 @@ define void @histogram_8bit(ptr noalias %buckets, ptr readonly %indices, i64 %N)
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 16 x i32>, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TMP8:%.*]] = zext <vscale x 16 x i32> [[WIDE_LOAD]] to <vscale x 16 x i64>
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 16 x i64> [[TMP8]]
@@ -280,7 +290,8 @@ define void @histogram_8bit(ptr noalias %buckets, ptr readonly %indices, i64 %N)
; CHECK-NEXT: br label [[VEC_EPILOG_VECTOR_BODY:%.*]]
; CHECK: vec.epilog.vector.body:
; CHECK-NEXT: [[INDEX4:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT6:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX4]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX4]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP18]]
; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <vscale x 8 x i32>, ptr [[TMP13]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = zext <vscale x 8 x i32> [[WIDE_LOAD5]] to <vscale x 8 x i64>
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 8 x i64> [[TMP14]]
@@ -296,7 +307,8 @@ define void @histogram_8bit(ptr noalias %buckets, ptr readonly %indices, i64 %N)
; CHECK-NEXT: br label [[FOR_BODY1:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT1:%.*]], [[FOR_BODY1]] ]
-; CHECK-NEXT: [[GEP_INDICES:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[IV1]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[IV1]], 2
+; CHECK-NEXT: [[GEP_INDICES:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP19]]
; CHECK-NEXT: [[L_IDX:%.*]] = load i32, ptr [[GEP_INDICES]], align 4
; CHECK-NEXT: [[IDXPROM1:%.*]] = zext i32 [[L_IDX]] to i64
; CHECK-NEXT: [[GEP_BUCKET:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[IDXPROM1]]
@@ -337,10 +349,12 @@ define void @histogram_float(ptr noalias %buckets, ptr readonly %indices, i64 %N
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[INDICES]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[IDXPROM1:%.*]] = zext i32 [[TMP0]] to i64
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw float, ptr [[BUCKETS]], i64 [[IDXPROM1]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[IDXPROM1]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INC:%.*]] = fadd fast float [[TMP1]], 1.000000e+00
; CHECK-NEXT: store float [[INC]], ptr [[ARRAYIDX2]], align 4
@@ -378,12 +392,15 @@ define void @histogram_varying_increment(ptr noalias %buckets, ptr readonly %ind
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[INDICES]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[IDXPROM1:%.*]] = zext i32 [[TMP0]] to i64
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[BUCKETS]], i64 [[IDXPROM1]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[IDXPROM1]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP4]]
; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
-; CHECK-NEXT: [[INCIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[INCVALS]], i64 [[IV]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[INCIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[INCVALS]], i64 [[TMP2]]
; CHECK-NEXT: [[INCVAL:%.*]] = load i32, ptr [[INCIDX]], align 4
; CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], [[INCVAL]]
; CHECK-NEXT: store i32 [[INC]], ptr [[ARRAYIDX2]], align 4
@@ -432,15 +449,18 @@ define void @simple_histogram_user_interleave(ptr noalias %buckets, ptr readonly
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP6]]
; CHECK-NEXT: [[DOTIDX:%.*]] = shl nuw nsw i64 [[TMP4]], 4
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP8]], i64 [[DOTIDX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 4 x i32>, ptr [[TMP17]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
; CHECK-NEXT: [[TMP19:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD1]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP19]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP9]], splat (i64 2)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP12]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP19]], splat (i64 2)
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP13]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP10]], i32 1, <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP21]], i32 1, <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
@@ -491,10 +511,12 @@ define void @histogram_array_3op_gep(i64 noundef %N) #0 {
; 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 @idx_array, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr @idx_array, i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <vscale x 4 x i32>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = sext <vscale x 4 x i32> [[WIDE_LOAD1]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr @data_array, <vscale x 4 x i64> [[TMP14]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw <vscale x 4 x i64> [[TMP14]], splat (i64 2)
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr @data_array, <vscale x 4 x i64> [[TMP7]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP11]], i32 1, <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP4]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -544,11 +566,13 @@ define void @histogram_array_4op_gep_nonzero_const_idx(i64 noundef %N, ptr reado
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP6:%.*]] = sext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds nuw i8, ptr [[DATA_STRUCT]], i64 8388608
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[DOTSPLIT]], <vscale x 4 x i64> [[TMP6]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw <vscale x 4 x i64> [[TMP6]], splat (i64 2)
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[DOTSPLIT]], <vscale x 4 x i64> [[TMP10]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP7]], i32 1, <vscale x 4 x i1> splat (i1 true))
; CHECK-NEXT: [[IV_NEXT]] = add nuw i64 [[IV]], [[TMP4]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[IV_NEXT]], [[N_VEC]]
@@ -592,10 +616,12 @@ define void @simple_histogram_tailfold(ptr noalias %buckets, ptr readonly %indic
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP2]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = call <vscale x 4 x i32> @llvm.masked.load.nxv4i32.p0(ptr align 4 [[TMP8]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i32> poison)
; CHECK-NEXT: [[TMP9:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP9]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP9]], splat (i64 2)
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP6]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP10]], i32 1, <vscale x 4 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP5]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX_NEXT]], i64 [[N]])
@@ -658,12 +684,15 @@ define void @simple_histogram_rtdepcheck(ptr noalias %buckets, ptr %array, ptr %
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i32> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 4 x i32>, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = zext <vscale x 4 x i32> [[WIDE_LOAD]] to <vscale x 4 x i64>
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP13]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nuw nsw <vscale x 4 x i64> [[TMP13]], splat (i64 2)
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 4 x i64> [[TMP17]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv4p0.i32(<vscale x 4 x ptr> [[TMP14]], i32 1, <vscale x 4 x i1> splat (i1 true))
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[ARRAY]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[ARRAY]], i64 [[TMP18]]
; CHECK-NEXT: store <vscale x 4 x i32> [[VEC_IND]], ptr [[TMP15]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP8]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i32> [[VEC_IND]], [[DOTSPLAT]]
@@ -706,10 +735,12 @@ define void @simple_histogram_unsafe_alias(ptr %buckets, ptr %indices, i64 %N) #
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[INDICES]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[IDXPROM1:%.*]] = zext i32 [[TMP12]] to i64
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[BUCKETS]], i64 [[IDXPROM1]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDXPROM1]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP13]], 1
; CHECK-NEXT: store i32 [[INC]], ptr [[ARRAYIDX2]], align 4
@@ -755,9 +786,11 @@ define void @simple_histogram_64b(ptr noalias %buckets, ptr readonly %indices, i
; 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 i64, ptr [[INDICES]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP8]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP5]], align 4
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[BUCKETS]], <vscale x 2 x i64> [[WIDE_LOAD]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw <vscale x 2 x i64> [[WIDE_LOAD]], splat (i64 3)
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[BUCKETS]], <vscale x 2 x i64> [[TMP9]]
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.nxv2p0.i64(<vscale x 2 x ptr> [[TMP6]], i64 1, <vscale x 2 x i1> splat (i1 true))
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP4]]
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -795,10 +828,12 @@ define void @histogram_generates_vectors_crash(ptr %data_array, ptr noalias %ind
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[GEP_INDICES:%.*]] = getelementptr [1048576 x i32], ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[IV]], 22
+; CHECK-NEXT: [[GEP_INDICES:%.*]] = getelementptr i8, ptr [[INDICES]], i64 [[TMP0]]
; CHECK-NEXT: [[L_IDX:%.*]] = load i32, ptr [[GEP_INDICES]], align 4
; CHECK-NEXT: [[IDXPROM5:%.*]] = sext i32 [[L_IDX]] to i64
-; CHECK-NEXT: [[GEP_BUCKET:%.*]] = getelementptr [1048576 x i32], ptr [[DATA_ARRAY]], i64 [[IDXPROM5]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IDXPROM5]], 22
+; CHECK-NEXT: [[GEP_BUCKET:%.*]] = getelementptr i8, ptr [[DATA_ARRAY]], i64 [[TMP1]]
; CHECK-NEXT: [[L_BUCKET:%.*]] = load i32, ptr [[GEP_BUCKET]], align 4
; CHECK-NEXT: [[INC:%.*]] = add i32 [[L_BUCKET]], 1
; CHECK-NEXT: store i32 [[INC]], ptr [[GEP_BUCKET]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll b/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll
index 7f8e901f1b361..212c1154d1c6a 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/uniform-args-call-variants.ll
@@ -17,15 +17,17 @@ define void @test_uniform(ptr noalias %dst, ptr readonly %src, i64 %uniform , i6
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 2 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP2]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
-; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @foo_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i64 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
-; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP3]], ptr align 8 [[TMP4]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP2]]
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP3]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
+; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x double> @foo_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i64 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP5]]
+; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP4]], ptr align 8 [[TMP6]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX_NEXT]], i64 [[N]])
-; CHECK-NEXT: [[TMP5:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
-; CHECK-NEXT: br i1 [[TMP5]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK-NEXT: [[TMP7:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT: br i1 [[TMP7]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
;
@@ -42,22 +44,26 @@ define void @test_uniform(ptr noalias %dst, ptr readonly %src, i64 %uniform , i6
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 2 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK2:%.*]] = phi <vscale x 2 x i1> [ [[ACTIVE_LANE_MASK_ENTRY1]], [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT4:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP4:%.*]] = getelementptr double, ptr [[TMP3]], i64 [[TMP1]]
-; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP3]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
-; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD3:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP4]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]], <vscale x 2 x double> poison)
-; INTERLEAVE-NEXT: [[TMP5:%.*]] = call <vscale x 2 x double> @foo_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i64 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; INTERLEAVE-NEXT: [[TMP6:%.*]] = call <vscale x 2 x double> @foo_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD3]], i64 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
-; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr inbounds double, ptr [[TMP7]], i64 [[TMP1]]
-; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP5]], ptr align 8 [[TMP7]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP6]], ptr align 8 [[TMP8]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP3]]
+; INTERLEAVE-NEXT: [[TMP5:%.*]] = shl i64 [[TMP0]], 4
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP4]], i64 [[TMP5]]
+; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP4]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
+; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD3:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP6]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]], <vscale x 2 x double> poison)
+; INTERLEAVE-NEXT: [[TMP7:%.*]] = call <vscale x 2 x double> @foo_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i64 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = call <vscale x 2 x double> @foo_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD3]], i64 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
+; INTERLEAVE-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP9]]
+; INTERLEAVE-NEXT: [[TMP11:%.*]] = shl i64 [[TMP0]], 4
+; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i64 [[TMP11]]
+; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP7]], ptr align 8 [[TMP10]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP8]], ptr align 8 [[TMP12]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP2]]
-; INTERLEAVE-NEXT: [[TMP9:%.*]] = add i64 [[INDEX_NEXT]], [[TMP1]]
+; INTERLEAVE-NEXT: [[TMP13:%.*]] = add i64 [[INDEX_NEXT]], [[TMP1]]
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX_NEXT]], i64 [[N]])
-; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT4]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP9]], i64 [[N]])
-; INTERLEAVE-NEXT: [[TMP10:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
-; INTERLEAVE-NEXT: br i1 [[TMP10]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
+; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT4]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP13]], i64 [[N]])
+; INTERLEAVE-NEXT: [[TMP14:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; INTERLEAVE-NEXT: br i1 [[TMP14]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP0:![0-9]+]]
; INTERLEAVE: for.cond.cleanup:
; INTERLEAVE-NEXT: ret void
;
@@ -90,15 +96,17 @@ define void @test_uniform_smaller_scalar(ptr noalias %dst, ptr readonly %src, i3
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 2 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP2]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
-; CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 2 x double> @bar_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i32 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
-; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP3]], ptr align 8 [[TMP4]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP2]]
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP3]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
+; CHECK-NEXT: [[TMP4:%.*]] = call <vscale x 2 x double> @bar_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i32 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP5]]
+; CHECK-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP4]], ptr align 8 [[TMP6]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
; CHECK-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP1]]
; CHECK-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX_NEXT]], i64 [[N]])
-; CHECK-NEXT: [[TMP5:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
-; CHECK-NEXT: br i1 [[TMP5]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK-NEXT: [[TMP7:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; CHECK-NEXT: br i1 [[TMP7]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
;
@@ -115,22 +123,26 @@ define void @test_uniform_smaller_scalar(ptr noalias %dst, ptr readonly %src, i3
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 2 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK2:%.*]] = phi <vscale x 2 x i1> [ [[ACTIVE_LANE_MASK_ENTRY1]], [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT4:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP4:%.*]] = getelementptr double, ptr [[TMP3]], i64 [[TMP1]]
-; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP3]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
-; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD3:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP4]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]], <vscale x 2 x double> poison)
-; INTERLEAVE-NEXT: [[TMP5:%.*]] = call <vscale x 2 x double> @bar_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i32 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; INTERLEAVE-NEXT: [[TMP6:%.*]] = call <vscale x 2 x double> @bar_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD3]], i32 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
-; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr inbounds double, ptr [[TMP7]], i64 [[TMP1]]
-; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP5]], ptr align 8 [[TMP7]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
-; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP6]], ptr align 8 [[TMP8]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP3]]
+; INTERLEAVE-NEXT: [[TMP5:%.*]] = shl i64 [[TMP0]], 4
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP4]], i64 [[TMP5]]
+; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP4]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x double> poison)
+; INTERLEAVE-NEXT: [[WIDE_MASKED_LOAD3:%.*]] = call <vscale x 2 x double> @llvm.masked.load.nxv2f64.p0(ptr align 8 [[TMP6]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]], <vscale x 2 x double> poison)
+; INTERLEAVE-NEXT: [[TMP7:%.*]] = call <vscale x 2 x double> @bar_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD]], i32 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = call <vscale x 2 x double> @bar_uniform(<vscale x 2 x double> [[WIDE_MASKED_LOAD3]], i32 [[UNIFORM]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
+; INTERLEAVE-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP9]]
+; INTERLEAVE-NEXT: [[TMP11:%.*]] = shl i64 [[TMP0]], 4
+; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i64 [[TMP11]]
+; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP7]], ptr align 8 [[TMP10]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
+; INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2f64.p0(<vscale x 2 x double> [[TMP8]], ptr align 8 [[TMP12]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP2]]
-; INTERLEAVE-NEXT: [[TMP9:%.*]] = add i64 [[INDEX_NEXT]], [[TMP1]]
+; INTERLEAVE-NEXT: [[TMP13:%.*]] = add i64 [[INDEX_NEXT]], [[TMP1]]
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX_NEXT]], i64 [[N]])
-; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT4]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP9]], i64 [[N]])
-; INTERLEAVE-NEXT: [[TMP10:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
-; INTERLEAVE-NEXT: br i1 [[TMP10]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
+; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT4]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP13]], i64 [[N]])
+; INTERLEAVE-NEXT: [[TMP14:%.*]] = extractelement <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], i64 0
+; INTERLEAVE-NEXT: br i1 [[TMP14]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP3:![0-9]+]]
; INTERLEAVE: for.cond.cleanup:
; INTERLEAVE-NEXT: ret void
;
@@ -160,10 +172,12 @@ define void @test_uniform_not_invariant(ptr noalias %dst, ptr readonly %src, i64
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[GEPSRC:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[GEPSRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP0]]
; CHECK-NEXT: [[DATA:%.*]] = load double, ptr [[GEPSRC]], align 8
; CHECK-NEXT: [[CALL:%.*]] = call double @foo(double [[DATA]], i64 [[INDVARS_IV]]) #[[ATTR4:[0-9]+]]
-; CHECK-NEXT: [[GEPDST:%.*]] = getelementptr inbounds nuw double, ptr [[DST]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[GEPDST:%.*]] = getelementptr inbounds nuw i8, ptr [[DST]], i64 [[TMP1]]
; CHECK-NEXT: store double [[CALL]], ptr [[GEPDST]], align 8
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
@@ -183,27 +197,31 @@ define void @test_uniform_not_invariant(ptr noalias %dst, ptr readonly %src, i64
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK2:%.*]] = phi i1 [ [[ACTIVE_LANE_MASK_ENTRY1]], [[ENTRY]] ], [ [[ACTIVE_LANE_MASK_NEXT5:%.*]], [[PRED_STORE_CONTINUE4]] ]
; INTERLEAVE-NEXT: br i1 [[ACTIVE_LANE_MASK]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; INTERLEAVE: pred.store.if:
-; INTERLEAVE-NEXT: [[TMP0:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: [[TMP1:%.*]] = load double, ptr [[TMP0]], align 8
-; INTERLEAVE-NEXT: [[TMP2:%.*]] = call double @foo(double [[TMP1]], i64 [[INDEX]]) #[[ATTR4:[0-9]+]]
-; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
-; INTERLEAVE-NEXT: store double [[TMP2]], ptr [[TMP3]], align 8
+; INTERLEAVE-NEXT: [[TMP0:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP0]]
+; INTERLEAVE-NEXT: [[TMP2:%.*]] = load double, ptr [[TMP1]], align 8
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = call double @foo(double [[TMP2]], i64 [[INDEX]]) #[[ATTR4:[0-9]+]]
+; INTERLEAVE-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP4]]
+; INTERLEAVE-NEXT: store double [[TMP3]], ptr [[TMP5]], align 8
; INTERLEAVE-NEXT: br label [[PRED_STORE_CONTINUE]]
; INTERLEAVE: pred.store.continue:
; INTERLEAVE-NEXT: br i1 [[ACTIVE_LANE_MASK2]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4]]
; INTERLEAVE: pred.store.if3:
-; INTERLEAVE-NEXT: [[TMP4:%.*]] = or disjoint i64 [[INDEX]], 1
-; INTERLEAVE-NEXT: [[TMP5:%.*]] = getelementptr double, ptr [[SRC]], i64 [[TMP4]]
-; INTERLEAVE-NEXT: [[TMP6:%.*]] = load double, ptr [[TMP5]], align 8
-; INTERLEAVE-NEXT: [[TMP7:%.*]] = call double @foo(double [[TMP6]], i64 [[TMP4]]) #[[ATTR4]]
-; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[TMP4]]
-; INTERLEAVE-NEXT: store double [[TMP7]], ptr [[TMP8]], align 8
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = or disjoint i64 [[INDEX]], 1
+; INTERLEAVE-NEXT: [[TMP7:%.*]] = shl i64 [[TMP6]], 3
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP7]]
+; INTERLEAVE-NEXT: [[TMP9:%.*]] = load double, ptr [[TMP8]], align 8
+; INTERLEAVE-NEXT: [[TMP10:%.*]] = call double @foo(double [[TMP9]], i64 [[TMP6]]) #[[ATTR4]]
+; INTERLEAVE-NEXT: [[TMP11:%.*]] = shl nsw i64 [[TMP6]], 3
+; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP11]]
+; INTERLEAVE-NEXT: store double [[TMP10]], ptr [[TMP12]], align 8
; INTERLEAVE-NEXT: br label [[PRED_STORE_CONTINUE4]]
; INTERLEAVE: pred.store.continue4:
; INTERLEAVE-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], 2
-; INTERLEAVE-NEXT: [[TMP9:%.*]] = add i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP13:%.*]] = add i64 [[INDEX]], 3
; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT:%.*]] = icmp ult i64 [[INDEX_NEXT]], [[N]]
-; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT5]] = icmp ult i64 [[TMP9]], [[N]]
+; INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT5]] = icmp ult i64 [[TMP13]], [[N]]
; INTERLEAVE-NEXT: br i1 [[ACTIVE_LANE_MASK_NEXT]], label [[VECTOR_BODY]], label [[FOR_COND_CLEANUP:%.*]], !llvm.loop [[LOOP4:![0-9]+]]
; INTERLEAVE: for.cond.cleanup:
; INTERLEAVE-NEXT: ret void
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll b/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll
index f0aa75aab1f91..428df30fe1e55 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/vector-reverse-mask4.ll
@@ -33,7 +33,8 @@ define void @vector_reverse_mask_v4i1(ptr noalias %a, ptr noalias %cond, i64 %N)
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = xor i64 [[INDEX]], -1
; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[N]], [[TMP0]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[COND:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP1]], 3
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[COND:%.*]], i64 [[TMP15]]
; 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
@@ -42,7 +43,8 @@ define void @vector_reverse_mask_v4i1(ptr noalias %a, ptr noalias %cond, i64 %N)
; CHECK-NEXT: [[REVERSE2:%.*]] = shufflevector <4 x double> [[WIDE_LOAD1]], <4 x double> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
; 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: [[TMP17:%.*]] = shl i64 [[TMP1]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP17]]
; 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>
@@ -69,12 +71,14 @@ define void @vector_reverse_mask_v4i1(ptr noalias %a, ptr noalias %cond, i64 %N)
; CHECK: for.body:
; CHECK-NEXT: [[I_08_IN:%.*]] = phi i64 [ [[I_08:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[I_08]] = add nsw i64 [[I_08_IN]], -1
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds double, ptr [[COND]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[I_08]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP16]]
; CHECK-NEXT: [[TMP13:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[TOBOOL:%.*]] = fcmp une double [[TMP13]], 0.000000e+00
; CHECK-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds double, ptr [[A]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[I_08]], 3
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; CHECK-NEXT: [[TMP14:%.*]] = load double, ptr [[ARRAYIDX1]], align 8
; CHECK-NEXT: [[ADD:%.*]] = fadd double [[TMP14]], 1.000000e+00
; CHECK-NEXT: store double [[ADD]], ptr [[ARRAYIDX1]], align 8
diff --git a/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll b/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll
index bc9cf4fe93622..16b7780e92eaf 100644
--- a/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll
+++ b/llvm/test/Transforms/LoopVectorize/AMDGPU/packed-math.ll
@@ -13,7 +13,8 @@ define half @vectorize_v2f16_loop(ptr addrspace(1) noalias %s) {
; GFX9-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[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: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 1
+; GFX9-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[S:%.*]], i64 [[TMP5]]
; GFX9-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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) [[TMP1]], align 2
@@ -38,7 +39,8 @@ define half @vectorize_v2f16_loop(ptr addrspace(1) noalias %s) {
; VI-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[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: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 1
+; VI-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr addrspace(1) [[S:%.*]], i64 [[TMP5]]
; VI-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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) [[TMP1]], align 2
@@ -60,7 +62,8 @@ define half @vectorize_v2f16_loop(ptr addrspace(1) noalias %s) {
; CI: for.body:
; CI-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; CI-NEXT: [[Q_04:%.*]] = phi half [ 0xH0000, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CI-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw half, ptr addrspace(1) [[S:%.*]], i64 [[INDVARS_IV]]
+; CI-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
+; CI-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(1) [[S:%.*]], i64 [[TMP1]]
; CI-NEXT: [[TMP0:%.*]] = load half, ptr addrspace(1) [[ARRAYIDX]], align 2
; CI-NEXT: [[ADD]] = fadd fast half [[Q_04]], [[TMP0]]
; CI-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-reduction-predselect.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-reduction-predselect.ll
index 706012c155fdf..03396ba2bb0c0 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-reduction-predselect.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-reduction-predselect.ll
@@ -14,7 +14,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP4]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[WIDE_MASKED_LOAD]])
; CHECK-NEXT: [[TMP2]] = add i32 [[VEC_PHI]], [[TMP1]]
@@ -55,9 +56,11 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP10]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> zeroinitializer)
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP11]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP2:%.*]] = select <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> [[VEC_IND]], <4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP3:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP2]])
@@ -107,9 +110,11 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1), [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
; CHECK-NEXT: [[TMP2:%.*]] = mul <4 x i32> [[VEC_PHI]], [[WIDE_MASKED_LOAD]]
; CHECK-NEXT: [[TMP3:%.*]] = mul <4 x i32> [[TMP2]], [[WIDE_MASKED_LOAD1]]
@@ -154,9 +159,11 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 -1), [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
; CHECK-NEXT: [[TMP2:%.*]] = and <4 x i32> [[WIDE_MASKED_LOAD]], [[WIDE_MASKED_LOAD1]]
; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> [[TMP2]], <4 x i32> splat (i32 -1)
@@ -201,9 +208,11 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_MASKED_LOAD1]], [[WIDE_MASKED_LOAD]]
; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> [[TMP2]], <4 x i32> zeroinitializer
@@ -248,9 +257,11 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_MASKED_LOAD1]], [[WIDE_MASKED_LOAD]]
; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> [[TMP2]], <4 x i32> zeroinitializer
@@ -295,9 +306,11 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x float> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x float> poison)
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x float> poison)
; CHECK-NEXT: [[TMP2:%.*]] = fadd fast <4 x float> [[VEC_PHI]], [[WIDE_MASKED_LOAD]]
; CHECK-NEXT: [[TMP3:%.*]] = fadd fast <4 x float> [[TMP2]], [[WIDE_MASKED_LOAD1]]
@@ -342,9 +355,11 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x float> [ <float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 257)
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x float> poison)
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x float> poison)
; CHECK-NEXT: [[TMP2:%.*]] = fmul fast <4 x float> [[VEC_PHI]], [[WIDE_MASKED_LOAD]]
; CHECK-NEXT: [[TMP3:%.*]] = fmul fast <4 x float> [[TMP2]], [[WIDE_MASKED_LOAD1]]
@@ -388,7 +403,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1000), [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP3]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[WIDE_LOAD]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -402,7 +418,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 256, [[SCALAR_PH]] ]
; CHECK-NEXT: [[RESULT_08:%.*]] = phi i32 [ [[V0:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i32 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP5]]
; CHECK-NEXT: [[L0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[V0]] = call i32 @llvm.smin.i32(i32 [[RESULT_08]], i32 [[L0]])
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i32 [[INDVARS_IV]], 1
@@ -439,7 +456,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1000), [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP3]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[WIDE_LOAD]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -453,7 +471,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 256, [[SCALAR_PH]] ]
; CHECK-NEXT: [[RESULT_08:%.*]] = phi i32 [ [[V0:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i32 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP5]]
; CHECK-NEXT: [[L0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[V0]] = call i32 @llvm.umax.i32(i32 [[RESULT_08]], i32 [[L0]])
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i32 [[INDVARS_IV]], 1
@@ -488,7 +507,8 @@ define float @reduction_fmax(ptr nocapture %A, ptr nocapture %B) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: [[RESULT_08:%.*]] = phi float [ [[V0:%.*]], [[FOR_BODY]] ], [ 1.000000e+03, [[ENTRY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i32 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP0]]
; CHECK-NEXT: [[L0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[C0:%.*]] = fcmp ogt float [[RESULT_08]], [[L0]]
; CHECK-NEXT: [[V0]] = select i1 [[C0]], float [[RESULT_08]], float [[L0]]
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions-interleave.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions-interleave.ll
index 27e87bc84dc86..56761eb4bfd33 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions-interleave.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions-interleave.ll
@@ -17,7 +17,8 @@ define i64 @add_i32_i64(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP11]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i32 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP8]], align 4
@@ -41,7 +42,8 @@ define i64 @add_i32_i64(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK: loop:
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[X]], i32 [[I_08]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i32 [[I_08]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 [[TMP10]]
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP5]] to i64
; CHECK-NEXT: [[ADD]] = add nsw i64 [[R_07]], [[CONV]]
@@ -84,11 +86,13 @@ define i64 @mla_i32_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP11:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP16]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i32 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP12]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[Y:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i32 [[TMP18]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i32 16
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4
@@ -114,9 +118,11 @@ define i64 @mla_i32_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK: loop:
; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[R_09:%.*]] = phi i64 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[X]], i32 [[I_010]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i32 [[I_010]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 [[TMP15]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[Y]], i32 [[I_010]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i32 [[I_010]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[Y]], i32 [[TMP17]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], [[TMP7]]
; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[MUL]] to i64
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll
index d9c441446750f..cf0f7187179d9 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll
@@ -13,7 +13,8 @@ define i64 @add_i64_i64(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i64, ptr [[X:%.*]], i32 [[I_08]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[I_08]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i32 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[ADD]] = add nsw i64 [[TMP0]], [[R_07]]
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1
@@ -58,7 +59,8 @@ define i64 @add_i32_i64(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = sext <4 x i32> [[WIDE_LOAD]] to <4 x i64>
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> [[TMP1]])
@@ -76,7 +78,8 @@ define i64 @add_i32_i64(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[X]], i32 [[I_08]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[I_08]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP6]]
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP5]] to i64
; CHECK-NEXT: [[ADD]] = add nsw i64 [[R_07]], [[CONV]]
@@ -123,7 +126,8 @@ define i64 @add_i16_i64(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP0]], align 2
; CHECK-NEXT: [[TMP1:%.*]] = sext <4 x i16> [[WIDE_LOAD]] to <4 x i64>
; CHECK-NEXT: [[TMP2:%.*]] = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> [[TMP1]])
@@ -141,7 +145,8 @@ define i64 @add_i16_i64(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_08]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i32 [[I_08]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP6]]
; CHECK-NEXT: [[TMP5:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP5]] to i64
; CHECK-NEXT: [[ADD]] = add nsw i64 [[R_07]], [[CONV]]
@@ -251,7 +256,8 @@ define i32 @add_i32_i32(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 [[N]])
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP4]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP0]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> zeroinitializer)
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[WIDE_MASKED_LOAD]])
; CHECK-NEXT: [[TMP2]] = add i32 [[VEC_PHI]], [[TMP1]]
@@ -295,7 +301,8 @@ define i32 @add_i16_i32(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <8 x i1> @llvm.get.active.lane.mask.v8i1.i32(i32 [[INDEX]], i32 [[N]])
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP6]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr align 2 [[TMP0]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> poison)
; CHECK-NEXT: [[TMP1:%.*]] = sext <8 x i16> [[WIDE_MASKED_LOAD]] to <8 x i32>
; CHECK-NEXT: [[TMP2:%.*]] = select <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i32> [[TMP1]], <8 x i32> zeroinitializer
@@ -389,7 +396,8 @@ define signext i16 @add_i16_i16(ptr nocapture readonly %x, i32 %n) #0 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i16 [ 0, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <8 x i1> @llvm.get.active.lane.mask.v8i1.i32(i32 [[INDEX]], i32 [[N]])
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP4]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr align 2 [[TMP0]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> zeroinitializer)
; CHECK-NEXT: [[TMP1:%.*]] = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> [[WIDE_MASKED_LOAD]])
; CHECK-NEXT: [[TMP2]] = add i16 [[VEC_PHI]], [[TMP1]]
@@ -519,9 +527,11 @@ define i64 @mla_i64_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK: for.body:
; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
; CHECK-NEXT: [[R_09:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i64, ptr [[X:%.*]], i32 [[I_010]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i32 [[I_010]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i32 [[TMP3]]
; CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i64, ptr [[Y:%.*]], i32 [[I_010]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[I_010]], 3
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i32 [[TMP2]]
; CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[ARRAYIDX1]], align 8
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP1]], [[TMP0]]
; CHECK-NEXT: [[ADD]] = add nsw i64 [[MUL]], [[R_09]]
@@ -570,9 +580,11 @@ define i64 @mla_i32_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[Y:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i32 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = mul nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP3:%.*]] = sext <4 x i32> [[TMP2]] to <4 x i64>
@@ -591,9 +603,11 @@ define i64 @mla_i32_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK: for.body:
; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[R_09:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[X]], i32 [[I_010]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i32 [[I_010]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP9]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i32, ptr [[Y]], i32 [[I_010]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i32 [[I_010]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i32 [[TMP11]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], [[TMP7]]
; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[MUL]] to i64
@@ -644,9 +658,11 @@ define i64 @mla_i16_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP1]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i32 [[TMP11]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <8 x i16>, ptr [[TMP2]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = sext <8 x i16> [[WIDE_LOAD1]] to <8 x i64>
; CHECK-NEXT: [[TMP3:%.*]] = sext <8 x i16> [[WIDE_LOAD]] to <8 x i64>
@@ -666,10 +682,12 @@ define i64 @mla_i16_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK: for.body:
; CHECK-NEXT: [[I_012:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[R_011:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_012]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nuw nsw i32 [[I_012]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP13]]
; CHECK-NEXT: [[TMP9:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP9]] to i32
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i16, ptr [[Y]], i32 [[I_012]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw i32 [[I_012]], 1
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i32 [[TMP12]]
; CHECK-NEXT: [[TMP10:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2
; CHECK-NEXT: [[CONV2:%.*]] = sext i16 [[TMP10]] to i32
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV2]], [[CONV]]
@@ -800,9 +818,11 @@ define i32 @mla_i32_i32(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 [[INDEX]], i32 [[N]])
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[Y:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP0]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP1]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[Y1:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i32 [[TMP8]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD2:%.*]] = call <4 x i32> @llvm.masked.load.v4i32.p0(ptr align 4 [[TMP7]], <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> poison)
; CHECK-NEXT: [[TMP2:%.*]] = mul nsw <4 x i32> [[WIDE_MASKED_LOAD2]], [[WIDE_MASKED_LOAD1]]
; CHECK-NEXT: [[TMP3:%.*]] = select <4 x i1> [[ACTIVE_LANE_MASK]], <4 x i32> [[TMP2]], <4 x i32> zeroinitializer
@@ -851,9 +871,11 @@ define i32 @mla_i16_i32(ptr nocapture readonly %x, ptr nocapture readonly %y, i3
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <8 x i1> @llvm.get.active.lane.mask.v8i1.i32(i32 [[INDEX]], i32 [[N]])
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP9]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr align 2 [[TMP0]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> poison)
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i32 [[TMP10]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr align 2 [[TMP2]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> poison)
; CHECK-NEXT: [[TMP3:%.*]] = sext <8 x i16> [[WIDE_MASKED_LOAD1]] to <8 x i32>
; CHECK-NEXT: [[TMP1:%.*]] = sext <8 x i16> [[WIDE_MASKED_LOAD]] to <8 x i32>
@@ -961,9 +983,11 @@ define signext i16 @mla_i16_i16(ptr nocapture readonly %x, ptr nocapture readonl
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i16 [ 0, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <8 x i1> @llvm.get.active.lane.mask.v8i1.i32(i32 [[INDEX]], i32 [[N]])
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP0]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr align 2 [[TMP1]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> poison)
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i16, ptr [[Y1:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i32 [[TMP8]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD2:%.*]] = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr align 2 [[TMP7]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> poison)
; CHECK-NEXT: [[TMP2:%.*]] = mul <8 x i16> [[WIDE_MASKED_LOAD2]], [[WIDE_MASKED_LOAD1]]
; CHECK-NEXT: [[TMP3:%.*]] = select <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> [[TMP2]], <8 x i16> zeroinitializer
@@ -1120,7 +1144,8 @@ define i32 @red_mla_ext_s8_s16_s32(ptr noalias nocapture readonly %A, ptr noalia
; CHECK-NEXT: [[ACTIVE_LANE_MASK:%.*]] = call <8 x i1> @llvm.get.active.lane.mask.v8i1.i32(i32 [[INDEX]], i32 [[N]])
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[INDEX]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <8 x i8> @llvm.masked.load.v8i8.p0(ptr align 1 [[TMP0]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i8> poison)
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP9]]
; CHECK-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <8 x i16> @llvm.masked.load.v8i16.p0(ptr align 2 [[TMP2]], <8 x i1> [[ACTIVE_LANE_MASK]], <8 x i16> poison)
; CHECK-NEXT: [[TMP3:%.*]] = sext <8 x i16> [[WIDE_MASKED_LOAD1]] to <8 x i32>
; CHECK-NEXT: [[TMP1:%.*]] = sext <8 x i8> [[WIDE_MASKED_LOAD]] to <8 x i32>
@@ -1181,10 +1206,12 @@ define i64 @red_mla_ext_s16_u16_s64(ptr noalias nocapture readonly %A, ptr noali
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP0]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP2]], align 1
; CHECK-NEXT: [[TMP1:%.*]] = sext <4 x i16> [[WIDE_LOAD]] to <4 x i32>
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i16, ptr [[B1:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i16>, ptr [[TMP11]], align 2
; CHECK-NEXT: [[TMP3:%.*]] = zext <4 x i16> [[WIDE_LOAD1]] to <4 x i32>
; CHECK-NEXT: [[TMP4:%.*]] = mul nsw <4 x i32> [[TMP3]], [[TMP1]]
@@ -1204,10 +1231,12 @@ define i64 @red_mla_ext_s16_u16_s64(ptr noalias nocapture readonly %A, ptr noali
; CHECK: for.body:
; CHECK-NEXT: [[I_011:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[S_010:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[B]], i32 [[I_011]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i32 [[I_011]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP14]]
; CHECK-NEXT: [[TMP9:%.*]] = load i16, ptr [[ARRAYIDX]], align 1
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP9]] to i32
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, ptr [[B1]], i32 [[I_011]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i32 [[I_011]], 1
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP13]]
; CHECK-NEXT: [[TMP10:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2
; CHECK-NEXT: [[CONV2:%.*]] = zext i16 [[TMP10]] to i32
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV2]], [[CONV]]
@@ -1355,10 +1384,12 @@ define i32 @reduction_interleave_group(i32 %n, ptr %arr) #0 {
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[RED_PHI:%.*]] = phi i32 [ [[RED_2:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[ARR]], i32 [[IV]]
+; CHECK-NEXT: [[ADD:%.*]] = shl i32 [[IV]], 2
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[ARR]], i32 [[ADD]]
; CHECK-NEXT: [[GEP_0:%.*]] = getelementptr i8, ptr [[TMP11]], i32 4
; CHECK-NEXT: [[L_0:%.*]] = load i32, ptr [[GEP_0]], align 4
-; CHECK-NEXT: [[GEP_1:%.*]] = getelementptr inbounds i32, ptr [[ARR]], i32 [[IV]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i32 [[IV]], 2
+; CHECK-NEXT: [[GEP_1:%.*]] = getelementptr inbounds i8, ptr [[ARR]], i32 [[TMP12]]
; CHECK-NEXT: [[L_1:%.*]] = load i32, ptr [[GEP_1]], align 4
; CHECK-NEXT: [[RED_1:%.*]] = add i32 [[L_0]], [[RED_PHI]]
; CHECK-NEXT: [[RED_2]] = add i32 [[RED_1]], [[L_1]]
@@ -1455,7 +1486,8 @@ define i64 @mla_xx_sext_zext(ptr nocapture noundef readonly %x, i32 %n) #0 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP2]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[TMP0]], align 2
; CHECK-NEXT: [[TMP1:%.*]] = sext <8 x i16> [[WIDE_LOAD]] to <8 x i64>
; CHECK-NEXT: [[TMP3:%.*]] = mul nsw <8 x i64> [[TMP1]], [[TMP1]]
@@ -1477,7 +1509,8 @@ define i64 @mla_xx_sext_zext(ptr nocapture noundef readonly %x, i32 %n) #0 {
; CHECK: for.body:
; CHECK-NEXT: [[I_011:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[S_010:%.*]] = phi i64 [ [[ADD]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_011]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i32 [[I_011]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP8]]
; CHECK-NEXT: [[TMP7:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP7]] to i32
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV]], [[CONV]]
@@ -1523,7 +1556,8 @@ define i64 @mla_and_add_together_16_64(ptr nocapture noundef readonly %x, i32 no
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i32 [[INDEX]], 1
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP1]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[TMP0]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = sext <8 x i16> [[WIDE_LOAD]] to <8 x i64>
; CHECK-NEXT: [[TMP2:%.*]] = mul nsw <8 x i64> [[TMP4]], [[TMP4]]
@@ -1553,7 +1587,8 @@ define i64 @mla_and_add_together_16_64(ptr nocapture noundef readonly %x, i32 no
; CHECK-NEXT: [[I_019:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[T_018:%.*]] = phi i64 [ [[ADD]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[S_017:%.*]] = phi i32 [ [[ADD6]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX2]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_019]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i32 [[I_019]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP11]]
; CHECK-NEXT: [[TMP9:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP9]] to i32
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV]], [[CONV]]
@@ -1601,20 +1636,24 @@ define i64 @interleave_doublereduct_i16_i64(ptr %x, ptr %y, i32 %n) {
; CHECK: for.body:
; CHECK-NEXT: [[I_025:%.*]] = phi i32 [ [[ADD13:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT: [[T_024:%.*]] = phi i64 [ [[ADD12]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i32 [[I_025]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i32 [[I_025]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i32 [[TMP5]]
; CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP0]] to i32
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i32 [[I_025]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i32 [[I_025]], 1
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i32 [[TMP7]]
; CHECK-NEXT: [[TMP1:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2
; CHECK-NEXT: [[CONV2:%.*]] = sext i16 [[TMP1]] to i32
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV2]], [[CONV]]
; CHECK-NEXT: [[CONV3:%.*]] = sext i32 [[MUL]] to i64
; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[T_024]], [[CONV3]]
; CHECK-NEXT: [[ADD4:%.*]] = or disjoint i32 [[I_025]], 1
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[ADD4]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i32 [[ADD4]], 1
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP4]]
; CHECK-NEXT: [[TMP2:%.*]] = load i16, ptr [[ARRAYIDX5]], align 2
; CHECK-NEXT: [[CONV6:%.*]] = sext i16 [[TMP2]] to i32
-; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds nuw i16, ptr [[Y]], i32 [[ADD4]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i32 [[ADD4]], 1
+; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i32 [[TMP6]]
; CHECK-NEXT: [[TMP3:%.*]] = load i16, ptr [[ARRAYIDX8]], align 2
; CHECK-NEXT: [[CONV9:%.*]] = sext i16 [[TMP3]] to i32
; CHECK-NEXT: [[MUL10:%.*]] = mul nsw i32 [[CONV9]], [[CONV6]]
@@ -1673,7 +1712,8 @@ define i64 @test_std_q31(ptr %x, i32 %n) #0 {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[FOR_BODY1]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[FOR_BODY1]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[FOR_BODY1]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP5]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = ashr <4 x i32> [[WIDE_LOAD]], splat (i32 8)
; CHECK-NEXT: [[TMP2:%.*]] = sext <4 x i32> [[TMP1]] to <4 x i64>
@@ -1702,7 +1742,8 @@ define i64 @test_std_q31(ptr %x, i32 %n) #0 {
; CHECK-NEXT: [[S_014:%.*]] = phi i64 [ [[ADD1]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[I_013:%.*]] = phi i32 [ [[ADD4:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[T_012:%.*]] = phi i64 [ [[ADD5]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX2]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[X]], i32 [[I_013]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i32 [[I_013]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[TMP11]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[SHR:%.*]] = ashr i32 [[TMP0]], 8
; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[SHR]] to i64
@@ -1756,12 +1797,13 @@ define i64 @test_fir_q15(ptr %x, ptr %y, i32 %n) #0 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP16:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP5]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <16 x i16>, ptr [[TMP3]], align 2
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <16 x i16> [[WIDE_VEC]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <16 x i16> [[WIDE_VEC]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i32 [[OFFSET_IDX]]
+; CHECK-NEXT: [[DOTIDX:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i32 [[DOTIDX]]
; CHECK-NEXT: [[WIDE_VEC2:%.*]] = load <16 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT: [[STRIDED_VEC3:%.*]] = shufflevector <16 x i16> [[WIDE_VEC2]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
; CHECK-NEXT: [[STRIDED_VEC4:%.*]] = shufflevector <16 x i16> [[WIDE_VEC2]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
@@ -1791,20 +1833,24 @@ define i64 @test_fir_q15(ptr %x, ptr %y, i32 %n) #0 {
; CHECK: for.body:
; CHECK-NEXT: [[I_025:%.*]] = phi i32 [ [[ADD13:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[S_024:%.*]] = phi i64 [ [[ADD12]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_025]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i32 [[I_025]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 [[TMP23]]
; CHECK-NEXT: [[TMP18:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP18]] to i32
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, ptr [[Y]], i32 [[I_025]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nsw i32 [[I_025]], 1
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[Y]], i32 [[TMP25]]
; CHECK-NEXT: [[TMP19:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2
; CHECK-NEXT: [[CONV2:%.*]] = sext i16 [[TMP19]] to i32
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV2]], [[CONV]]
; CHECK-NEXT: [[CONV3:%.*]] = sext i32 [[MUL]] to i64
; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[S_024]], [[CONV3]]
; CHECK-NEXT: [[ADD4:%.*]] = or disjoint i32 [[I_025]], 1
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[ADD4]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i32 [[ADD4]], 1
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 [[TMP22]]
; CHECK-NEXT: [[TMP20:%.*]] = load i16, ptr [[ARRAYIDX5]], align 2
; CHECK-NEXT: [[CONV6:%.*]] = sext i16 [[TMP20]] to i32
-; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i16, ptr [[Y]], i32 [[ADD4]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl nsw i32 [[ADD4]], 1
+; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, ptr [[Y]], i32 [[TMP24]]
; CHECK-NEXT: [[TMP21:%.*]] = load i16, ptr [[ARRAYIDX8]], align 2
; CHECK-NEXT: [[CONV9:%.*]] = sext i16 [[TMP21]] to i32
; CHECK-NEXT: [[MUL10:%.*]] = mul nsw i32 [[CONV9]], [[CONV6]]
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/tail-fold-multiple-icmps.ll b/llvm/test/Transforms/LoopVectorize/ARM/tail-fold-multiple-icmps.ll
index 6f7b45beeb6e7..fbaed97d418db 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/tail-fold-multiple-icmps.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/tail-fold-multiple-icmps.ll
@@ -18,7 +18,8 @@ define arm_aapcs_vfpcc i32 @minmaxval4(ptr nocapture readonly %x, ptr nocapture
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 2147483647), [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ splat (i32 -2147483648), [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i32 [[TMP8]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[WIDE_LOAD]], <4 x i32> [[VEC_PHI1]])
; CHECK-NEXT: [[TMP2]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[WIDE_LOAD]], <4 x i32> [[VEC_PHI]])
@@ -44,7 +45,8 @@ define arm_aapcs_vfpcc i32 @minmaxval4(ptr nocapture readonly %x, ptr nocapture
; CHECK-NEXT: [[I_029:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[MIN_028:%.*]] = phi i32 [ [[COND9]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[MAX_027:%.*]] = phi i32 [ [[COND]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX2]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[X]], i32 [[I_029]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i32 [[I_029]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 [[TMP7]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[COND]] = call i32 @llvm.smax.i32(i32 [[TMP6]], i32 [[MAX_027]])
; CHECK-NEXT: [[COND9]] = call i32 @llvm.smin.i32(i32 [[TMP6]], i32 [[MIN_028]])
diff --git a/llvm/test/Transforms/LoopVectorize/PowerPC/vsx-tsvc-s173.ll b/llvm/test/Transforms/LoopVectorize/PowerPC/vsx-tsvc-s173.ll
index 372dfbc2e3cd2..bb8ba148bbd1e 100644
--- a/llvm/test/Transforms/LoopVectorize/PowerPC/vsx-tsvc-s173.ll
+++ b/llvm/test/Transforms/LoopVectorize/PowerPC/vsx-tsvc-s173.ll
@@ -24,7 +24,8 @@ define signext i32 @s173() #0 {
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr @global_data, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr @global_data, i64 [[TMP36]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 32
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 48
@@ -40,7 +41,8 @@ define signext i32 @s173() #0 {
; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <4 x float>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[WIDE_LOAD6:%.*]] = load <4 x float>, ptr [[TMP7]], align 4
; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x float>, ptr [[TMP8]], align 4
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr getelementptr inbounds nuw (i8, ptr @global_data, i64 128016), i64 [[INDEX]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @global_data, i64 128016), i64 [[TMP37]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP9]], i64 16
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP9]], i64 32
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP9]], i64 48
@@ -64,7 +66,8 @@ define signext i32 @s173() #0 {
; CHECK-NEXT: [[TMP22:%.*]] = fadd <4 x float> [[WIDE_LOAD5]], [[WIDE_LOAD13]]
; CHECK-NEXT: [[TMP23:%.*]] = fadd <4 x float> [[WIDE_LOAD6]], [[WIDE_LOAD14]]
; CHECK-NEXT: [[TMP24:%.*]] = fadd <4 x float> [[WIDE_LOAD7]], [[WIDE_LOAD15]]
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr float, ptr @global_data, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP38:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr i8, ptr @global_data, i64 [[TMP38]]
; CHECK-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[TMP25]], i64 64000
; CHECK-NEXT: [[TMP27:%.*]] = getelementptr i8, ptr [[TMP25]], i64 64016
; CHECK-NEXT: [[TMP28:%.*]] = getelementptr i8, ptr [[TMP25]], i64 64032
diff --git a/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll b/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll
index 8dcc211d22cc4..22c2df0f16675 100644
--- a/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll
+++ b/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-call.ll
@@ -10,7 +10,8 @@ define dso_local double @test(ptr %Arr) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x double> [ zeroinitializer, [[ENTRY]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw float, ptr [[ARR:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = fpext <2 x float> [[WIDE_LOAD]] to <2 x double>
; CHECK-NEXT: [[TMP2:%.*]] = tail call fast <2 x double> @__sind2(<2 x double> [[TMP1]])
diff --git a/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-vfabi-attr.ll b/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-vfabi-attr.ll
index 172d77d849223..9b76816543bea 100644
--- a/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-vfabi-attr.ll
+++ b/llvm/test/Transforms/LoopVectorize/PowerPC/widened-massv-vfabi-attr.ll
@@ -9,7 +9,8 @@ define dso_local double @test(ptr %Arr) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x double> [ zeroinitializer, [[ENTRY]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[ARR:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[ARR:%.*]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = fpext <2 x float> [[WIDE_LOAD]] to <2 x double>
; CHECK-NEXT: [[TMP3:%.*]] = call fast <2 x double> @__sind2_P8(<2 x double> [[TMP2]])
diff --git a/llvm/test/Transforms/LoopVectorize/SystemZ/addressing.ll b/llvm/test/Transforms/LoopVectorize/SystemZ/addressing.ll
index 1153d18656004..a64785f78fa62 100644
--- a/llvm/test/Transforms/LoopVectorize/SystemZ/addressing.ll
+++ b/llvm/test/Transforms/LoopVectorize/SystemZ/addressing.ll
@@ -14,9 +14,9 @@ define void @foo(ptr nocapture %A) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[DOTIDX1:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[DOTIDX]]
-; CHECK-NEXT: [[DOTIDX1:%.*]] = shl i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[A]], i64 [[DOTIDX1]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 16
; CHECK-NEXT: store i32 4, ptr [[TMP2]], align 4
@@ -57,8 +57,10 @@ define void @foo1(ptr nocapture noalias %A, ptr nocapture %PtrPtr) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds ptr, ptr [[PTRPTR:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr ptr, ptr [[PTRPTR]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[PTRPTR:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[PTRPTR]], i64 [[TMP12]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP11]], i64 8
; CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP1]], align 8
; CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP2]], align 8
@@ -66,7 +68,8 @@ define void @foo1(ptr nocapture noalias %A, ptr nocapture %PtrPtr) {
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <2 x i32> poison, i32 [[TMP5]], i64 0
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <2 x i32> [[TMP7]], i32 [[TMP6]], i64 1
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP13]]
; CHECK-NEXT: store <2 x i32> [[TMP8]], ptr [[TMP9]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10000
diff --git a/llvm/test/Transforms/LoopVectorize/X86/consecutive-ptr-uniforms.ll b/llvm/test/Transforms/LoopVectorize/X86/consecutive-ptr-uniforms.ll
index 7bea9e07d7224..1f0f84279f6f3 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/consecutive-ptr-uniforms.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/consecutive-ptr-uniforms.ll
@@ -41,7 +41,8 @@ define void @PR31671(float %x, ptr %d) #0 {
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <80 x float>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <80 x float> [[WIDE_VEC]], <80 x float> poison, <16 x i32> <i32 0, i32 5, i32 10, i32 15, i32 20, i32 25, i32 30, i32 35, i32 40, i32 45, i32 50, i32 55, i32 60, i32 65, i32 70, i32 75>
; CHECK-NEXT: [[TMP1:%.*]] = fmul <16 x float> [[BROADCAST_SPLAT]], [[STRIDED_VEC]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[D]], <16 x i64> [[VEC_IND]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw <16 x i64> [[VEC_IND]], splat (i64 2)
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[D]], <16 x i64> [[TMP6]]
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <16 x ptr> [[TMP2]], i64 0
; CHECK-NEXT: [[WIDE_VEC1:%.*]] = load <80 x float>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = shufflevector <80 x float> [[WIDE_VEC1]], <80 x float> poison, <16 x i32> <i32 0, i32 5, i32 10, i32 15, i32 20, i32 25, i32 30, i32 35, i32 40, i32 45, i32 50, i32 55, i32 60, i32 65, i32 70, i32 75>
@@ -179,7 +180,11 @@ define void @PR40816() #1 {
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[TMP0:%.*]] = phi i32 [ 0, %[[ENTRY]] ], [ [[INC:%.*]], %[[FOR_BODY]] ]
; CHECK-NEXT: store i32 [[TMP0]], ptr @b, align 1
-; CHECK-NEXT: [[CMP2:%.*]] = icmp eq i32 [[TMP0]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[TMP0]] to i64
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr @a, i64 [[TMP2]]
+; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX1]], align 1
+; CHECK-NEXT: [[CMP2:%.*]] = icmp eq i32 [[TMP3]], 0
; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[TMP0]], 1
; CHECK-NEXT: br i1 [[CMP2]], label %[[RETURN:.*]], label %[[FOR_BODY]]
; CHECK: [[RETURN]]:
diff --git a/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll b/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll
index e2329fe31cd56..1efb093a0bd22 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/interleaving.ll
@@ -13,9 +13,9 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; SSE-NEXT: br label [[VECTOR_BODY:%.*]]
; SSE: vector.body:
; SSE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; SSE-NEXT: [[DOTIDX5:%.*]] = shl i64 [[INDEX]], 3
; SSE-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[INDEX]], 3
; SSE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[DOTIDX]]
-; SSE-NEXT: [[DOTIDX5:%.*]] = shl i64 [[INDEX]], 3
; SSE-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[B]], i64 [[DOTIDX5]]
; SSE-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 32
; SSE-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP2]], align 4
@@ -26,7 +26,8 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; SSE-NEXT: [[STRIDED_VEC4:%.*]] = shufflevector <8 x i32> [[WIDE_VEC2]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[STRIDED_VEC1]], [[STRIDED_VEC]]
; SSE-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[STRIDED_VEC4]], [[STRIDED_VEC3]]
-; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; SSE-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; SSE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP9]]
; SSE-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP6]], i64 16
; SSE-NEXT: store <4 x i32> [[TMP4]], ptr [[TMP6]], align 4
; SSE-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP7]], align 4
@@ -45,15 +46,15 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; AVX1-NEXT: br label [[VECTOR_BODY:%.*]]
; AVX1: vector.body:
; AVX1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX1-NEXT: [[DOTIDX11:%.*]] = shl i64 [[INDEX]], 3
+; AVX1-NEXT: [[DOTIDX12:%.*]] = shl i64 [[INDEX]], 3
+; AVX1-NEXT: [[DOTIDX13:%.*]] = shl i64 [[INDEX]], 3
; AVX1-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[INDEX]], 3
; AVX1-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[DOTIDX]]
-; AVX1-NEXT: [[DOTIDX11:%.*]] = shl i64 [[INDEX]], 3
; AVX1-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[B]], i64 [[DOTIDX11]]
; AVX1-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP1]], i64 32
-; AVX1-NEXT: [[DOTIDX12:%.*]] = shl i64 [[INDEX]], 3
; AVX1-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[B]], i64 [[DOTIDX12]]
; AVX1-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP3]], i64 64
-; AVX1-NEXT: [[DOTIDX13:%.*]] = shl i64 [[INDEX]], 3
; AVX1-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[DOTIDX13]]
; AVX1-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP5]], i64 96
; AVX1-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP6]], align 4
@@ -72,7 +73,8 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; AVX1-NEXT: [[TMP11:%.*]] = add nsw <4 x i32> [[STRIDED_VEC4]], [[STRIDED_VEC3]]
; AVX1-NEXT: [[TMP12:%.*]] = add nsw <4 x i32> [[STRIDED_VEC7]], [[STRIDED_VEC6]]
; AVX1-NEXT: [[TMP13:%.*]] = add nsw <4 x i32> [[STRIDED_VEC10]], [[STRIDED_VEC9]]
-; AVX1-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; AVX1-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; AVX1-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP19]]
; AVX1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP14]], i64 16
; AVX1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP14]], i64 32
; AVX1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP14]], i64 48
@@ -95,15 +97,15 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; AVX2-NEXT: br label [[VECTOR_BODY:%.*]]
; AVX2: vector.body:
; AVX2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; AVX2-NEXT: [[DOTIDX11:%.*]] = shl i64 [[INDEX]], 3
+; AVX2-NEXT: [[DOTIDX12:%.*]] = shl i64 [[INDEX]], 3
+; AVX2-NEXT: [[DOTIDX13:%.*]] = shl i64 [[INDEX]], 3
; AVX2-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[INDEX]], 3
; AVX2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[DOTIDX]]
-; AVX2-NEXT: [[DOTIDX11:%.*]] = shl i64 [[INDEX]], 3
; AVX2-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[B]], i64 [[DOTIDX11]]
; AVX2-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP1]], i64 64
-; AVX2-NEXT: [[DOTIDX12:%.*]] = shl i64 [[INDEX]], 3
; AVX2-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[B]], i64 [[DOTIDX12]]
; AVX2-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP3]], i64 128
-; AVX2-NEXT: [[DOTIDX13:%.*]] = shl i64 [[INDEX]], 3
; AVX2-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[DOTIDX13]]
; AVX2-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP5]], i64 192
; AVX2-NEXT: [[WIDE_VEC:%.*]] = load <16 x i32>, ptr [[TMP6]], align 4
@@ -122,7 +124,8 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; AVX2-NEXT: [[TMP11:%.*]] = add nsw <8 x i32> [[STRIDED_VEC4]], [[STRIDED_VEC3]]
; AVX2-NEXT: [[TMP12:%.*]] = add nsw <8 x i32> [[STRIDED_VEC7]], [[STRIDED_VEC6]]
; AVX2-NEXT: [[TMP13:%.*]] = add nsw <8 x i32> [[STRIDED_VEC10]], [[STRIDED_VEC9]]
-; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; AVX2-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; AVX2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP19]]
; AVX2-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP14]], i64 32
; AVX2-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP14]], i64 64
; AVX2-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP14]], i64 96
@@ -145,14 +148,16 @@ define void @foo(ptr noalias nocapture %a, ptr noalias nocapture readonly %b) {
; ATOM-NEXT: ret void
; ATOM: for.body:
; ATOM-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; ATOM-NEXT: [[TMP0:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
-; ATOM-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[TMP0]]
+; ATOM-NEXT: [[TMP0:%.*]] = shl i64 [[INDVARS_IV]], 3
+; ATOM-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 3
+; ATOM-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; ATOM-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; ATOM-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[TMP0]]
-; ATOM-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 4
+; ATOM-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP0]]
+; ATOM-NEXT: [[ARRAYIDX3:%.*]] = getelementptr i8, ptr [[TMP4]], i64 4
; ATOM-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARRAYIDX3]], align 4
; ATOM-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP3]], [[TMP1]]
-; ATOM-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; ATOM-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; ATOM-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP5]]
; ATOM-NEXT: store i32 [[ADD4]], ptr [[ARRAYIDX6]], align 4
; ATOM-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; ATOM-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 1024
diff --git a/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll b/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
index d514ab6bc72b7..30a749ce5213a 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/metadata-enable.ll
@@ -807,10 +807,11 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O1-NEXT: br label [[FOR_BODY:%.*]]
; O1: for.body:
; O1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; O1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; O1-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; O1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; O1-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; O1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; O1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; O1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; O1-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; O1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; O1-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 64
@@ -1164,10 +1165,11 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; Oz-NEXT: br label [[FOR_BODY:%.*]]
; Oz: for.body:
; Oz-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; Oz-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; Oz-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; Oz-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; Oz-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; Oz-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; Oz-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; Oz-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; Oz-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; Oz-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; Oz-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 64
@@ -1185,13 +1187,14 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O1VEC2-NEXT: br label [[VECTOR_BODY:%.*]]
; O1VEC2: vector.body:
; O1VEC2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; O1VEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDEX]]
+; O1VEC2-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; O1VEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP0]]
; O1VEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i64 4
; O1VEC2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; O1VEC2-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
; O1VEC2-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; O1VEC2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[BROADCAST_SPLAT]]
-; O1VEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDEX]]
+; O1VEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP0]]
; O1VEC2-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP6]], i64 4
; O1VEC2-NEXT: store <4 x i32> [[TMP4]], ptr [[TMP6]], align 4
; O1VEC2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP8]], align 4
@@ -1213,13 +1216,14 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; OzVEC2-NEXT: br label [[VECTOR_BODY:%.*]]
; OzVEC2: vector.body:
; OzVEC2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; OzVEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDEX]]
+; OzVEC2-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; OzVEC2-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP0]]
; OzVEC2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i64 4
; OzVEC2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; OzVEC2-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
; OzVEC2-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; OzVEC2-NEXT: [[TMP5:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[BROADCAST_SPLAT]]
-; OzVEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDEX]]
+; OzVEC2-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP0]]
; OzVEC2-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP6]], i64 4
; OzVEC2-NEXT: store <4 x i32> [[TMP4]], ptr [[TMP6]], align 4
; OzVEC2-NEXT: store <4 x i32> [[TMP5]], ptr [[TMP8]], align 4
@@ -1237,10 +1241,11 @@ define i32 @nopragma(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O3DIS-NEXT: br label [[FOR_BODY:%.*]]
; O3DIS: for.body:
; O3DIS-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; O3DIS-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; O3DIS-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; O3DIS-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; O3DIS-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; O3DIS-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; O3DIS-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; O3DIS-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; O3DIS-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; O3DIS-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; O3DIS-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 64
@@ -1274,10 +1279,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O1-NEXT: br label [[FOR_BODY:%.*]]
; O1: for.body:
; O1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; O1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; O1-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; O1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; O1-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; O1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; O1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; O1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; O1-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; O1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; O1-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
@@ -1291,10 +1297,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O2-NEXT: br label [[FOR_BODY:%.*]]
; O2: for.body:
; O2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; O2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; O2-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; O2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; O2-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; O2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; O2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; O2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; O2-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; O2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; O2-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
@@ -1308,10 +1315,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O3-NEXT: br label [[FOR_BODY:%.*]]
; O3: for.body:
; O3-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; O3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; O3-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; O3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; O3-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; O3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; O3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; O3-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; O3-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; O3-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; O3-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
@@ -1390,10 +1398,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; Os-NEXT: br label [[FOR_BODY:%.*]]
; Os: for.body:
; Os-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; Os-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; Os-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; Os-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; Os-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; Os-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; Os-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; Os-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; Os-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; Os-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; Os-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
@@ -1407,10 +1416,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; Oz-NEXT: br label [[FOR_BODY:%.*]]
; Oz: for.body:
; Oz-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; Oz-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; Oz-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; Oz-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; Oz-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; Oz-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; Oz-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; Oz-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; Oz-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; Oz-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; Oz-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
@@ -1424,10 +1434,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O1VEC2-NEXT: br label [[FOR_BODY:%.*]]
; O1VEC2: for.body:
; O1VEC2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; O1VEC2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; O1VEC2-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; O1VEC2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; O1VEC2-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; O1VEC2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; O1VEC2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; O1VEC2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; O1VEC2-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; O1VEC2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; O1VEC2-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
@@ -1441,10 +1452,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; OzVEC2-NEXT: br label [[FOR_BODY:%.*]]
; OzVEC2: for.body:
; OzVEC2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; OzVEC2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; OzVEC2-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; OzVEC2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; OzVEC2-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; OzVEC2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; OzVEC2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; OzVEC2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; OzVEC2-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; OzVEC2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; OzVEC2-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
@@ -1458,10 +1470,11 @@ define i32 @disabled(ptr noalias nocapture %a, ptr noalias nocapture readonly %b
; O3DIS-NEXT: br label [[FOR_BODY:%.*]]
; O3DIS: for.body:
; O3DIS-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; O3DIS-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; O3DIS-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; O3DIS-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP2]]
; O3DIS-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; O3DIS-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], [[N:%.*]]
-; O3DIS-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; O3DIS-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP2]]
; O3DIS-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX2]], align 4
; O3DIS-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; O3DIS-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 48
diff --git a/llvm/test/Transforms/LoopVectorize/X86/parallel-loops.ll b/llvm/test/Transforms/LoopVectorize/X86/parallel-loops.ll
index 66809ebb715f0..c2105a49197f0 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/parallel-loops.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/parallel-loops.ll
@@ -19,15 +19,19 @@ define void @loop(ptr nocapture %a, ptr nocapture %b) nounwind uwtable {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP7]]
; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[IDXPROM3:%.*]] = sext i32 [[TMP1]] to i64
-; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IDXPROM3]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[IDXPROM3]], 2
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP4]]
; CHECK-NEXT: store i32 [[TMP0]], ptr [[ARRAYIDX4]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT]], 2
+; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX6]], align 4
; CHECK-NEXT: store i32 [[TMP2]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = and i64 [[INDVARS_IV_NEXT]], 4294967295
@@ -71,20 +75,25 @@ define void @parallel_loop(ptr nocapture %a, ptr nocapture %b) nounwind uwtable
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4, !llvm.access.group [[ACC_GRP0:![0-9]+]]
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i32> [[WIDE_LOAD]], i64 0
; CHECK-NEXT: [[TMP21:%.*]] = extractelement <4 x i32> [[WIDE_LOAD]], i64 1
; CHECK-NEXT: [[TMP22:%.*]] = extractelement <4 x i32> [[WIDE_LOAD]], i64 2
; CHECK-NEXT: [[TMP23:%.*]] = extractelement <4 x i32> [[WIDE_LOAD]], i64 3
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP17]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP2]], i64 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[TMP4]], i64 8
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP24]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP6]], i64 12
-; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4, !llvm.access.group [[ACC_GRP0]]
+; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP18]], align 4, !llvm.access.group [[ACC_GRP0]]
; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP8]], align 4, !llvm.access.group [[ACC_GRP0]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP26]], align 4, !llvm.access.group [[ACC_GRP0]]
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4, !llvm.access.group [[ACC_GRP0]]
@@ -92,18 +101,23 @@ define void @parallel_loop(ptr nocapture %a, ptr nocapture %b) nounwind uwtable
; CHECK-NEXT: [[TMP14:%.*]] = sext i32 [[TMP10]] to i64
; CHECK-NEXT: [[TMP15:%.*]] = sext i32 [[TMP11]] to i64
; CHECK-NEXT: [[TMP16:%.*]] = sext i32 [[TMP12]] to i64
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP13]]
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP15]]
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP16]]
-; CHECK-NEXT: store i32 [[TMP1]], ptr [[TMP17]], align 4, !llvm.access.group [[ACC_GRP1:![0-9]+]]
-; CHECK-NEXT: store i32 [[TMP21]], ptr [[TMP18]], align 4, !llvm.access.group [[ACC_GRP1]]
-; CHECK-NEXT: store i32 [[TMP22]], ptr [[TMP19]], align 4, !llvm.access.group [[ACC_GRP1]]
-; CHECK-NEXT: store i32 [[TMP23]], ptr [[TMP20]], align 4, !llvm.access.group [[ACC_GRP1]]
-; CHECK-NEXT: [[TMP27:%.*]] = getelementptr i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl nsw i64 [[TMP13]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP34]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[TMP14]], 2
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP36]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl nsw i64 [[TMP15]], 2
+; CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP37]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nsw i64 [[TMP16]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: store i32 [[TMP1]], ptr [[TMP35]], align 4, !llvm.access.group [[ACC_GRP1:![0-9]+]]
+; CHECK-NEXT: store i32 [[TMP21]], ptr [[TMP28]], align 4, !llvm.access.group [[ACC_GRP1]]
+; CHECK-NEXT: store i32 [[TMP22]], ptr [[TMP30]], align 4, !llvm.access.group [[ACC_GRP1]]
+; CHECK-NEXT: store i32 [[TMP23]], ptr [[TMP32]], align 4, !llvm.access.group [[ACC_GRP1]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP27:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP33]]
; CHECK-NEXT: [[TMP25:%.*]] = getelementptr i8, ptr [[TMP27]], i64 4
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP25]], align 4, !llvm.access.group [[ACC_GRP0]]
-; CHECK-NEXT: store <4 x i32> [[WIDE_LOAD1]], ptr [[TMP5]], align 4, !llvm.access.group [[ACC_GRP0]]
+; CHECK-NEXT: store <4 x i32> [[WIDE_LOAD1]], ptr [[TMP18]], align 4, !llvm.access.group [[ACC_GRP0]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], 512
; CHECK-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
@@ -148,20 +162,24 @@ define void @mixed_metadata(ptr nocapture %a, ptr nocapture %b) nounwind uwtable
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4, !llvm.access.group [[ACC_GRP7]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4, !llvm.access.group [[ACC_GRP6]]
; CHECK-NEXT: [[IDXPROM3:%.*]] = sext i32 [[TMP1]] to i64
-; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IDXPROM3]]
-; CHECK-NEXT: store i32 [[TMP0]], ptr [[ARRAYIDX4]], align 4, !llvm.access.group [[ACC_GRP8:![0-9]+]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[IDXPROM3]], 2
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP4]]
+; CHECK-NEXT: store i32 [[TMP7]], ptr [[ARRAYIDX4]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV_NEXT]]
-; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX6]], align 4, !llvm.access.group [[ACC_GRP7]]
-; CHECK-NEXT: store i32 [[TMP2]], ptr [[ARRAYIDX2]], align 4, !llvm.access.group [[ACC_GRP7]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT]], 2
+; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX6]], align 4, !llvm.access.group [[ACC_GRP6]]
+; CHECK-NEXT: store i32 [[TMP6]], ptr [[ARRAYIDX2]], align 4, !llvm.access.group [[ACC_GRP6]]
; CHECK-NEXT: [[TMP3:%.*]] = and i64 [[INDVARS_IV_NEXT]], 4294967295
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[TMP3]], 512
-; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/LoopVectorize/X86/small-size.ll b/llvm/test/Transforms/LoopVectorize/X86/small-size.ll
index 6e74d5afe57ca..bfc6cd5a7115c 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/small-size.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/small-size.ll
@@ -28,19 +28,22 @@ define void @example1() optsize {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; 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: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr @b, i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr @c, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr @c, i64 [[TMP8]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr @a, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr @a, i64 [[TMP9]]
; CHECK-NEXT: store <4 x i32> [[TMP3]], ptr [[TMP4]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[TMP6:%.*]]
-; CHECK: 6:
+; CHECK: 9:
; CHECK-NEXT: ret void
;
br label %1
@@ -85,14 +88,16 @@ define void @example2(i32 %n, i32 %x) optsize {
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i1> [[TMP3]], i64 0
; CHECK-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr @b, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr @b, i64 [[TMP51]]
; CHECK-NEXT: store i32 [[X:%.*]], ptr [[TMP5]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE]]
; CHECK: pred.store.continue:
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP3]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2:%.*]]
; CHECK: pred.store.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr @b, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr @b, i64 [[TMP53]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 4
; CHECK-NEXT: store i32 [[X]], ptr [[TMP8]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE2]]
@@ -100,7 +105,8 @@ define void @example2(i32 %n, i32 %x) optsize {
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP3]], i64 2
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]]
; CHECK: pred.store.if3:
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr @b, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr @b, i64 [[TMP54]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP10]], i64 8
; CHECK-NEXT: store i32 [[X]], ptr [[TMP11]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE4]]
@@ -108,7 +114,8 @@ define void @example2(i32 %n, i32 %x) optsize {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP3]], i64 3
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6]]
; CHECK: pred.store.if5:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr @b, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP55:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr @b, i64 [[TMP55]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 12
; CHECK-NEXT: store i32 [[X]], ptr [[TMP14]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE6]]
@@ -146,11 +153,14 @@ define void @example2(i32 %n, i32 %x) optsize {
; CHECK-NEXT: [[TMP19:%.*]] = extractelement <4 x i1> [[TMP18]], i64 0
; CHECK-NEXT: br i1 [[TMP19]], label [[PRED_STORE_IF19:%.*]], label [[PRED_STORE_CONTINUE20:%.*]]
; CHECK: pred.store.if18:
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr @b, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl nsw i64 [[OFFSET_IDX]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr @b, i64 [[TMP56]]
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr @c, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP58:%.*]] = shl nsw i64 [[OFFSET_IDX]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr @c, i64 [[TMP58]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds i32, ptr @a, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP59:%.*]] = shl nsw i64 [[OFFSET_IDX]], 2
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr @a, i64 [[TMP59]]
; CHECK-NEXT: [[TMP25:%.*]] = and i32 [[TMP23]], [[TMP21]]
; CHECK-NEXT: store i32 [[TMP25]], ptr [[TMP24]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE20]]
@@ -159,11 +169,14 @@ define void @example2(i32 %n, i32 %x) optsize {
; CHECK-NEXT: br i1 [[TMP26]], label [[PRED_STORE_IF21:%.*]], label [[PRED_STORE_CONTINUE22:%.*]]
; CHECK: pred.store.if20:
; CHECK-NEXT: [[TMP27:%.*]] = add i64 [[OFFSET_IDX]], 1
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr @b, i64 [[TMP27]]
+; CHECK-NEXT: [[TMP61:%.*]] = shl nsw i64 [[TMP27]], 2
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr @b, i64 [[TMP61]]
; CHECK-NEXT: [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
-; CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds i32, ptr @c, i64 [[TMP27]]
+; CHECK-NEXT: [[TMP62:%.*]] = shl nsw i64 [[TMP27]], 2
+; CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr @c, i64 [[TMP62]]
; CHECK-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr @a, i64 [[TMP27]]
+; CHECK-NEXT: [[TMP64:%.*]] = shl nsw i64 [[TMP27]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr @a, i64 [[TMP64]]
; CHECK-NEXT: [[TMP33:%.*]] = and i32 [[TMP31]], [[TMP29]]
; CHECK-NEXT: store i32 [[TMP33]], ptr [[TMP32]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE22]]
@@ -172,11 +185,14 @@ define void @example2(i32 %n, i32 %x) optsize {
; CHECK-NEXT: br i1 [[TMP34]], label [[PRED_STORE_IF23:%.*]], label [[PRED_STORE_CONTINUE24:%.*]]
; CHECK: pred.store.if22:
; CHECK-NEXT: [[TMP35:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr @b, i64 [[TMP35]]
+; CHECK-NEXT: [[TMP65:%.*]] = shl nsw i64 [[TMP35]], 2
+; CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr @b, i64 [[TMP65]]
; CHECK-NEXT: [[TMP37:%.*]] = load i32, ptr [[TMP36]], align 4
-; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds i32, ptr @c, i64 [[TMP35]]
+; CHECK-NEXT: [[TMP66:%.*]] = shl nsw i64 [[TMP35]], 2
+; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr @c, i64 [[TMP66]]
; CHECK-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP38]], align 4
-; CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds i32, ptr @a, i64 [[TMP35]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP35]], 2
+; CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr @a, i64 [[TMP52]]
; CHECK-NEXT: [[TMP41:%.*]] = and i32 [[TMP39]], [[TMP37]]
; CHECK-NEXT: store i32 [[TMP41]], ptr [[TMP40]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE24]]
@@ -185,11 +201,14 @@ define void @example2(i32 %n, i32 %x) optsize {
; CHECK-NEXT: br i1 [[TMP42]], label [[PRED_STORE_IF25:%.*]], label [[PRED_STORE_CONTINUE26]]
; CHECK: pred.store.if24:
; CHECK-NEXT: [[TMP43:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds i32, ptr @b, i64 [[TMP43]]
+; CHECK-NEXT: [[TMP57:%.*]] = shl nsw i64 [[TMP43]], 2
+; CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds i8, ptr @b, i64 [[TMP57]]
; CHECK-NEXT: [[TMP45:%.*]] = load i32, ptr [[TMP44]], align 4
-; CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds i32, ptr @c, i64 [[TMP43]]
+; CHECK-NEXT: [[TMP60:%.*]] = shl nsw i64 [[TMP43]], 2
+; CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds i8, ptr @c, i64 [[TMP60]]
; CHECK-NEXT: [[TMP47:%.*]] = load i32, ptr [[TMP46]], align 4
-; CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds i32, ptr @a, i64 [[TMP43]]
+; CHECK-NEXT: [[TMP63:%.*]] = shl nsw i64 [[TMP43]], 2
+; CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds i8, ptr @a, i64 [[TMP63]]
; CHECK-NEXT: [[TMP49:%.*]] = and i32 [[TMP47]], [[TMP45]]
; CHECK-NEXT: store i32 [[TMP49]], ptr [[TMP48]], align 4
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE26]]
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 cfc588a7f5d89..8d2d2c5b996c5 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
@@ -25,7 +25,8 @@ define dso_local void @test1(ptr noalias nocapture %points, ptr noalias nocaptur
; DISABLED_MASKED_STRIDED: vector.body:
; DISABLED_MASKED_STRIDED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; DISABLED_MASKED_STRIDED-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[INDEX]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP29:%.*]] = shl nsw i64 [[INDEX]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP29]]
; DISABLED_MASKED_STRIDED-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP0]], align 2
; DISABLED_MASKED_STRIDED-NEXT: [[TMP10:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: [[TMP11:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 1
@@ -36,15 +37,20 @@ define dso_local void @test1(ptr noalias nocapture %points, ptr noalias nocaptur
; DISABLED_MASKED_STRIDED-NEXT: [[TMP4:%.*]] = extractelement <4 x i64> [[TMP1]], i64 1
; DISABLED_MASKED_STRIDED-NEXT: [[TMP6:%.*]] = extractelement <4 x i64> [[TMP1]], i64 2
; DISABLED_MASKED_STRIDED-NEXT: [[TMP8:%.*]] = extractelement <4 x i64> [[TMP1]], i64 3
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS:%.*]], i64 [[TMP2]]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP4]]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP6]]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP8]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP31:%.*]] = shl nuw nsw i64 [[TMP2]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS:%.*]], i64 [[TMP31]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP33:%.*]] = shl nuw nsw i64 [[TMP4]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP33]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP35:%.*]] = shl nuw nsw i64 [[TMP6]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP35]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP37:%.*]] = shl nuw nsw i64 [[TMP8]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP37]]
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP10]], ptr [[TMP3]], align 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP11]], ptr [[TMP5]], align 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP12]], ptr [[TMP7]], align 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP13]], ptr [[TMP9]], align 2
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP14:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[INDEX]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP38:%.*]] = shl nsw i64 [[INDEX]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i64 [[TMP38]]
; DISABLED_MASKED_STRIDED-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i16>, ptr [[TMP14]], align 2
; DISABLED_MASKED_STRIDED-NEXT: [[TMP24:%.*]] = extractelement <4 x i16> [[WIDE_LOAD1]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: [[TMP25:%.*]] = extractelement <4 x i16> [[WIDE_LOAD1]], i64 1
@@ -55,10 +61,14 @@ define dso_local void @test1(ptr noalias nocapture %points, ptr noalias nocaptur
; DISABLED_MASKED_STRIDED-NEXT: [[TMP18:%.*]] = extractelement <4 x i64> [[TMP15]], i64 1
; DISABLED_MASKED_STRIDED-NEXT: [[TMP20:%.*]] = extractelement <4 x i64> [[TMP15]], i64 2
; DISABLED_MASKED_STRIDED-NEXT: [[TMP22:%.*]] = extractelement <4 x i64> [[TMP15]], i64 3
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP16]]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP18]]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP20]]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP22]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP30:%.*]] = shl nuw nsw i64 [[TMP16]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP30]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP32:%.*]] = shl nuw nsw i64 [[TMP18]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP32]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP34:%.*]] = shl nuw nsw i64 [[TMP20]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP34]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP36:%.*]] = shl nuw nsw i64 [[TMP22]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP36]]
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP24]], ptr [[TMP17]], align 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP25]], ptr [[TMP19]], align 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP26]], ptr [[TMP21]], align 2
@@ -75,11 +85,13 @@ define dso_local void @test1(ptr noalias nocapture %points, ptr noalias nocaptur
; 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]] ]
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[INDEX]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP5]]
; ENABLED_MASKED_STRIDED-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP0]], align 2
; ENABLED_MASKED_STRIDED-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 3
; ENABLED_MASKED_STRIDED-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[POINTS:%.*]], i64 [[TMP1]]
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[INDEX]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDEX]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i64 [[TMP3]]
; ENABLED_MASKED_STRIDED-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i16>, ptr [[TMP2]], align 2
; ENABLED_MASKED_STRIDED-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i16> [[WIDE_LOAD]], <4 x i16> [[WIDE_LOAD1]], <16 x i32> <i32 0, i32 4, i32 poison, i32 poison, i32 1, i32 5, i32 poison, i32 poison, i32 2, i32 6, i32 poison, i32 poison, i32 3, i32 7, i32 poison, i32 poison>
; ENABLED_MASKED_STRIDED-NEXT: call void @llvm.masked.store.v16i16.p0(<16 x i16> [[INTERLEAVED_VEC]], ptr align 2 [[GEP]], <16 x i1> <i1 true, i1 true, i1 false, i1 false, i1 true, i1 true, i1 false, i1 false, i1 true, i1 true, i1 false, i1 false, i1 true, i1 true, i1 false, i1 false>)
@@ -139,14 +151,16 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; DISABLED_MASKED_STRIDED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE15:%.*]] ]
; DISABLED_MASKED_STRIDED-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_STORE_CONTINUE15]] ]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = icmp ule <4 x i64> [[VEC_IND]], [[BROADCAST_SPLAT]]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[INDEX]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP38:%.*]] = shl nsw i64 [[INDEX]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP38]]
; DISABLED_MASKED_STRIDED-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr align 2 [[TMP1]], <4 x i1> [[TMP0]], <4 x i16> poison)
; DISABLED_MASKED_STRIDED-NEXT: [[TMP2:%.*]] = shl nsw <4 x i64> [[VEC_IND]], splat (i64 2)
; DISABLED_MASKED_STRIDED-NEXT: [[TMP3:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP3]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP4:%.*]] = extractelement <4 x i64> [[TMP2]], i64 0
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i16, ptr [[POINTS:%.*]], i64 [[TMP4]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP40:%.*]] = shl nsw i64 [[TMP4]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[POINTS:%.*]], i64 [[TMP40]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP6:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP6]], ptr [[TMP5]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE]]
@@ -155,7 +169,8 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP7]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if1:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP8:%.*]] = extractelement <4 x i64> [[TMP2]], i64 1
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds i16, ptr [[POINTS]], i64 [[TMP8]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP41:%.*]] = shl nsw i64 [[TMP8]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[POINTS]], i64 [[TMP41]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP10:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD]], i64 1
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP10]], ptr [[TMP9]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE2]]
@@ -164,7 +179,8 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP11]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if3:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP12:%.*]] = extractelement <4 x i64> [[TMP2]], i64 2
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP13:%.*]] = getelementptr inbounds i16, ptr [[POINTS]], i64 [[TMP12]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP42:%.*]] = shl nsw i64 [[TMP12]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[POINTS]], i64 [[TMP42]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP14:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD]], i64 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP14]], ptr [[TMP13]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE4]]
@@ -173,19 +189,22 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP15]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if5:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP16:%.*]] = extractelement <4 x i64> [[TMP2]], i64 3
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds i16, ptr [[POINTS]], i64 [[TMP16]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP43:%.*]] = shl nsw i64 [[TMP16]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[POINTS]], i64 [[TMP43]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP18:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD]], i64 3
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP18]], ptr [[TMP17]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE6]]
; DISABLED_MASKED_STRIDED: pred.store.continue6:
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP19:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[INDEX]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP45:%.*]] = shl nsw i64 [[INDEX]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i64 [[TMP45]]
; DISABLED_MASKED_STRIDED-NEXT: [[WIDE_MASKED_LOAD7:%.*]] = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr align 2 [[TMP19]], <4 x i1> [[TMP0]], <4 x i16> poison)
; DISABLED_MASKED_STRIDED-NEXT: [[TMP20:%.*]] = or disjoint <4 x i64> [[TMP2]], splat (i64 1)
; DISABLED_MASKED_STRIDED-NEXT: [[TMP21:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP21]], label [[PRED_STORE_IF8:%.*]], label [[PRED_STORE_CONTINUE9:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if8:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP22:%.*]] = extractelement <4 x i64> [[TMP20]], i64 0
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP23:%.*]] = getelementptr inbounds i16, ptr [[POINTS]], i64 [[TMP22]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP46:%.*]] = shl nsw i64 [[TMP22]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[POINTS]], i64 [[TMP46]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP24:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD7]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP24]], ptr [[TMP23]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE9]]
@@ -194,7 +213,8 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP25]], label [[PRED_STORE_IF10:%.*]], label [[PRED_STORE_CONTINUE11:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if10:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP26:%.*]] = extractelement <4 x i64> [[TMP20]], i64 1
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP27:%.*]] = getelementptr inbounds i16, ptr [[POINTS]], i64 [[TMP26]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP47:%.*]] = shl nsw i64 [[TMP26]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[POINTS]], i64 [[TMP47]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP28:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD7]], i64 1
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP28]], ptr [[TMP27]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE11]]
@@ -203,7 +223,8 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP29]], label [[PRED_STORE_IF12:%.*]], label [[PRED_STORE_CONTINUE13:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if12:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP30:%.*]] = extractelement <4 x i64> [[TMP20]], i64 2
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP31:%.*]] = getelementptr inbounds i16, ptr [[POINTS]], i64 [[TMP30]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP39:%.*]] = shl nsw i64 [[TMP30]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[POINTS]], i64 [[TMP39]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP32:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD7]], i64 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP32]], ptr [[TMP31]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE13]]
@@ -212,7 +233,8 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP33]], label [[PRED_STORE_IF14:%.*]], label [[PRED_STORE_CONTINUE15]]
; DISABLED_MASKED_STRIDED: pred.store.if14:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP34:%.*]] = extractelement <4 x i64> [[TMP20]], i64 3
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP35:%.*]] = getelementptr inbounds i16, ptr [[POINTS]], i64 [[TMP34]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP44:%.*]] = shl nsw i64 [[TMP34]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[POINTS]], i64 [[TMP44]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP36:%.*]] = extractelement <4 x i16> [[WIDE_MASKED_LOAD7]], i64 3
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP36]], ptr [[TMP35]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE15]]
@@ -244,11 +266,13 @@ define dso_local void @test2(ptr noalias nocapture %points, i32 %numPoints, ptr
; ENABLED_MASKED_STRIDED-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT1]], <4 x i64> poison, <4 x i32> zeroinitializer
; ENABLED_MASKED_STRIDED-NEXT: [[VEC_IV:%.*]] = or disjoint <4 x i64> [[BROADCAST_SPLAT2]], <i64 0, i64 1, i64 2, i64 3>
; ENABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = icmp ule <4 x i64> [[VEC_IV]], [[BROADCAST_SPLAT]]
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[INDEX]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP7]]
; ENABLED_MASKED_STRIDED-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr align 2 [[TMP1]], <4 x i1> [[TMP0]], <4 x i16> poison)
; ENABLED_MASKED_STRIDED-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDEX]], 3
; ENABLED_MASKED_STRIDED-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[POINTS:%.*]], i64 [[TMP2]]
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i64 [[INDEX]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[Y:%.*]], i64 [[TMP4]]
; ENABLED_MASKED_STRIDED-NEXT: [[WIDE_MASKED_LOAD3:%.*]] = call <4 x i16> @llvm.masked.load.v4i16.p0(ptr align 2 [[TMP3]], <4 x i1> [[TMP0]], <4 x i16> poison)
; ENABLED_MASKED_STRIDED-NEXT: [[INTERLEAVED_VEC:%.*]] = shufflevector <4 x i16> [[WIDE_MASKED_LOAD]], <4 x i16> [[WIDE_MASKED_LOAD3]], <16 x i32> <i32 0, i32 4, i32 poison, i32 poison, i32 1, i32 5, i32 poison, i32 poison, i32 2, i32 6, i32 poison, i32 poison, i32 3, i32 7, i32 poison, i32 poison>
; ENABLED_MASKED_STRIDED-NEXT: [[INTERLEAVED_MASK:%.*]] = shufflevector <4 x i1> [[TMP0]], <4 x i1> poison, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1, i32 2, i32 2, i32 2, i32 2, i32 3, i32 3, i32 3, i32 3>
@@ -311,7 +335,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; DISABLED_MASKED_STRIDED: vector.body:
; DISABLED_MASKED_STRIDED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE6:%.*]] ]
; DISABLED_MASKED_STRIDED-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_STORE_CONTINUE6]] ]
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[INDEX]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDEX]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP20]]
; DISABLED_MASKED_STRIDED-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP0]], align 2
; DISABLED_MASKED_STRIDED-NEXT: [[TMP1:%.*]] = icmp sgt <4 x i16> [[WIDE_LOAD]], zeroinitializer
; DISABLED_MASKED_STRIDED-NEXT: [[TMP2:%.*]] = mul nuw nsw <4 x i64> [[VEC_IND]], splat (i64 3)
@@ -319,7 +344,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP3]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP4:%.*]] = extractelement <4 x i64> [[TMP2]], i64 0
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS:%.*]], i64 [[TMP4]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP22:%.*]] = shl nuw nsw i64 [[TMP4]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS:%.*]], i64 [[TMP22]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP6:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 0
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP6]], ptr [[TMP5]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE]]
@@ -328,7 +354,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP7]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if1:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP8:%.*]] = extractelement <4 x i64> [[TMP2]], i64 1
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP8]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP23:%.*]] = shl nuw nsw i64 [[TMP8]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP23]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP10:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 1
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP10]], ptr [[TMP9]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE2]]
@@ -337,7 +364,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP11]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]]
; DISABLED_MASKED_STRIDED: pred.store.if3:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP12:%.*]] = extractelement <4 x i64> [[TMP2]], i64 2
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP12]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP24:%.*]] = shl nuw nsw i64 [[TMP12]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP24]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP14:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 2
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP14]], ptr [[TMP13]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE4]]
@@ -346,7 +374,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; DISABLED_MASKED_STRIDED-NEXT: br i1 [[TMP15]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6]]
; DISABLED_MASKED_STRIDED: pred.store.if5:
; DISABLED_MASKED_STRIDED-NEXT: [[TMP16:%.*]] = extractelement <4 x i64> [[TMP2]], i64 3
-; DISABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP16]]
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP21:%.*]] = shl nuw nsw i64 [[TMP16]], 1
+; DISABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP21]]
; DISABLED_MASKED_STRIDED-NEXT: [[TMP18:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 3
; DISABLED_MASKED_STRIDED-NEXT: store i16 [[TMP18]], ptr [[TMP17]], align 2
; DISABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE6]]
@@ -364,7 +393,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; ENABLED_MASKED_STRIDED: vector.body:
; ENABLED_MASKED_STRIDED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE6:%.*]] ]
; ENABLED_MASKED_STRIDED-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[ENTRY]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_STORE_CONTINUE6]] ]
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i64 [[INDEX]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDEX]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X:%.*]], i64 [[TMP20]]
; ENABLED_MASKED_STRIDED-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP0]], align 2
; ENABLED_MASKED_STRIDED-NEXT: [[TMP1:%.*]] = icmp sgt <4 x i16> [[WIDE_LOAD]], zeroinitializer
; ENABLED_MASKED_STRIDED-NEXT: [[TMP2:%.*]] = mul nuw nsw <4 x i64> [[VEC_IND]], splat (i64 3)
@@ -372,7 +402,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; ENABLED_MASKED_STRIDED-NEXT: br i1 [[TMP3]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; ENABLED_MASKED_STRIDED: pred.store.if:
; ENABLED_MASKED_STRIDED-NEXT: [[TMP4:%.*]] = extractelement <4 x i64> [[TMP2]], i64 0
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS:%.*]], i64 [[TMP4]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP22:%.*]] = shl nuw nsw i64 [[TMP4]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS:%.*]], i64 [[TMP22]]
; ENABLED_MASKED_STRIDED-NEXT: [[TMP6:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 0
; ENABLED_MASKED_STRIDED-NEXT: store i16 [[TMP6]], ptr [[TMP5]], align 2
; ENABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE]]
@@ -381,7 +412,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; ENABLED_MASKED_STRIDED-NEXT: br i1 [[TMP7]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2:%.*]]
; ENABLED_MASKED_STRIDED: pred.store.if1:
; ENABLED_MASKED_STRIDED-NEXT: [[TMP8:%.*]] = extractelement <4 x i64> [[TMP2]], i64 1
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP8]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP23:%.*]] = shl nuw nsw i64 [[TMP8]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP23]]
; ENABLED_MASKED_STRIDED-NEXT: [[TMP10:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 1
; ENABLED_MASKED_STRIDED-NEXT: store i16 [[TMP10]], ptr [[TMP9]], align 2
; ENABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE2]]
@@ -390,7 +422,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; ENABLED_MASKED_STRIDED-NEXT: br i1 [[TMP11]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]]
; ENABLED_MASKED_STRIDED: pred.store.if3:
; ENABLED_MASKED_STRIDED-NEXT: [[TMP12:%.*]] = extractelement <4 x i64> [[TMP2]], i64 2
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP12]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP24:%.*]] = shl nuw nsw i64 [[TMP12]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP24]]
; ENABLED_MASKED_STRIDED-NEXT: [[TMP14:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 2
; ENABLED_MASKED_STRIDED-NEXT: store i16 [[TMP14]], ptr [[TMP13]], align 2
; ENABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE4]]
@@ -399,7 +432,8 @@ define dso_local void @test(ptr noalias nocapture %points, ptr noalias nocapture
; ENABLED_MASKED_STRIDED-NEXT: br i1 [[TMP15]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6]]
; ENABLED_MASKED_STRIDED: pred.store.if5:
; ENABLED_MASKED_STRIDED-NEXT: [[TMP16:%.*]] = extractelement <4 x i64> [[TMP2]], i64 3
-; ENABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i16, ptr [[POINTS]], i64 [[TMP16]]
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP21:%.*]] = shl nuw nsw i64 [[TMP16]], 1
+; ENABLED_MASKED_STRIDED-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[POINTS]], i64 [[TMP21]]
; ENABLED_MASKED_STRIDED-NEXT: [[TMP18:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 3
; ENABLED_MASKED_STRIDED-NEXT: store i16 [[TMP18]], ptr [[TMP17]], align 2
; ENABLED_MASKED_STRIDED-NEXT: br label [[PRED_STORE_CONTINUE6]]
diff --git a/llvm/test/Transforms/LoopVectorize/bsd_regex.ll b/llvm/test/Transforms/LoopVectorize/bsd_regex.ll
index 784ccd22f98d7..d583dca5dad3a 100644
--- a/llvm/test/Transforms/LoopVectorize/bsd_regex.ll
+++ b/llvm/test/Transforms/LoopVectorize/bsd_regex.ll
@@ -24,10 +24,14 @@ define void @foo(ptr nocapture %A) {
; CHECK-NEXT: [[TMP1:%.*]] = add <2 x i64> [[STEP_ADD]], splat (i64 8)
; CHECK-NEXT: [[TMP8:%.*]] = extractelement <2 x i64> [[TMP1]], i64 0
; CHECK-NEXT: [[TMP11:%.*]] = extractelement <2 x i64> [[TMP1]], i64 1
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[TMP4]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP13]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[TMP6]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP8]], 2
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP16]]
; CHECK-NEXT: store i32 4, ptr [[TMP5]], align 4
; CHECK-NEXT: store i32 4, ptr [[TMP7]], align 4
; CHECK-NEXT: store i32 4, ptr [[TMP9]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll b/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll
index 93ac34003d4b7..15c00e2a6533f 100644
--- a/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll
+++ b/llvm/test/Transforms/LoopVectorize/extract-last-veclane.ll
@@ -11,10 +11,12 @@ define void @inv_store_last_lane(ptr noalias nocapture %a, ptr noalias nocapture
; 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 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = shl nsw <4 x i32> [[WIDE_LOAD]], splat (i32 1)
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP8]]
; CHECK-NEXT: store <4 x i32> [[TMP1]], ptr [[TMP2]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -28,10 +30,12 @@ define void @inv_store_last_lane(ptr noalias nocapture %a, ptr noalias nocapture
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP7]]
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL:%.*]] = shl nsw i32 [[TMP5]], 1
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
@@ -72,10 +76,12 @@ define float @ret_last_lane(ptr noalias nocapture %a, ptr noalias nocapture read
; 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 float, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = fmul <4 x float> [[WIDE_LOAD]], splat (float 2.000000e+00)
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP8]]
; CHECK-NEXT: store <4 x float> [[TMP1]], ptr [[TMP2]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -89,10 +95,12 @@ define float @ret_last_lane(ptr noalias nocapture %a, ptr noalias nocapture read
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP7]]
; CHECK-NEXT: [[TMP5:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL:%.*]] = fmul float [[TMP5]], 2.000000e+00
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: store float [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
diff --git a/llvm/test/Transforms/LoopVectorize/float-induction.ll b/llvm/test/Transforms/LoopVectorize/float-induction.ll
index f56699a45320d..c9b35ed89dd22 100644
--- a/llvm/test/Transforms/LoopVectorize/float-induction.ll
+++ b/llvm/test/Transforms/LoopVectorize/float-induction.ll
@@ -45,7 +45,8 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC4_INTERL1: vector.body:
; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; VEC4_INTERL1-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP4]], align 4
; VEC4_INTERL1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VEC4_INTERL1-NEXT: [[VEC_IND_NEXT]] = fsub fast <4 x float> [[VEC_IND]], [[DOTSPLAT5]]
@@ -61,7 +62,8 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC4_INTERL1: for.body:
; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL1-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP8]]
; VEC4_INTERL1-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL1-NEXT: [[ADD]] = fsub fast float [[X_05]], [[FPINC]]
; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -99,7 +101,8 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP8]]
; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw 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
@@ -117,7 +120,8 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC4_INTERL2: for.body:
; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL2-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP9]]
; VEC4_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL2-NEXT: [[ADD]] = fsub fast float [[X_05]], [[FPINC]]
; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -150,8 +154,10 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC1_INTERL2-NEXT: [[TMP3:%.*]] = fmul fast float [[FPINC]], [[DOTCAST2]]
; VEC1_INTERL2-NEXT: [[OFFSET_IDX:%.*]] = fsub fast float [[INIT]], [[TMP3]]
; VEC1_INTERL2-NEXT: [[TMP4:%.*]] = fsub fast float [[OFFSET_IDX]], [[FPINC]]
-; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC1_INTERL2-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
+; VEC1_INTERL2-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP10]]
; VEC1_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP9]], i64 4
; VEC1_INTERL2-NEXT: store float [[OFFSET_IDX]], ptr [[TMP8]], align 4
; VEC1_INTERL2-NEXT: store float [[TMP4]], ptr [[TMP6]], align 4
@@ -168,7 +174,8 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC1_INTERL2: for.body:
; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC1_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC1_INTERL2-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
; VEC1_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC1_INTERL2-NEXT: [[ADD]] = fsub fast float [[X_05]], [[FPINC]]
; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -207,7 +214,8 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC2_INTERL1_PRED_STORE: vector.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND:%.*]] = phi <2 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> [[VEC_IND]], ptr [[TMP4]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND_NEXT]] = fsub fast <2 x float> [[VEC_IND]], [[DOTSPLAT5]]
@@ -219,7 +227,8 @@ define void @fp_iv_loop1_fast_FMF(float %init, ptr noalias nocapture %A, i32 %N)
; VEC2_INTERL1_PRED_STORE: for.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_LR_PH]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[X_05:%.*]] = phi float [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[INIT]], [[FOR_BODY_LR_PH]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[ADD]] = fsub fast float [[X_05]], [[FPINC]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -290,7 +299,8 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC4_INTERL1: vector.body:
; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; VEC4_INTERL1-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP4]], align 4
; VEC4_INTERL1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VEC4_INTERL1-NEXT: [[VEC_IND_NEXT]] = fsub reassoc <4 x float> [[VEC_IND]], [[DOTSPLAT5]]
@@ -306,7 +316,8 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC4_INTERL1: for.body:
; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL1-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP8]]
; VEC4_INTERL1-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL1-NEXT: [[ADD]] = fsub reassoc float [[X_05]], [[FPINC]]
; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -344,7 +355,8 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP8]]
; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw 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
@@ -362,7 +374,8 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC4_INTERL2: for.body:
; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL2-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP9]]
; VEC4_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL2-NEXT: [[ADD]] = fsub reassoc float [[X_05]], [[FPINC]]
; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -395,8 +408,10 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC1_INTERL2-NEXT: [[TMP3:%.*]] = fmul reassoc float [[FPINC]], [[DOTCAST2]]
; VEC1_INTERL2-NEXT: [[OFFSET_IDX:%.*]] = fsub reassoc float [[INIT]], [[TMP3]]
; VEC1_INTERL2-NEXT: [[TMP6:%.*]] = fsub reassoc float [[OFFSET_IDX]], [[FPINC]]
-; VEC1_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP10:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC1_INTERL2-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
+; VEC1_INTERL2-NEXT: [[TMP12:%.*]] = shl i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP12]]
; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP10]], i64 4
; VEC1_INTERL2-NEXT: store float [[OFFSET_IDX]], ptr [[TMP7]], align 4
; VEC1_INTERL2-NEXT: store float [[TMP6]], ptr [[TMP8]], align 4
@@ -413,7 +428,8 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC1_INTERL2: for.body:
; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC1_INTERL2-NEXT: [[X_05:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC1_INTERL2-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
; VEC1_INTERL2-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC1_INTERL2-NEXT: [[ADD]] = fsub reassoc float [[X_05]], [[FPINC]]
; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -452,7 +468,8 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC2_INTERL1_PRED_STORE: vector.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND:%.*]] = phi <2 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> [[VEC_IND]], ptr [[TMP4]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND_NEXT]] = fsub reassoc <2 x float> [[VEC_IND]], [[DOTSPLAT5]]
@@ -464,7 +481,8 @@ define void @fp_iv_loop1_reassoc_FMF(float %init, ptr noalias nocapture %A, i32
; VEC2_INTERL1_PRED_STORE: for.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_LR_PH]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[X_05:%.*]] = phi float [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[INIT]], [[FOR_BODY_LR_PH]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[X_05]], ptr [[ARRAYIDX]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[ADD]] = fsub reassoc float [[X_05]], [[FPINC]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -530,7 +548,8 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC4_INTERL1: vector.body:
; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; VEC4_INTERL1-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP2]], align 4
; VEC4_INTERL1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VEC4_INTERL1-NEXT: [[VEC_IND_NEXT]] = fadd fast <4 x float> [[VEC_IND]], splat (float 2.000000e+00)
@@ -546,7 +565,8 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC4_INTERL1: for.body:
; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC4_INTERL1-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP6]]
; VEC4_INTERL1-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL1-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -579,7 +599,8 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], splat (float 2.000000e+00)
-; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw 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
@@ -597,7 +618,8 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC4_INTERL2: for.body:
; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC4_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP7]]
; VEC4_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -629,8 +651,10 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC1_INTERL2-NEXT: [[TMP3:%.*]] = fmul fast float [[DOTCAST2]], 5.000000e-01
; VEC1_INTERL2-NEXT: [[OFFSET_IDX:%.*]] = fadd fast float [[INIT]], [[TMP3]]
; VEC1_INTERL2-NEXT: [[TMP4:%.*]] = fadd fast float [[OFFSET_IDX]], 5.000000e-01
-; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC1_INTERL2-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
+; VEC1_INTERL2-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP10]]
; VEC1_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP9]], i64 4
; VEC1_INTERL2-NEXT: store float [[OFFSET_IDX]], ptr [[TMP8]], align 4
; VEC1_INTERL2-NEXT: store float [[TMP4]], ptr [[TMP6]], align 4
@@ -647,7 +671,8 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC1_INTERL2: for.body:
; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC1_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC1_INTERL2-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
; VEC1_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC1_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -679,7 +704,8 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC2_INTERL1_PRED_STORE: vector.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND:%.*]] = phi <2 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> [[VEC_IND]], ptr [[TMP2]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND_NEXT]] = fadd fast <2 x float> [[VEC_IND]], splat (float 1.000000e+00)
@@ -691,7 +717,8 @@ define void @fp_iv_loop2(float %init, ptr noalias nocapture %A, i32 %N) #0 {
; VEC2_INTERL1_PRED_STORE: for.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[INIT]], [[FOR_BODY_PREHEADER]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP6]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -774,14 +801,17 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ <float 0x3FB99999A0000000, float 0xBFD99999A0000000, float 0xBFECCCCCC0000000, float 0xBFF6666660000000>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[VEC_IND9:%.*]] = phi <4 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT10:%.*]], [[VECTOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP14]]
; VEC4_INTERL1-NEXT: store <4 x float> [[VEC_IND9]], ptr [[TMP6]], align 4
; VEC4_INTERL1-NEXT: [[TMP7:%.*]] = fadd fast <4 x float> [[VEC_IND9]], [[DOTSPLAT]]
; VEC4_INTERL1-NEXT: [[TMP8:%.*]] = fadd fast <4 x float> [[VEC_IND]], splat (float -5.000000e-01)
; VEC4_INTERL1-NEXT: [[TMP9:%.*]] = fadd fast <4 x float> [[TMP8]], [[TMP7]]
-; VEC4_INTERL1-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP16:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP16]]
; VEC4_INTERL1-NEXT: store <4 x float> [[TMP9]], ptr [[TMP10]], align 4
-; VEC4_INTERL1-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[C:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP15:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i64 [[TMP15]]
; VEC4_INTERL1-NEXT: store <4 x float> [[TMP8]], ptr [[TMP11]], align 4
; VEC4_INTERL1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VEC4_INTERL1-NEXT: [[VEC_IND_NEXT]] = fadd fast <4 x float> [[VEC_IND]], splat (float -2.000000e+00)
@@ -800,14 +830,17 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[CONV1:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL1-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL1-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP18]]
; VEC4_INTERL1-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL1-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]]
; VEC4_INTERL1-NEXT: [[CONV1]] = fadd fast float [[Y_012]], -5.000000e-01
; VEC4_INTERL1-NEXT: [[ADD2:%.*]] = fadd fast float [[CONV1]], [[ADD]]
-; VEC4_INTERL1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]]
+; VEC4_INTERL1-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP19]]
; VEC4_INTERL1-NEXT: store float [[ADD2]], ptr [[ARRAYIDX4]], align 4
-; VEC4_INTERL1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]]
+; VEC4_INTERL1-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP20]]
; VEC4_INTERL1-NEXT: store float [[CONV1]], ptr [[ARRAYIDX6]], align 4
; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; VEC4_INTERL1-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
@@ -850,7 +883,8 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC4_INTERL2-NEXT: [[VEC_IND:%.*]] = phi <4 x float> [ <float 0x3FB99999A0000000, float 0xBFD99999A0000000, float 0xBFECCCCCC0000000, float 0xBFF6666660000000>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], [[DOTSPLAT5]]
-; VEC4_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP21]]
; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw 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
@@ -860,11 +894,13 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC4_INTERL2-NEXT: [[TMP11:%.*]] = fadd fast <4 x float> [[VEC_IND]], splat (float -2.500000e+00)
; 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: [[TMP22:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP22]]
; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw 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: [[TMP20:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i64 [[TMP20]]
; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw 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
@@ -885,14 +921,17 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[CONV1:%.*]], [[FOR_BODY]] ]
; VEC4_INTERL2-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL2-NEXT: [[TMP24:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP24]]
; VEC4_INTERL2-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL2-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]]
; VEC4_INTERL2-NEXT: [[CONV1]] = fadd fast float [[Y_012]], -5.000000e-01
; VEC4_INTERL2-NEXT: [[ADD2:%.*]] = fadd fast float [[CONV1]], [[ADD]]
-; VEC4_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]]
+; VEC4_INTERL2-NEXT: [[TMP25:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP25]]
; VEC4_INTERL2-NEXT: store float [[ADD2]], ptr [[ARRAYIDX4]], align 4
-; VEC4_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]]
+; VEC4_INTERL2-NEXT: [[TMP26:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP26]]
; VEC4_INTERL2-NEXT: store float [[CONV1]], ptr [[ARRAYIDX6]], align 4
; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; VEC4_INTERL2-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
@@ -930,8 +969,10 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC1_INTERL2-NEXT: [[TMP6:%.*]] = fmul fast float [[TMP0]], [[DOTCAST6]]
; VEC1_INTERL2-NEXT: [[OFFSET_IDX7:%.*]] = fadd fast float [[INIT]], [[TMP6]]
; VEC1_INTERL2-NEXT: [[TMP7:%.*]] = fadd fast float [[OFFSET_IDX7]], [[TMP0]]
-; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP4]]
+; VEC1_INTERL2-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP21]]
+; VEC1_INTERL2-NEXT: [[TMP23:%.*]] = shl nsw i64 [[TMP4]], 2
+; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP23]]
; VEC1_INTERL2-NEXT: store float [[OFFSET_IDX7]], ptr [[TMP8]], align 4
; VEC1_INTERL2-NEXT: store float [[TMP7]], ptr [[TMP9]], align 4
; VEC1_INTERL2-NEXT: [[TMP10:%.*]] = fadd fast float [[OFFSET_IDX7]], [[TMP0]]
@@ -940,12 +981,16 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC1_INTERL2-NEXT: [[TMP13:%.*]] = fadd fast float [[TMP5]], 0xBFECCCCCC0000000
; VEC1_INTERL2-NEXT: [[TMP14:%.*]] = fadd fast float [[TMP12]], [[TMP10]]
; VEC1_INTERL2-NEXT: [[TMP15:%.*]] = fadd fast float [[TMP13]], [[TMP11]]
-; VEC1_INTERL2-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP4]]
+; VEC1_INTERL2-NEXT: [[TMP25:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP25]]
+; VEC1_INTERL2-NEXT: [[TMP22:%.*]] = shl nsw i64 [[TMP4]], 2
+; VEC1_INTERL2-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP22]]
; VEC1_INTERL2-NEXT: store float [[TMP14]], ptr [[TMP16]], align 4
; VEC1_INTERL2-NEXT: store float [[TMP15]], ptr [[TMP17]], align 4
-; VEC1_INTERL2-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[C:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP19:%.*]] = getelementptr inbounds float, ptr [[C]], i64 [[TMP4]]
+; VEC1_INTERL2-NEXT: [[TMP24:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i64 [[TMP24]]
+; VEC1_INTERL2-NEXT: [[TMP26:%.*]] = shl nsw i64 [[TMP4]], 2
+; VEC1_INTERL2-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[C]], i64 [[TMP26]]
; VEC1_INTERL2-NEXT: store float [[TMP12]], ptr [[TMP18]], align 4
; VEC1_INTERL2-NEXT: store float [[TMP13]], ptr [[TMP19]], align 4
; VEC1_INTERL2-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
@@ -963,14 +1008,17 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; VEC1_INTERL2-NEXT: [[Y_012:%.*]] = phi float [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[CONV1:%.*]], [[FOR_BODY]] ]
; VEC1_INTERL2-NEXT: [[X_011:%.*]] = phi float [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC1_INTERL2-NEXT: [[TMP29:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP29]]
; VEC1_INTERL2-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4
; VEC1_INTERL2-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]]
; VEC1_INTERL2-NEXT: [[CONV1]] = fadd fast float [[Y_012]], -5.000000e-01
; VEC1_INTERL2-NEXT: [[ADD2:%.*]] = fadd fast float [[CONV1]], [[ADD]]
-; VEC1_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]]
+; VEC1_INTERL2-NEXT: [[TMP30:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC1_INTERL2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP30]]
; VEC1_INTERL2-NEXT: store float [[ADD2]], ptr [[ARRAYIDX4]], align 4
-; VEC1_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]]
+; VEC1_INTERL2-NEXT: [[TMP31:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC1_INTERL2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP31]]
; VEC1_INTERL2-NEXT: store float [[CONV1]], ptr [[ARRAYIDX6]], align 4
; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; VEC1_INTERL2-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
@@ -1014,14 +1062,17 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND:%.*]] = phi <2 x float> [ <float 0x3FB99999A0000000, float 0xBFD99999A0000000>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND9:%.*]] = phi <2 x float> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT10:%.*]], [[VECTOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP14]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> [[VEC_IND9]], ptr [[TMP6]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP7:%.*]] = fadd fast <2 x float> [[VEC_IND9]], [[DOTSPLAT]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP8:%.*]] = fadd fast <2 x float> [[VEC_IND]], splat (float -5.000000e-01)
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP9:%.*]] = fadd fast <2 x float> [[TMP8]], [[TMP7]]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP16:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP16]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> [[TMP9]], ptr [[TMP10]], align 4
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[C:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP15:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[C:%.*]], i64 [[TMP15]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> [[TMP8]], ptr [[TMP11]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND_NEXT]] = fadd fast <2 x float> [[VEC_IND]], splat (float -1.000000e+00)
@@ -1035,14 +1086,17 @@ define void @fp_iv_loop3(float %init, ptr noalias nocapture %A, ptr noalias noca
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_LR_PH]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[Y_012:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ 0x3FB99999A0000000, [[FOR_BODY_LR_PH]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[X_011:%.*]] = phi float [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[IND_END3]], [[MIDDLE_BLOCK]] ], [ [[INIT]], [[FOR_BODY_LR_PH]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[X_011]], ptr [[ARRAYIDX]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[ADD]] = fadd fast float [[X_011]], [[TMP0]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[CONV1]] = fadd fast float [[Y_012]], -5.000000e-01
; VEC2_INTERL1_PRED_STORE-NEXT: [[ADD2:%.*]] = fadd fast float [[CONV1]], [[ADD]]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP19]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[ADD2]], ptr [[ARRAYIDX4]], align 4
-; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds float, ptr [[C]], i64 [[INDVARS_IV]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i8, ptr [[C]], i64 [[TMP20]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[CONV1]], ptr [[ARRAYIDX6]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; VEC2_INTERL1_PRED_STORE-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
@@ -1112,7 +1166,8 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC4_INTERL1: vector.body:
; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC4_INTERL1-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_INTERL1-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; VEC4_INTERL1-NEXT: store <4 x float> [[VEC_IND]], ptr [[TMP2]], align 4
; VEC4_INTERL1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VEC4_INTERL1-NEXT: [[VEC_IND_NEXT]] = fadd fast <4 x float> [[VEC_IND]], splat (float 2.000000e+00)
@@ -1128,7 +1183,8 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC4_INTERL1: for.body:
; VEC4_INTERL1-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC4_INTERL1-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP6]]
; VEC4_INTERL1-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL1-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC4_INTERL1-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -1158,7 +1214,8 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], splat (float 2.000000e+00)
-; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw 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
@@ -1176,7 +1233,8 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC4_INTERL2: for.body:
; VEC4_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC4_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC4_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP7]]
; VEC4_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC4_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC4_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -1208,8 +1266,10 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC1_INTERL2-NEXT: [[TMP3:%.*]] = fmul fast float [[DOTCAST2]], 5.000000e-01
; VEC1_INTERL2-NEXT: [[OFFSET_IDX:%.*]] = fadd fast float [[TMP3]], 1.000000e+00
; VEC1_INTERL2-NEXT: [[TMP4:%.*]] = fadd fast float [[TMP3]], 1.500000e+00
-; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC1_INTERL2-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
+; VEC1_INTERL2-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP10]]
; VEC1_INTERL2-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP9]], i64 4
; VEC1_INTERL2-NEXT: store float [[OFFSET_IDX]], ptr [[TMP8]], align 4
; VEC1_INTERL2-NEXT: store float [[TMP4]], ptr [[TMP6]], align 4
@@ -1226,7 +1286,8 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC1_INTERL2: for.body:
; VEC1_INTERL2-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC1_INTERL2-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC1_INTERL2-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC1_INTERL2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
; VEC1_INTERL2-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC1_INTERL2-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC1_INTERL2-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -1255,7 +1316,8 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC2_INTERL1_PRED_STORE: vector.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND:%.*]] = phi <2 x float> [ <float 1.000000e+00, float 1.500000e+00>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> [[VEC_IND]], ptr [[TMP2]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC2_INTERL1_PRED_STORE-NEXT: [[VEC_IND_NEXT]] = fadd fast <2 x float> [[VEC_IND]], splat (float 1.000000e+00)
@@ -1267,7 +1329,8 @@ define void @fp_iv_loop4(ptr noalias nocapture %A, i32 %N) {
; VEC2_INTERL1_PRED_STORE: for.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[X_06:%.*]] = phi float [ [[CONV1:%.*]], [[FOR_BODY]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ 1.000000e+00, [[FOR_BODY_PREHEADER]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP6]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[X_06]], ptr [[ARRAYIDX]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[CONV1]] = fadd fast float [[X_06]], 5.000000e-01
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -1316,20 +1379,23 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL1: vector.body:
; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE7:%.*]] ]
; VEC4_INTERL1-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float
-; VEC4_INTERL1-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP18]]
; VEC4_INTERL1-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP0]], align 4
; VEC4_INTERL1-NEXT: [[TMP1:%.*]] = fcmp fast oeq <4 x float> [[WIDE_LOAD]], zeroinitializer
; VEC4_INTERL1-NEXT: [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; VEC4_INTERL1-NEXT: br i1 [[TMP2]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; VEC4_INTERL1: pred.store.if:
-; VEC4_INTERL1-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; VEC4_INTERL1-NEXT: store float [[DOTCAST2]], ptr [[TMP3]], align 4
; VEC4_INTERL1-NEXT: br label [[PRED_STORE_CONTINUE]]
; VEC4_INTERL1: pred.store.continue:
; VEC4_INTERL1-NEXT: [[TMP4:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
; VEC4_INTERL1-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF2:%.*]], label [[PRED_STORE_CONTINUE3:%.*]]
; VEC4_INTERL1: pred.store.if2:
-; VEC4_INTERL1-NEXT: [[TMP5:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP20:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP20]]
; VEC4_INTERL1-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i64 4
; VEC4_INTERL1-NEXT: [[TMP7:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00
; VEC4_INTERL1-NEXT: store float [[TMP7]], ptr [[TMP6]], align 4
@@ -1338,7 +1404,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL1-NEXT: [[TMP8:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
; VEC4_INTERL1-NEXT: br i1 [[TMP8]], label [[PRED_STORE_IF4:%.*]], label [[PRED_STORE_CONTINUE5:%.*]]
; VEC4_INTERL1: pred.store.if4:
-; VEC4_INTERL1-NEXT: [[TMP9:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP21:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP21]]
; VEC4_INTERL1-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[TMP9]], i64 8
; VEC4_INTERL1-NEXT: [[TMP11:%.*]] = fadd fast float [[DOTCAST2]], 2.000000e+00
; VEC4_INTERL1-NEXT: store float [[TMP11]], ptr [[TMP10]], align 4
@@ -1347,7 +1414,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL1-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
; VEC4_INTERL1-NEXT: br i1 [[TMP12]], label [[PRED_STORE_IF6:%.*]], label [[PRED_STORE_CONTINUE7]]
; VEC4_INTERL1: pred.store.if6:
-; VEC4_INTERL1-NEXT: [[TMP13:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP17:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP17]]
; VEC4_INTERL1-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 12
; VEC4_INTERL1-NEXT: [[TMP15:%.*]] = fadd fast float [[DOTCAST2]], 3.000000e+00
; VEC4_INTERL1-NEXT: store float [[TMP15]], ptr [[TMP14]], align 4
@@ -1366,7 +1434,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL1: for.body:
; VEC4_INTERL1-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC4_INTERL1-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; VEC4_INTERL1-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[I]]
+; VEC4_INTERL1-NEXT: [[TMP22:%.*]] = shl nsw i64 [[I]], 2
+; VEC4_INTERL1-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP22]]
; VEC4_INTERL1-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC4_INTERL1-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC4_INTERL1-NEXT: br i1 [[VAR2]], label [[IF_PRED:%.*]], label [[FOR_INC]]
@@ -1393,7 +1462,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2: vector.body:
; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE16:%.*]] ]
; VEC4_INTERL2-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float
-; VEC4_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP0]]
; VEC4_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw 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
@@ -1402,14 +1472,16 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[TMP5:%.*]] = extractelement <4 x i1> [[TMP3]], i64 0
; VEC4_INTERL2-NEXT: br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; VEC4_INTERL2: pred.store.if:
-; VEC4_INTERL2-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP36:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP36]]
; VEC4_INTERL2-NEXT: store float [[DOTCAST2]], ptr [[TMP35]], align 4
; VEC4_INTERL2-NEXT: br label [[PRED_STORE_CONTINUE]]
; VEC4_INTERL2: pred.store.continue:
; VEC4_INTERL2-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP3]], i64 1
; VEC4_INTERL2-NEXT: br i1 [[TMP6]], label [[PRED_STORE_IF3:%.*]], label [[PRED_STORE_CONTINUE4:%.*]]
; VEC4_INTERL2: pred.store.if3:
-; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP37:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP37]]
; VEC4_INTERL2-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 4
; VEC4_INTERL2-NEXT: [[TMP9:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00
; VEC4_INTERL2-NEXT: store float [[TMP9]], ptr [[TMP8]], align 4
@@ -1418,7 +1490,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[TMP10:%.*]] = extractelement <4 x i1> [[TMP3]], i64 2
; VEC4_INTERL2-NEXT: br i1 [[TMP10]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6:%.*]]
; VEC4_INTERL2: pred.store.if5:
-; VEC4_INTERL2-NEXT: [[TMP11:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP38:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP38]]
; VEC4_INTERL2-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP11]], i64 8
; VEC4_INTERL2-NEXT: [[TMP13:%.*]] = fadd fast float [[DOTCAST2]], 2.000000e+00
; VEC4_INTERL2-NEXT: store float [[TMP13]], ptr [[TMP12]], align 4
@@ -1427,7 +1500,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[TMP14:%.*]] = extractelement <4 x i1> [[TMP3]], i64 3
; VEC4_INTERL2-NEXT: br i1 [[TMP14]], label [[PRED_STORE_IF7:%.*]], label [[PRED_STORE_CONTINUE8:%.*]]
; VEC4_INTERL2: pred.store.if7:
-; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP40:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP40]]
; VEC4_INTERL2-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[TMP15]], i64 12
; VEC4_INTERL2-NEXT: [[TMP17:%.*]] = fadd fast float [[DOTCAST2]], 3.000000e+00
; VEC4_INTERL2-NEXT: store float [[TMP17]], ptr [[TMP16]], align 4
@@ -1436,7 +1510,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP4]], i64 0
; VEC4_INTERL2-NEXT: br i1 [[TMP18]], label [[PRED_STORE_IF9:%.*]], label [[PRED_STORE_CONTINUE10:%.*]]
; VEC4_INTERL2: pred.store.if9:
-; VEC4_INTERL2-NEXT: [[TMP19:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP41:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP41]]
; VEC4_INTERL2-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i64 16
; VEC4_INTERL2-NEXT: [[TMP21:%.*]] = fadd fast float [[DOTCAST2]], 4.000000e+00
; VEC4_INTERL2-NEXT: store float [[TMP21]], ptr [[TMP20]], align 4
@@ -1445,7 +1520,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[TMP22:%.*]] = extractelement <4 x i1> [[TMP4]], i64 1
; VEC4_INTERL2-NEXT: br i1 [[TMP22]], label [[PRED_STORE_IF11:%.*]], label [[PRED_STORE_CONTINUE12:%.*]]
; VEC4_INTERL2: pred.store.if11:
-; VEC4_INTERL2-NEXT: [[TMP23:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP42:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP42]]
; VEC4_INTERL2-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[TMP23]], i64 20
; VEC4_INTERL2-NEXT: [[TMP25:%.*]] = fadd fast float [[DOTCAST2]], 5.000000e+00
; VEC4_INTERL2-NEXT: store float [[TMP25]], ptr [[TMP24]], align 4
@@ -1454,7 +1530,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[TMP26:%.*]] = extractelement <4 x i1> [[TMP4]], i64 2
; VEC4_INTERL2-NEXT: br i1 [[TMP26]], label [[PRED_STORE_IF13:%.*]], label [[PRED_STORE_CONTINUE14:%.*]]
; VEC4_INTERL2: pred.store.if13:
-; VEC4_INTERL2-NEXT: [[TMP27:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP43:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP27:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP43]]
; VEC4_INTERL2-NEXT: [[TMP28:%.*]] = getelementptr i8, ptr [[TMP27]], i64 24
; VEC4_INTERL2-NEXT: [[TMP29:%.*]] = fadd fast float [[DOTCAST2]], 6.000000e+00
; VEC4_INTERL2-NEXT: store float [[TMP29]], ptr [[TMP28]], align 4
@@ -1463,7 +1540,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2-NEXT: [[TMP30:%.*]] = extractelement <4 x i1> [[TMP4]], i64 3
; VEC4_INTERL2-NEXT: br i1 [[TMP30]], label [[PRED_STORE_IF15:%.*]], label [[PRED_STORE_CONTINUE16]]
; VEC4_INTERL2: pred.store.if15:
-; VEC4_INTERL2-NEXT: [[TMP31:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP39:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP31:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP39]]
; VEC4_INTERL2-NEXT: [[TMP32:%.*]] = getelementptr i8, ptr [[TMP31]], i64 28
; VEC4_INTERL2-NEXT: [[TMP33:%.*]] = fadd fast float [[DOTCAST2]], 7.000000e+00
; VEC4_INTERL2-NEXT: store float [[TMP33]], ptr [[TMP32]], align 4
@@ -1482,7 +1560,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC4_INTERL2: for.body:
; VEC4_INTERL2-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC4_INTERL2-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL17]], [[SCALAR_PH]] ]
-; VEC4_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[I]]
+; VEC4_INTERL2-NEXT: [[TMP44:%.*]] = shl nsw i64 [[I]], 2
+; VEC4_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP44]]
; VEC4_INTERL2-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC4_INTERL2-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC4_INTERL2-NEXT: br i1 [[VAR2]], label [[IF_PRED:%.*]], label [[FOR_INC]]
@@ -1509,8 +1588,10 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC1_INTERL2: vector.body:
; VEC1_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE3:%.*]] ]
; VEC1_INTERL2-NEXT: [[DOTCAST1:%.*]] = sitofp i64 [[INDEX]] to float
-; VEC1_INTERL2-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC1_INTERL2-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP9]]
+; VEC1_INTERL2-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP10]]
; VEC1_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP1]], i64 4
; VEC1_INTERL2-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP0]], align 4
; VEC1_INTERL2-NEXT: [[TMP4:%.*]] = load float, ptr [[TMP2]], align 4
@@ -1540,7 +1621,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC1_INTERL2: for.body:
; VEC1_INTERL2-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; VEC1_INTERL2-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ]
-; VEC1_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[I]]
+; VEC1_INTERL2-NEXT: [[TMP11:%.*]] = shl nsw i64 [[I]], 2
+; VEC1_INTERL2-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
; VEC1_INTERL2-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC1_INTERL2-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC1_INTERL2-NEXT: br i1 [[VAR2]], label [[IF_PRED:%.*]], label [[FOR_INC]]
@@ -1567,20 +1649,23 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC2_INTERL1_PRED_STORE: vector.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE3:%.*]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[DOTCAST2:%.*]] = sitofp i64 [[INDEX]] to float
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP9]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP0]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP1:%.*]] = fcmp fast oeq <2 x float> [[WIDE_LOAD]], zeroinitializer
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP2:%.*]] = extractelement <2 x i1> [[TMP1]], i64 0
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[TMP2]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; VEC2_INTERL1_PRED_STORE: pred.store.if:
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP10]]
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[DOTCAST2]], ptr [[TMP3]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: br label [[PRED_STORE_CONTINUE]]
; VEC2_INTERL1_PRED_STORE: pred.store.continue:
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP4:%.*]] = extractelement <2 x i1> [[TMP1]], i64 1
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[TMP4]], label [[PRED_STORE_IF2:%.*]], label [[PRED_STORE_CONTINUE3]]
; VEC2_INTERL1_PRED_STORE: pred.store.if2:
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP5:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP11:%.*]] = shl i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP11]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i64 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP7:%.*]] = fadd fast float [[DOTCAST2]], 1.000000e+00
; VEC2_INTERL1_PRED_STORE-NEXT: store float [[TMP7]], ptr [[TMP6]], align 4
@@ -1595,7 +1680,8 @@ define void @non_primary_iv_float_scalar(ptr %A, i64 %N) {
; VEC2_INTERL1_PRED_STORE: for.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
; VEC2_INTERL1_PRED_STORE-NEXT: [[J:%.*]] = phi float [ [[J_NEXT:%.*]], [[FOR_INC]] ], [ [[DOTCAST]], [[MIDDLE_BLOCK]] ], [ 0.000000e+00, [[ENTRY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR0:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP12:%.*]] = shl nsw i64 [[I]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR0:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP12]]
; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR1:%.*]] = load float, ptr [[VAR0]], align 4
; VEC2_INTERL1_PRED_STORE-NEXT: [[VAR2:%.*]] = fcmp fast oeq float [[VAR1]], 0.000000e+00
; VEC2_INTERL1_PRED_STORE-NEXT: br i1 [[VAR2]], label [[IF_PRED:%.*]], label [[FOR_INC]]
@@ -1643,7 +1729,8 @@ define i32 @float_induction_with_dbg_on_fadd(ptr %dst) {
; VEC4_INTERL1-NEXT: br label [[VECTOR_BODY:%.*]]
; VEC4_INTERL1: vector.body:
; VEC4_INTERL1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC4_INTERL1-NEXT: [[TMP0:%.*]] = getelementptr float, ptr null, i64 [[INDEX]]
+; VEC4_INTERL1-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL1-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr null, i64 [[TMP2]]
; VEC4_INTERL1-NEXT: store <4 x float> poison, ptr [[TMP0]], align 8
; VEC4_INTERL1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; VEC4_INTERL1-NEXT: [[TMP1:%.*]] = icmp eq i64 [[INDEX_NEXT]], 200
@@ -1660,7 +1747,8 @@ define i32 @float_induction_with_dbg_on_fadd(ptr %dst) {
; VEC4_INTERL2-NEXT: br label [[VECTOR_BODY:%.*]]
; VEC4_INTERL2: vector.body:
; VEC4_INTERL2-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC4_INTERL2-NEXT: [[TMP0:%.*]] = getelementptr float, ptr null, i64 [[INDEX]]
+; VEC4_INTERL2-NEXT: [[TMP3:%.*]] = shl i64 [[INDEX]], 2
+; VEC4_INTERL2-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr null, i64 [[TMP3]]
; VEC4_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i64 16
; VEC4_INTERL2-NEXT: store <4 x float> poison, ptr [[TMP0]], align 8
; VEC4_INTERL2-NEXT: store <4 x float> zeroinitializer, ptr [[TMP1]], align 8
@@ -1682,8 +1770,10 @@ define i32 @float_induction_with_dbg_on_fadd(ptr %dst) {
; VEC1_INTERL2-NEXT: [[DOTCAST:%.*]] = sitofp i64 [[INDEX]] to float
; VEC1_INTERL2-NEXT: [[TMP0:%.*]] = call reassoc float @llvm.copysign.f32(float 0.000000e+00, float [[DOTCAST]])
; VEC1_INTERL2-NEXT: [[OFFSET_IDX:%.*]] = fadd reassoc float [[TMP0]], 0.000000e+00
-; VEC1_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr float, ptr null, i64 [[INDEX]]
-; VEC1_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr float, ptr null, i64 [[INDEX]]
+; VEC1_INTERL2-NEXT: [[TMP5:%.*]] = shl i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr null, i64 [[TMP5]]
+; VEC1_INTERL2-NEXT: [[TMP6:%.*]] = shl i64 [[INDEX]], 2
+; VEC1_INTERL2-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr null, i64 [[TMP6]]
; VEC1_INTERL2-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[TMP2]], i64 4
; VEC1_INTERL2-NEXT: store float poison, ptr [[TMP1]], align 8
; VEC1_INTERL2-NEXT: store float [[OFFSET_IDX]], ptr [[TMP4]], align 8
@@ -1700,7 +1790,8 @@ define i32 @float_induction_with_dbg_on_fadd(ptr %dst) {
; VEC2_INTERL1_PRED_STORE-NEXT: br label [[VECTOR_BODY:%.*]]
; VEC2_INTERL1_PRED_STORE: vector.body:
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP0:%.*]] = getelementptr float, ptr null, i64 [[INDEX]]
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 2
+; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr null, i64 [[TMP2]]
; VEC2_INTERL1_PRED_STORE-NEXT: store <2 x float> poison, ptr [[TMP0]], align 8
; VEC2_INTERL1_PRED_STORE-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; VEC2_INTERL1_PRED_STORE-NEXT: [[TMP1:%.*]] = icmp eq i64 [[INDEX_NEXT]], 200
diff --git a/llvm/test/Transforms/LoopVectorize/forked-pointers.ll b/llvm/test/Transforms/LoopVectorize/forked-pointers.ll
index 226379243c4d9..9d08882da76ef 100644
--- a/llvm/test/Transforms/LoopVectorize/forked-pointers.ll
+++ b/llvm/test/Transforms/LoopVectorize/forked-pointers.ll
@@ -42,7 +42,8 @@ define dso_local void @forked_ptrs_different_base_same_offset(ptr nocapture read
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[PREDS]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[PREDS]], i64 [[TMP3]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP8:%.*]] = select <4 x i1> [[TMP7]], <4 x ptr> [[BROADCAST_SPLAT]], <4 x ptr> [[BROADCAST_SPLAT9]]
@@ -50,12 +51,16 @@ define dso_local void @forked_ptrs_different_base_same_offset(ptr nocapture read
; CHECK-NEXT: [[TMP11:%.*]] = extractelement <4 x ptr> [[TMP8]], i64 1
; CHECK-NEXT: [[TMP13:%.*]] = extractelement <4 x ptr> [[TMP8]], i64 2
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x ptr> [[TMP8]], i64 3
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[TMP9]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP29:%.*]] = getelementptr float, ptr [[TMP11]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP32:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP9]], i64 [[TMP32]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP29:%.*]] = getelementptr i8, ptr [[TMP11]], i64 [[TMP34]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP29]], i64 4
-; CHECK-NEXT: [[TMP30:%.*]] = getelementptr float, ptr [[TMP13]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP30:%.*]] = getelementptr i8, ptr [[TMP13]], i64 [[TMP37]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP30]], i64 8
-; CHECK-NEXT: [[TMP31:%.*]] = getelementptr float, ptr [[TMP15]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP38:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP31:%.*]] = getelementptr i8, ptr [[TMP15]], i64 [[TMP38]]
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[TMP31]], i64 12
; CHECK-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP10]], align 4
; CHECK-NEXT: [[TMP18:%.*]] = load float, ptr [[TMP12]], align 4
@@ -65,7 +70,8 @@ define dso_local void @forked_ptrs_different_base_same_offset(ptr nocapture read
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x float> [[TMP21]], float [[TMP18]], i64 1
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x float> [[TMP22]], float [[TMP19]], i64 2
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x float> [[TMP23]], float [[TMP20]], i64 3
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds float, ptr [[DEST]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP39:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[DEST]], i64 [[TMP39]]
; CHECK-NEXT: store <4 x float> [[TMP24]], ptr [[TMP25]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP26:%.*]] = icmp eq i64 [[INDEX_NEXT]], 100
@@ -78,13 +84,16 @@ define dso_local void @forked_ptrs_different_base_same_offset(ptr nocapture read
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[PREDS]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[PREDS]], i64 [[TMP33]]
; CHECK-NEXT: [[TMP27:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CMP1_NOT:%.*]] = icmp eq i32 [[TMP27]], 0
; CHECK-NEXT: [[SPEC_SELECT:%.*]] = select i1 [[CMP1_NOT]], ptr [[BASE2]], ptr [[BASE1]]
-; CHECK-NEXT: [[DOTSINK_IN:%.*]] = getelementptr inbounds nuw float, ptr [[SPEC_SELECT]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[DOTSINK_IN:%.*]] = getelementptr inbounds nuw i8, ptr [[SPEC_SELECT]], i64 [[TMP35]]
; CHECK-NEXT: [[DOTSINK:%.*]] = load float, ptr [[DOTSINK_IN]], align 4
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw float, ptr [[DEST]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw i8, ptr [[DEST]], i64 [[TMP36]]
; CHECK-NEXT: store float [[DOTSINK]], ptr [[TMP28]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 100
@@ -132,7 +141,8 @@ define dso_local void @forked_ptrs_same_base_different_offset(ptr nocapture read
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[I_014:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[PREDS:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[PREDS:%.*]], i64 [[TMP4]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CMP1_NOT:%.*]] = icmp eq i32 [[TMP0]], 0
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -140,9 +150,11 @@ define dso_local void @forked_ptrs_same_base_different_offset(ptr nocapture read
; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[INDVARS_IV]] to i32
; CHECK-NEXT: [[OFFSET_0:%.*]] = select i1 [[CMP1_NOT]], i32 [[ADD]], i32 [[TMP1]]
; CHECK-NEXT: [[IDXPROM213:%.*]] = zext i32 [[OFFSET_0]] to i64
-; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw float, ptr [[BASE:%.*]], i64 [[IDXPROM213]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[IDXPROM213]], 2
+; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw i8, ptr [[BASE:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[ARRAYIDX3]], align 4
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw float, ptr [[DEST:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i8, ptr [[DEST:%.*]], i64 [[TMP5]]
; CHECK-NEXT: store float [[TMP2]], ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 100
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]]
diff --git a/llvm/test/Transforms/LoopVectorize/histograms.ll b/llvm/test/Transforms/LoopVectorize/histograms.ll
index 5bb8722d734f4..db33341531e8e 100644
--- a/llvm/test/Transforms/LoopVectorize/histograms.ll
+++ b/llvm/test/Transforms/LoopVectorize/histograms.ll
@@ -12,13 +12,16 @@ define void @simple_histogram(ptr noalias %buckets, ptr readonly %indices, i64 %
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = zext <2 x i32> [[WIDE_LOAD]] to <2 x i64>
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i64 0
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <2 x i64> [[TMP1]], i64 1
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i32, ptr [[BUCKETS]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i32, ptr [[BUCKETS]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw i64 [[TMP2]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[TMP4]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP11]]
; CHECK-NEXT: [[TMP6:%.*]] = insertelement <2 x ptr> poison, ptr [[TMP3]], i64 0
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <2 x ptr> [[TMP6]], ptr [[TMP5]], i64 1
; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.v2p0.i32(<2 x ptr> [[TMP7]], i32 1, <2 x i1> splat (i1 true))
@@ -33,10 +36,12 @@ define void @simple_histogram(ptr noalias %buckets, ptr readonly %indices, i64 %
; CHECK-NEXT: br label [[FOR_BODY1:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV1:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT1:%.*]], [[FOR_BODY1]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[IV1]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[IV1]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[INDICES]], i64 [[TMP14]]
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[IDXPROM1:%.*]] = zext i32 [[TMP12]] to i64
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[BUCKETS]], i64 [[IDXPROM1]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[IDXPROM1]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[BUCKETS]], i64 [[TMP15]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP13]], 1
; CHECK-NEXT: store i32 [[INC]], ptr [[ARRAYIDX2]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/if-conversion-nest.ll b/llvm/test/Transforms/LoopVectorize/if-conversion-nest.ll
index 51255b2e35707..387abd90ff6d8 100644
--- a/llvm/test/Transforms/LoopVectorize/if-conversion-nest.ll
+++ b/llvm/test/Transforms/LoopVectorize/if-conversion-nest.ll
@@ -28,17 +28,19 @@ define void @foo(ptr nocapture %A, ptr nocapture %B, i32 %n) {
; 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 [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP5]], align 4, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP6]], align 4, !alias.scope [[META3]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP5]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP6]], align 4, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP15]]
+; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP17]], align 4, !alias.scope [[META3]]
; CHECK-NEXT: [[TMP7:%.*]] = icmp sgt <4 x i32> [[WIDE_LOAD]], [[WIDE_LOAD2]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp sgt <4 x i32> [[WIDE_LOAD]], splat (i32 19)
; CHECK-NEXT: [[TMP9:%.*]] = icmp slt <4 x i32> [[WIDE_LOAD2]], splat (i32 4)
; CHECK-NEXT: [[TMP10:%.*]] = select <4 x i1> [[TMP9]], <4 x i32> splat (i32 4), <4 x i32> splat (i32 5)
; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP8]], <4 x i32> splat (i32 3), <4 x i32> [[TMP10]]
; CHECK-NEXT: [[PREDPHI3:%.*]] = select <4 x i1> [[TMP7]], <4 x i32> [[PREDPHI]], <4 x i32> splat (i32 9)
-; CHECK-NEXT: store <4 x i32> [[PREDPHI3]], ptr [[TMP5]], align 4, !alias.scope [[META0]], !noalias [[META3]]
+; CHECK-NEXT: store <4 x i32> [[PREDPHI3]], ptr [[TMP6]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
@@ -50,9 +52,11 @@ define void @foo(ptr nocapture %A, ptr nocapture %B, i32 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[IF_END14:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP14]]
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP16]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[CMP3:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]]
; CHECK-NEXT: br i1 [[CMP3]], label [[IF_THEN:%.*]], label [[IF_END14]]
@@ -135,10 +139,12 @@ define void @multi_variable_if_nest(ptr nocapture %A, ptr nocapture %B, i32 %n)
; 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 [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP5]], align 4, !alias.scope [[META9:![0-9]+]], !noalias [[META12:![0-9]+]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP6]], align 4, !alias.scope [[META12]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP5]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP6]], align 4, !alias.scope [[META9:![0-9]+]], !noalias [[META12:![0-9]+]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP9]]
+; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4, !alias.scope [[META12]]
; CHECK-NEXT: [[TMP7:%.*]] = icmp sgt <4 x i32> [[WIDE_LOAD]], [[WIDE_LOAD2]]
; CHECK-NEXT: [[TMP8:%.*]] = icmp sgt <4 x i32> [[WIDE_LOAD]], splat (i32 19)
; CHECK-NEXT: [[TMP11:%.*]] = icmp slt <4 x i32> [[WIDE_LOAD2]], splat (i32 4)
@@ -149,8 +155,8 @@ define void @multi_variable_if_nest(ptr nocapture %A, ptr nocapture %B, i32 %n)
; CHECK-NEXT: [[PREDPHI3:%.*]] = select <4 x i1> [[TMP7]], <4 x i32> [[PREDPHI]], <4 x i32> splat (i32 9)
; CHECK-NEXT: [[PREDPHI4:%.*]] = select <4 x i1> [[TMP14]], <4 x i32> splat (i32 7), <4 x i32> [[TMP13]]
; CHECK-NEXT: [[PREDPHI5:%.*]] = select <4 x i1> [[TMP7]], <4 x i32> [[PREDPHI4]], <4 x i32> splat (i32 18)
-; CHECK-NEXT: store <4 x i32> [[PREDPHI3]], ptr [[TMP5]], align 4, !alias.scope [[META9]], !noalias [[META12]]
-; CHECK-NEXT: store <4 x i32> [[PREDPHI5]], ptr [[TMP6]], align 4, !alias.scope [[META12]]
+; CHECK-NEXT: store <4 x i32> [[PREDPHI3]], ptr [[TMP6]], align 4, !alias.scope [[META9]], !noalias [[META12]]
+; CHECK-NEXT: store <4 x i32> [[PREDPHI5]], ptr [[TMP10]], align 4, !alias.scope [[META12]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
@@ -162,9 +168,11 @@ define void @multi_variable_if_nest(ptr nocapture %A, ptr nocapture %B, i32 %n)
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[IF_END14:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP18]]
; CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[CMP3:%.*]] = icmp sgt i32 [[TMP16]], [[TMP17]]
; CHECK-NEXT: br i1 [[CMP3]], label [[IF_THEN:%.*]], label [[IF_END14]]
diff --git a/llvm/test/Transforms/LoopVectorize/if-conversion.ll b/llvm/test/Transforms/LoopVectorize/if-conversion.ll
index 350c267445e54..7ac2a309ad27a 100644
--- a/llvm/test/Transforms/LoopVectorize/if-conversion.ll
+++ b/llvm/test/Transforms/LoopVectorize/if-conversion.ll
@@ -54,15 +54,17 @@ define void @function0(ptr nocapture %a, ptr nocapture %b, i32 %start, i32 %end)
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[INDEX]], [[TMP0]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x i32>, ptr [[TMP14]], align 4, !alias.scope [[META3]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[OFFSET_IDX]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP13]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP14]], align 4, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[OFFSET_IDX]], 2
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP22]]
+; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x i32>, ptr [[TMP24]], align 4, !alias.scope [[META3]]
; CHECK-NEXT: [[DOTNOT:%.*]] = icmp sgt <4 x i32> [[WIDE_LOAD]], [[WIDE_LOAD4]]
; CHECK-NEXT: [[TMP15:%.*]] = mul <4 x i32> [[WIDE_LOAD]], splat (i32 5)
; CHECK-NEXT: [[TMP16:%.*]] = add <4 x i32> [[TMP15]], splat (i32 3)
; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[DOTNOT]], <4 x i32> [[TMP16]], <4 x i32> [[WIDE_LOAD]]
-; CHECK-NEXT: store <4 x i32> [[PREDPHI]], ptr [[TMP13]], align 4, !alias.scope [[META0]], !noalias [[META3]]
+; CHECK-NEXT: store <4 x i32> [[PREDPHI]], ptr [[TMP14]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; 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]+]]
@@ -74,9 +76,11 @@ define void @function0(ptr nocapture %a, ptr nocapture %b, i32 %start, i32 %end)
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[IF_END:.*]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP23]]
; CHECK-NEXT: [[TMP19:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
; CHECK-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP18]], [[TMP19]]
; CHECK-NEXT: br i1 [[CMP5]], label %[[IF_THEN:.*]], label %[[IF_END]]
@@ -157,7 +161,8 @@ define i32 @reduction_func(ptr nocapture %A, i32 %n) nounwind uwtable readonly s
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, %[[VECTOR_PH]] ], [ [[PREDPHI:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp sgt <4 x i32> [[WIDE_LOAD]], splat (i32 30)
; CHECK-NEXT: [[TMP3:%.*]] = add <4 x i32> [[VEC_PHI]], splat (i32 2)
@@ -177,7 +182,8 @@ define i32 @reduction_func(ptr nocapture %A, i32 %n) nounwind uwtable readonly s
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_INC:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
; CHECK-NEXT: [[SUM_011:%.*]] = phi i32 [ [[SUM_1:%.*]], %[[FOR_INC]] ], [ [[BC_MERGE_RDX]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i32 [[TMP7]], 30
; CHECK-NEXT: br i1 [[CMP1]], label %[[IF_THEN:.*]], label %[[FOR_INC]]
diff --git a/llvm/test/Transforms/LoopVectorize/induction.ll b/llvm/test/Transforms/LoopVectorize/induction.ll
index c1b912fd54b7d..74dff1c852749 100644
--- a/llvm/test/Transforms/LoopVectorize/induction.ll
+++ b/llvm/test/Transforms/LoopVectorize/induction.ll
@@ -67,7 +67,8 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 190, i32 191>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; IND-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; IND-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2
+; IND-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP3]], align 4
; IND-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <2 x i32> [[VEC_IND]], splat (i32 2)
@@ -83,7 +84,8 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; IND: for.body:
; IND-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; IND-NEXT: [[COUNT_09:%.*]] = phi i32 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
-; IND-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; IND-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; IND-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP7]]
; IND-NEXT: store i32 [[COUNT_09]], ptr [[ARRAYIDX2]], align 4
; IND-NEXT: [[INC]] = add nsw i32 [[COUNT_09]], 1
; IND-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -109,7 +111,8 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], splat (i32 2)
-; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; UNROLL-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw 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
@@ -127,7 +130,8 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; UNROLL: for.body:
; UNROLL-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; UNROLL-NEXT: [[COUNT_09:%.*]] = phi i32 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; UNROLL-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; UNROLL-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; UNROLL-NEXT: store i32 [[COUNT_09]], ptr [[ARRAYIDX2]], align 4
; UNROLL-NEXT: [[INC]] = add nsw i32 [[COUNT_09]], 1
; UNROLL-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -198,7 +202,8 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], splat (i32 4)
-; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; INTERLEAVE-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw 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
@@ -216,7 +221,8 @@ define void @multi_int_induction(ptr %A, i32 %N) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; INTERLEAVE-NEXT: [[COUNT_09:%.*]] = phi i32 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
-; INTERLEAVE-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; INTERLEAVE-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; INTERLEAVE-NEXT: store i32 [[COUNT_09]], ptr [[ARRAYIDX2]], align 4
; INTERLEAVE-NEXT: [[INC]] = add nsw i32 [[COUNT_09]], 1
; INTERLEAVE-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -339,15 +345,17 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; IND-NEXT: [[TMP5:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
-; IND-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[TMP5]], i64 [[OFFSET]]
-; IND-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP6]], align 4, !alias.scope [[META4:![0-9]+]], !noalias [[META7:![0-9]+]]
-; IND-NEXT: [[TMP7:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
-; IND-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP7]], i64 [[OFFSET2]]
-; IND-NEXT: [[WIDE_LOAD4:%.*]] = load <2 x float>, ptr [[TMP8]], align 4, !alias.scope [[META7]]
+; IND-NEXT: [[TMP5:%.*]] = add i64 [[INDEX]], [[OFFSET]]
+; IND-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP5]], 2
+; IND-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP6]]
+; IND-NEXT: [[WIDE_LOAD:%.*]] = load <2 x float>, ptr [[TMP7]], align 4, !alias.scope [[META4:![0-9]+]], !noalias [[META7:![0-9]+]]
+; IND-NEXT: [[TMP8:%.*]] = add i64 [[INDEX]], [[OFFSET2]]
+; IND-NEXT: [[TMP12:%.*]] = shl nsw i64 [[TMP8]], 2
+; IND-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP12]]
+; IND-NEXT: [[WIDE_LOAD4:%.*]] = load <2 x float>, ptr [[TMP13]], align 4, !alias.scope [[META7]]
; IND-NEXT: [[TMP9:%.*]] = fmul fast <2 x float> [[BROADCAST_SPLAT]], [[WIDE_LOAD4]]
; IND-NEXT: [[TMP10:%.*]] = fadd fast <2 x float> [[WIDE_LOAD]], [[TMP9]]
-; IND-NEXT: store <2 x float> [[TMP10]], ptr [[TMP6]], align 4, !alias.scope [[META4]], !noalias [[META7]]
+; IND-NEXT: store <2 x float> [[TMP10]], ptr [[TMP7]], align 4, !alias.scope [[META4]], !noalias [[META7]]
; IND-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; IND-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; IND-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
@@ -359,11 +367,13 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; IND-NEXT: br label [[FOR_BODY:%.*]]
; IND: for.body:
; IND-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; IND-NEXT: [[TMP12:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; IND-NEXT: [[ARR_IDX:%.*]] = getelementptr float, ptr [[TMP12]], i64 [[OFFSET]]
+; IND-NEXT: [[IND_SUM:%.*]] = add i64 [[IV]], [[OFFSET]]
+; IND-NEXT: [[TMP14:%.*]] = shl nsw i64 [[IND_SUM]], 2
+; IND-NEXT: [[ARR_IDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP14]]
; IND-NEXT: [[L1:%.*]] = load float, ptr [[ARR_IDX]], align 4
-; IND-NEXT: [[TMP13:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; IND-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP13]], i64 [[OFFSET2]]
+; IND-NEXT: [[IND_SUM2:%.*]] = add i64 [[IV]], [[OFFSET2]]
+; IND-NEXT: [[TMP15:%.*]] = shl nsw i64 [[IND_SUM2]], 2
+; IND-NEXT: [[ARR_IDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP15]]
; IND-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4
; IND-NEXT: [[M:%.*]] = fmul fast float [[B]], [[L2]]
; IND-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]]
@@ -399,13 +409,15 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; 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: [[TMP5:%.*]] = add i64 [[INDEX]], [[OFFSET]]
+; UNROLL-NEXT: [[TMP8:%.*]] = shl nsw i64 [[TMP5]], 2
+; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; UNROLL-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw 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: [[TMP16:%.*]] = add i64 [[INDEX]], [[OFFSET2]]
+; UNROLL-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP16]], 2
+; UNROLL-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP17]]
; UNROLL-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw 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]]
@@ -426,11 +438,13 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; UNROLL-NEXT: br label [[FOR_BODY:%.*]]
; UNROLL: for.body:
; UNROLL-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT: [[TMP16:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; UNROLL-NEXT: [[ARR_IDX:%.*]] = getelementptr float, ptr [[TMP16]], i64 [[OFFSET]]
+; UNROLL-NEXT: [[IND_SUM:%.*]] = add i64 [[IV]], [[OFFSET]]
+; UNROLL-NEXT: [[TMP18:%.*]] = shl nsw i64 [[IND_SUM]], 2
+; UNROLL-NEXT: [[ARR_IDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; UNROLL-NEXT: [[L1:%.*]] = load float, ptr [[ARR_IDX]], align 4
-; UNROLL-NEXT: [[TMP17:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; UNROLL-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP17]], i64 [[OFFSET2]]
+; UNROLL-NEXT: [[IND_SUM2:%.*]] = add i64 [[IV]], [[OFFSET2]]
+; UNROLL-NEXT: [[TMP19:%.*]] = shl nsw i64 [[IND_SUM2]], 2
+; UNROLL-NEXT: [[ARR_IDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; UNROLL-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4
; UNROLL-NEXT: [[M:%.*]] = fmul fast float [[B]], [[L2]]
; UNROLL-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]]
@@ -534,13 +548,15 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; INTERLEAVE-NEXT: br label [[VECTOR_BODY:%.*]]
; INTERLEAVE: vector.body:
; 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: [[TMP5:%.*]] = add i64 [[INDEX]], [[OFFSET]]
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = shl nsw i64 [[TMP5]], 2
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw 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: [[TMP16:%.*]] = add i64 [[INDEX]], [[OFFSET2]]
+; INTERLEAVE-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP16]], 2
+; INTERLEAVE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP17]]
; INTERLEAVE-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw 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]]
@@ -561,11 +577,13 @@ define void @scalar_use(ptr %a, float %b, i64 %offset, i64 %offset2, i64 %n) {
; INTERLEAVE-NEXT: br label [[FOR_BODY:%.*]]
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; INTERLEAVE-NEXT: [[TMP16:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; INTERLEAVE-NEXT: [[ARR_IDX:%.*]] = getelementptr float, ptr [[TMP16]], i64 [[OFFSET]]
+; INTERLEAVE-NEXT: [[IND_SUM:%.*]] = add i64 [[IV]], [[OFFSET]]
+; INTERLEAVE-NEXT: [[TMP18:%.*]] = shl nsw i64 [[IND_SUM]], 2
+; INTERLEAVE-NEXT: [[ARR_IDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; INTERLEAVE-NEXT: [[L1:%.*]] = load float, ptr [[ARR_IDX]], align 4
-; INTERLEAVE-NEXT: [[TMP17:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; INTERLEAVE-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP17]], i64 [[OFFSET2]]
+; INTERLEAVE-NEXT: [[IND_SUM2:%.*]] = add i64 [[IV]], [[OFFSET2]]
+; INTERLEAVE-NEXT: [[TMP19:%.*]] = shl nsw i64 [[IND_SUM2]], 2
+; INTERLEAVE-NEXT: [[ARR_IDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; INTERLEAVE-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4
; INTERLEAVE-NEXT: [[M:%.*]] = fmul fast float [[B]], [[L2]]
; INTERLEAVE-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]]
@@ -659,7 +677,8 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[VEC_PHI:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; IND-NEXT: [[TMP0:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[INDEX]]
+; IND-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 3
+; IND-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP8]]
; IND-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP0]], align 8
; IND-NEXT: [[TMP1]] = add <2 x i64> [[WIDE_LOAD]], [[VEC_PHI]]
; IND-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
@@ -676,7 +695,8 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[SUM:%.*]] = phi i64 [ [[TMP6:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; IND-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i64, ptr [[A]], i64 [[I]]
+; IND-NEXT: [[TMP9:%.*]] = shl nsw i64 [[I]], 3
+; IND-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP9]]
; IND-NEXT: [[TMP5:%.*]] = load i64, ptr [[TMP4]], align 8
; IND-NEXT: [[TMP6]] = add i64 [[TMP5]], [[SUM]]
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -698,7 +718,8 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 3
+; UNROLL-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP10]]
; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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
@@ -719,7 +740,8 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[SUM:%.*]] = phi i64 [ [[TMP8:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i64, ptr [[A]], i64 [[I]]
+; UNROLL-NEXT: [[TMP11:%.*]] = shl nsw i64 [[I]], 3
+; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
; UNROLL-NEXT: [[TMP7:%.*]] = load i64, ptr [[TMP6]], align 8
; UNROLL-NEXT: [[TMP8]] = add i64 [[TMP7]], [[SUM]]
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -785,7 +807,8 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP10]]
; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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
@@ -806,7 +829,8 @@ define i64 @scalarize_induction_variable_01(ptr %a, i64 %n) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[SUM:%.*]] = phi i64 [ [[TMP8:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i64, ptr [[A]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP11:%.*]] = shl nsw i64 [[I]], 3
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
; INTERLEAVE-NEXT: [[TMP7:%.*]] = load i64, ptr [[TMP6]], align 8
; INTERLEAVE-NEXT: [[TMP8]] = add i64 [[TMP7]], [[SUM]]
; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -925,14 +949,18 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; IND-NEXT: [[VEC_PHI:%.*]] = phi <2 x float> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
; IND-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 8
-; IND-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; IND-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP3]]
+; IND-NEXT: [[TMP28:%.*]] = shl i64 [[INDEX]], 5
+; IND-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP28]]
+; IND-NEXT: [[TMP30:%.*]] = shl nsw i64 [[TMP3]], 2
+; IND-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP30]]
; IND-NEXT: [[TMP6:%.*]] = load float, ptr [[TMP4]], align 4
; IND-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP5]], align 4
; IND-NEXT: [[TMP8:%.*]] = insertelement <2 x float> poison, float [[TMP6]], i64 0
; IND-NEXT: [[TMP9:%.*]] = insertelement <2 x float> [[TMP8]], float [[TMP7]], i64 1
-; IND-NEXT: [[TMP10:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP3]]
+; IND-NEXT: [[TMP31:%.*]] = shl i64 [[INDEX]], 5
+; IND-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP31]]
+; IND-NEXT: [[TMP32:%.*]] = shl nsw i64 [[TMP3]], 2
+; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP32]]
; IND-NEXT: [[TMP12:%.*]] = load float, ptr [[TMP10]], align 4
; IND-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP11]], align 4
; IND-NEXT: [[TMP14:%.*]] = insertelement <2 x float> poison, float [[TMP12]], i64 0
@@ -954,9 +982,11 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[FOR_BODY]] ]
; IND-NEXT: [[S:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[TMP27:%.*]], [[FOR_BODY]] ]
-; IND-NEXT: [[TMP21:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I]]
+; IND-NEXT: [[TMP33:%.*]] = shl nsw i64 [[I]], 2
+; IND-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP33]]
; IND-NEXT: [[TMP22:%.*]] = load float, ptr [[TMP21]], align 4
-; IND-NEXT: [[TMP23:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[I]]
+; IND-NEXT: [[TMP29:%.*]] = shl nsw i64 [[I]], 2
+; IND-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP29]]
; IND-NEXT: [[TMP24:%.*]] = load float, ptr [[TMP23]], align 4
; IND-NEXT: [[TMP25:%.*]] = fadd fast float [[S]], 1.000000e+00
; IND-NEXT: [[TMP26:%.*]] = fadd fast float [[TMP25]], [[TMP22]]
@@ -988,10 +1018,14 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; UNROLL-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 8
; UNROLL-NEXT: [[TMP4:%.*]] = or disjoint i64 [[OFFSET_IDX]], 16
; UNROLL-NEXT: [[TMP5:%.*]] = or disjoint i64 [[OFFSET_IDX]], 24
-; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; UNROLL-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP3]]
-; UNROLL-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP4]]
-; UNROLL-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP5]]
+; UNROLL-NEXT: [[TMP45:%.*]] = shl i64 [[INDEX]], 5
+; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP45]]
+; UNROLL-NEXT: [[TMP46:%.*]] = shl nsw i64 [[TMP3]], 2
+; UNROLL-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP46]]
+; UNROLL-NEXT: [[TMP48:%.*]] = shl nsw i64 [[TMP4]], 2
+; UNROLL-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP48]]
+; UNROLL-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP5]], 2
+; UNROLL-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP49]]
; UNROLL-NEXT: [[TMP10:%.*]] = load float, ptr [[TMP6]], align 4
; UNROLL-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP7]], align 4
; UNROLL-NEXT: [[TMP12:%.*]] = insertelement <2 x float> poison, float [[TMP10]], i64 0
@@ -1000,10 +1034,14 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; UNROLL-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP9]], align 4
; UNROLL-NEXT: [[TMP16:%.*]] = insertelement <2 x float> poison, float [[TMP14]], i64 0
; UNROLL-NEXT: [[TMP17:%.*]] = insertelement <2 x float> [[TMP16]], float [[TMP15]], i64 1
-; UNROLL-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; UNROLL-NEXT: [[TMP19:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP3]]
-; UNROLL-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP4]]
-; UNROLL-NEXT: [[TMP21:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP5]]
+; UNROLL-NEXT: [[TMP51:%.*]] = shl i64 [[INDEX]], 5
+; UNROLL-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP51]]
+; UNROLL-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP3]], 2
+; UNROLL-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP52]]
+; UNROLL-NEXT: [[TMP53:%.*]] = shl nsw i64 [[TMP4]], 2
+; UNROLL-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP53]]
+; UNROLL-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP5]], 2
+; UNROLL-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP54]]
; UNROLL-NEXT: [[TMP22:%.*]] = load float, ptr [[TMP18]], align 4
; UNROLL-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP19]], align 4
; UNROLL-NEXT: [[TMP24:%.*]] = insertelement <2 x float> poison, float [[TMP22]], i64 0
@@ -1033,9 +1071,11 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[FOR_BODY]] ]
; UNROLL-NEXT: [[S:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[TMP44:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT: [[TMP38:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I]]
+; UNROLL-NEXT: [[TMP47:%.*]] = shl nsw i64 [[I]], 2
+; UNROLL-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP47]]
; UNROLL-NEXT: [[TMP39:%.*]] = load float, ptr [[TMP38]], align 4
-; UNROLL-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[I]]
+; UNROLL-NEXT: [[TMP50:%.*]] = shl nsw i64 [[I]], 2
+; UNROLL-NEXT: [[TMP40:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP50]]
; UNROLL-NEXT: [[TMP41:%.*]] = load float, ptr [[TMP40]], align 4
; UNROLL-NEXT: [[TMP42:%.*]] = fadd fast float [[S]], 1.000000e+00
; UNROLL-NEXT: [[TMP43:%.*]] = fadd fast float [[TMP42]], [[TMP39]]
@@ -1148,14 +1188,18 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; INTERLEAVE-NEXT: [[VEC_PHI1:%.*]] = phi <4 x float> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
; INTERLEAVE-NEXT: [[TMP5:%.*]] = or disjoint i64 [[OFFSET_IDX]], 32
-; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP5]]
+; INTERLEAVE-NEXT: [[TMP25:%.*]] = shl i64 [[INDEX]], 5
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP25]]
+; INTERLEAVE-NEXT: [[TMP27:%.*]] = shl nsw i64 [[TMP5]], 2
+; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP27]]
; INTERLEAVE-NEXT: [[WIDE_VEC:%.*]] = load <32 x float>, ptr [[TMP6]], align 4
; INTERLEAVE-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <32 x float> [[WIDE_VEC]], <32 x float> poison, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
; INTERLEAVE-NEXT: [[WIDE_VEC2:%.*]] = load <32 x float>, ptr [[TMP7]], align 4
; INTERLEAVE-NEXT: [[STRIDED_VEC3:%.*]] = shufflevector <32 x float> [[WIDE_VEC2]], <32 x float> poison, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
-; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; INTERLEAVE-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP5]]
+; INTERLEAVE-NEXT: [[TMP28:%.*]] = shl i64 [[INDEX]], 5
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP28]]
+; INTERLEAVE-NEXT: [[TMP29:%.*]] = shl nsw i64 [[TMP5]], 2
+; INTERLEAVE-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP29]]
; INTERLEAVE-NEXT: [[WIDE_VEC4:%.*]] = load <32 x float>, ptr [[TMP8]], align 4
; INTERLEAVE-NEXT: [[STRIDED_VEC5:%.*]] = shufflevector <32 x float> [[WIDE_VEC4]], <32 x float> poison, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
; INTERLEAVE-NEXT: [[WIDE_VEC6:%.*]] = load <32 x float>, ptr [[TMP9]], align 4
@@ -1180,9 +1224,11 @@ define float @scalarize_induction_variable_02(ptr %a, ptr %b, i64 %n) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[FOR_BODY]] ]
; INTERLEAVE-NEXT: [[S:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[TMP24:%.*]], [[FOR_BODY]] ]
-; INTERLEAVE-NEXT: [[TMP18:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP30:%.*]] = shl nsw i64 [[I]], 2
+; INTERLEAVE-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP30]]
; INTERLEAVE-NEXT: [[TMP19:%.*]] = load float, ptr [[TMP18]], align 4
-; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP26:%.*]] = shl nsw i64 [[I]], 2
+; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP26]]
; INTERLEAVE-NEXT: [[TMP21:%.*]] = load float, ptr [[TMP20]], align 4
; INTERLEAVE-NEXT: [[TMP22:%.*]] = fadd fast float [[S]], 1.000000e+00
; INTERLEAVE-NEXT: [[TMP23:%.*]] = fadd fast float [[TMP22]], [[TMP19]]
@@ -1283,9 +1329,11 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; IND-NEXT: br label [[VECTOR_BODY:%.*]]
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; IND-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; IND-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 3
+; IND-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP0]]
; IND-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; IND-NEXT: [[TMP13:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; IND-NEXT: [[TMP14:%.*]] = shl i64 [[INDEX]], 3
+; IND-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP14]]
; IND-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP13]], i64 12
; IND-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 8
; IND-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP2]], align 8
@@ -1307,7 +1355,8 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; IND-NEXT: br label [[FOR_BODY:%.*]]
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; IND-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; IND-NEXT: [[TMP15:%.*]] = shl nsw i64 [[I]], 3
+; IND-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP15]]
; IND-NEXT: [[F:%.*]] = getelementptr inbounds nuw i8, ptr [[F_SPLIT]], i64 4
; IND-NEXT: [[TMP11:%.*]] = load i32, ptr [[F]], align 8
; IND-NEXT: [[TMP12:%.*]] = xor i32 [[TMP11]], [[Y]]
@@ -1330,13 +1379,17 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; UNROLL-NEXT: br label [[VECTOR_BODY:%.*]]
; UNROLL: vector.body:
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 3
+; UNROLL-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP0]]
; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 3
+; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP2]]
; UNROLL-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[TMP1]], i64 12
-; UNROLL-NEXT: [[TMP24:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP27:%.*]] = shl i64 [[INDEX]], 3
+; UNROLL-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP27]]
; UNROLL-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[TMP24]], i64 20
-; UNROLL-NEXT: [[TMP25:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP28:%.*]] = shl i64 [[INDEX]], 3
+; UNROLL-NEXT: [[TMP25:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP28]]
; UNROLL-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP25]], i64 28
; UNROLL-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP3]], align 8
; UNROLL-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP4]], align 8
@@ -1367,7 +1420,8 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; UNROLL-NEXT: br label [[FOR_BODY:%.*]]
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; UNROLL-NEXT: [[TMP26:%.*]] = shl nsw i64 [[I]], 3
+; UNROLL-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP26]]
; UNROLL-NEXT: [[F:%.*]] = getelementptr inbounds nuw i8, ptr [[F_SPLIT]], i64 4
; UNROLL-NEXT: [[TMP22:%.*]] = load i32, ptr [[F]], align 8
; UNROLL-NEXT: [[TMP23:%.*]] = xor i32 [[TMP22]], [[Y]]
@@ -1452,21 +1506,29 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; INTERLEAVE-NEXT: br label [[VECTOR_BODY:%.*]]
; INTERLEAVE: vector.body:
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; INTERLEAVE-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP2]]
; INTERLEAVE-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP4:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP4]]
; INTERLEAVE-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[TMP3]], i64 12
-; INTERLEAVE-NEXT: [[TMP5:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP8]]
; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP5]], i64 20
-; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP34:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP34]]
; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP7]], i64 28
-; INTERLEAVE-NEXT: [[TMP30:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP35:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP30:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP35]]
; INTERLEAVE-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP30]], i64 36
-; INTERLEAVE-NEXT: [[TMP31:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP37:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP31:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP37]]
; INTERLEAVE-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP31]], i64 44
-; INTERLEAVE-NEXT: [[TMP32:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP38:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP32:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP38]]
; INTERLEAVE-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[TMP32]], i64 52
-; INTERLEAVE-NEXT: [[TMP33:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP39:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP33:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP39]]
; INTERLEAVE-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[TMP33]], i64 60
; INTERLEAVE-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP9]], align 8
; INTERLEAVE-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
@@ -1500,7 +1562,8 @@ define void @scalarize_induction_variable_03(ptr %p, i32 %y, i64 %n) {
; INTERLEAVE-NEXT: br label [[FOR_BODY:%.*]]
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; INTERLEAVE-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP36:%.*]] = shl nsw i64 [[I]], 3
+; INTERLEAVE-NEXT: [[F_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP36]]
; INTERLEAVE-NEXT: [[F:%.*]] = getelementptr inbounds nuw i8, ptr [[F_SPLIT]], i64 4
; INTERLEAVE-NEXT: [[TMP28:%.*]] = load i32, ptr [[F]], align 8
; INTERLEAVE-NEXT: [[TMP29:%.*]] = xor i32 [[TMP28]], [[Y]]
@@ -1631,16 +1694,20 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; IND-NEXT: [[TMP10:%.*]] = shl nsw <2 x i64> [[VEC_IND]], splat (i64 2)
; IND-NEXT: [[TMP11:%.*]] = extractelement <2 x i64> [[TMP10]], i64 0
; IND-NEXT: [[TMP13:%.*]] = extractelement <2 x i64> [[TMP10]], i64 1
-; IND-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[TMP11]]
-; IND-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[TMP13]]
-; IND-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP12]], align 1, !alias.scope [[META17:![0-9]+]]
-; IND-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP24]], align 1, !alias.scope [[META17]]
-; IND-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P]], i64 [[INDEX]]
+; IND-NEXT: [[TMP12:%.*]] = shl nsw i64 [[TMP11]], 2
+; IND-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP12]]
+; IND-NEXT: [[TMP14:%.*]] = shl nsw i64 [[TMP13]], 2
+; IND-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP14]]
+; IND-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 1, !alias.scope [[META17:![0-9]+]]
+; IND-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP15]], align 1, !alias.scope [[META17]]
+; IND-NEXT: [[TMP28:%.*]] = shl nsw i64 [[INDEX]], 3
+; IND-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP28]]
; IND-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; IND-NEXT: [[TMP17:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; IND-NEXT: [[TMP29:%.*]] = shl i64 [[INDEX]], 3
+; IND-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP29]]
; IND-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[TMP17]], i64 12
-; IND-NEXT: store i32 [[TMP14]], ptr [[TMP16]], align 1, !alias.scope [[META20:![0-9]+]], !noalias [[META17]]
-; IND-NEXT: store i32 [[TMP15]], ptr [[TMP18]], align 1, !alias.scope [[META20]], !noalias [[META17]]
+; IND-NEXT: store i32 [[TMP25]], ptr [[TMP16]], align 1, !alias.scope [[META20:![0-9]+]], !noalias [[META17]]
+; IND-NEXT: store i32 [[TMP27]], ptr [[TMP18]], align 1, !alias.scope [[META20]], !noalias [[META17]]
; IND-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <2 x i64> [[VEC_IND]], splat (i64 2)
; IND-NEXT: [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -1656,7 +1723,8 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; IND-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[I]], 4
; IND-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[DOTIDX]]
; IND-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 1
-; IND-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; IND-NEXT: [[TMP26:%.*]] = shl nsw i64 [[I]], 3
+; IND-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP26]]
; IND-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT4]], i64 4
; IND-NEXT: store i32 [[TMP21]], ptr [[TMP22]], align 1
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -1700,26 +1768,34 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; UNROLL-NEXT: [[TMP13:%.*]] = add <2 x i64> [[STEP_ADD]], splat (i64 8)
; UNROLL-NEXT: [[TMP20:%.*]] = extractelement <2 x i64> [[TMP13]], i64 0
; UNROLL-NEXT: [[TMP35:%.*]] = extractelement <2 x i64> [[TMP13]], i64 1
-; UNROLL-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP16]]
-; UNROLL-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP18]]
-; UNROLL-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP20]]
-; UNROLL-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP35]]
-; UNROLL-NEXT: [[TMP37:%.*]] = load i32, ptr [[TMP17]], align 1, !alias.scope [[META17:![0-9]+]]
-; UNROLL-NEXT: [[TMP38:%.*]] = load i32, ptr [[TMP19]], align 1, !alias.scope [[META17]]
-; UNROLL-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP21]], align 1, !alias.scope [[META17]]
-; UNROLL-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP36]], align 1, !alias.scope [[META17]]
-; UNROLL-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP16]], 2
+; UNROLL-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP15]]
+; UNROLL-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP18]], 2
+; UNROLL-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP17]]
+; UNROLL-NEXT: [[TMP19:%.*]] = shl nsw i64 [[TMP20]], 2
+; UNROLL-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
+; UNROLL-NEXT: [[TMP21:%.*]] = shl nsw i64 [[TMP35]], 2
+; UNROLL-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP21]]
+; UNROLL-NEXT: [[TMP39:%.*]] = load i32, ptr [[TMP36]], align 1, !alias.scope [[META17:![0-9]+]]
+; UNROLL-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP37]], align 1, !alias.scope [[META17]]
+; UNROLL-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP38]], align 1, !alias.scope [[META17]]
+; UNROLL-NEXT: [[TMP43:%.*]] = load i32, ptr [[TMP22]], align 1, !alias.scope [[META17]]
+; UNROLL-NEXT: [[TMP44:%.*]] = shl nsw i64 [[INDEX]], 3
+; UNROLL-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP44]]
; UNROLL-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; UNROLL-NEXT: [[TMP24:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP45:%.*]] = shl i64 [[INDEX]], 3
+; UNROLL-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP45]]
; UNROLL-NEXT: [[TMP25:%.*]] = getelementptr i8, ptr [[TMP24]], i64 12
-; UNROLL-NEXT: [[TMP26:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP46:%.*]] = shl i64 [[INDEX]], 3
+; UNROLL-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP46]]
; UNROLL-NEXT: [[TMP27:%.*]] = getelementptr i8, ptr [[TMP26]], i64 20
-; UNROLL-NEXT: [[TMP28:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP47:%.*]] = shl i64 [[INDEX]], 3
+; UNROLL-NEXT: [[TMP28:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP47]]
; UNROLL-NEXT: [[TMP29:%.*]] = getelementptr i8, ptr [[TMP28]], i64 28
-; UNROLL-NEXT: store i32 [[TMP37]], ptr [[TMP23]], align 1, !alias.scope [[META20:![0-9]+]], !noalias [[META17]]
-; UNROLL-NEXT: store i32 [[TMP38]], ptr [[TMP25]], align 1, !alias.scope [[META20]], !noalias [[META17]]
-; UNROLL-NEXT: store i32 [[TMP39]], ptr [[TMP27]], align 1, !alias.scope [[META20]], !noalias [[META17]]
-; UNROLL-NEXT: store i32 [[TMP22]], ptr [[TMP29]], align 1, !alias.scope [[META20]], !noalias [[META17]]
+; UNROLL-NEXT: store i32 [[TMP39]], ptr [[TMP23]], align 1, !alias.scope [[META20:![0-9]+]], !noalias [[META17]]
+; UNROLL-NEXT: store i32 [[TMP40]], ptr [[TMP25]], align 1, !alias.scope [[META20]], !noalias [[META17]]
+; UNROLL-NEXT: store i32 [[TMP42]], ptr [[TMP27]], align 1, !alias.scope [[META20]], !noalias [[META17]]
+; UNROLL-NEXT: store i32 [[TMP43]], ptr [[TMP29]], align 1, !alias.scope [[META20]], !noalias [[META17]]
; UNROLL-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; UNROLL-NEXT: [[VEC_IND_NEXT]] = add nuw <2 x i64> [[VEC_IND]], splat (i64 4)
; UNROLL-NEXT: [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -1735,7 +1811,8 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; UNROLL-NEXT: [[DOTIDX:%.*]] = shl nsw i64 [[I]], 4
; UNROLL-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[DOTIDX]]
; UNROLL-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 1
-; UNROLL-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; UNROLL-NEXT: [[TMP41:%.*]] = shl nsw i64 [[I]], 3
+; UNROLL-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP41]]
; UNROLL-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT6]], i64 4
; UNROLL-NEXT: store i32 [[TMP32]], ptr [[TMP33]], align 1
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -1868,21 +1945,29 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; INTERLEAVE-NEXT: [[TMP33:%.*]] = extractelement <16 x i32> [[WIDE_VEC3]], i64 4
; INTERLEAVE-NEXT: [[TMP34:%.*]] = extractelement <16 x i32> [[WIDE_VEC3]], i64 8
; INTERLEAVE-NEXT: [[TMP35:%.*]] = extractelement <16 x i32> [[WIDE_VEC3]], i64 12
-; INTERLEAVE-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP44:%.*]] = shl nsw i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP44]]
; INTERLEAVE-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; INTERLEAVE-NEXT: [[TMP15:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP45:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP45]]
; INTERLEAVE-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[TMP15]], i64 12
-; INTERLEAVE-NEXT: [[TMP17:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP46:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP46]]
; INTERLEAVE-NEXT: [[TMP42:%.*]] = getelementptr i8, ptr [[TMP17]], i64 20
-; INTERLEAVE-NEXT: [[TMP43:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP48:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP43:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP48]]
; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP43]], i64 28
-; INTERLEAVE-NEXT: [[DOTSPLIT9:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP14]]
+; INTERLEAVE-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP14]], 3
+; INTERLEAVE-NEXT: [[DOTSPLIT9:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP49]]
; INTERLEAVE-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT9]], i64 4
-; INTERLEAVE-NEXT: [[TMP22:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP50:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP22:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP50]]
; INTERLEAVE-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[TMP22]], i64 44
-; INTERLEAVE-NEXT: [[TMP24:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP51:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP51]]
; INTERLEAVE-NEXT: [[TMP25:%.*]] = getelementptr i8, ptr [[TMP24]], i64 52
-; INTERLEAVE-NEXT: [[TMP26:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP52:%.*]] = shl i64 [[INDEX]], 3
+; INTERLEAVE-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP52]]
; INTERLEAVE-NEXT: [[TMP27:%.*]] = getelementptr i8, ptr [[TMP26]], i64 60
; INTERLEAVE-NEXT: store i32 [[TMP28]], ptr [[TMP41]], align 1, !alias.scope [[META20:![0-9]+]], !noalias [[META17]]
; INTERLEAVE-NEXT: store i32 [[TMP29]], ptr [[TMP16]], align 1, !alias.scope [[META20]], !noalias [[META17]]
@@ -1905,7 +1990,8 @@ define void @scalarize_induction_variable_04(ptr %a, ptr %p, i32 %n) {
; INTERLEAVE-NEXT: [[DOTIDX6:%.*]] = shl nsw i64 [[I]], 4
; INTERLEAVE-NEXT: [[TMP37:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[DOTIDX6]]
; INTERLEAVE-NEXT: [[TMP38:%.*]] = load i32, ptr [[TMP37]], align 1
-; INTERLEAVE-NEXT: [[DOTSPLIT14:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP47:%.*]] = shl nsw i64 [[I]], 3
+; INTERLEAVE-NEXT: [[DOTSPLIT14:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP47]]
; INTERLEAVE-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT14]], i64 4
; INTERLEAVE-NEXT: store i32 [[TMP38]], ptr [[TMP39]], align 1
; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -2027,7 +2113,8 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; IND-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_UDIV_CONTINUE2:%.*]] ]
; IND-NEXT: [[VEC_PHI:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[PRED_UDIV_CONTINUE2]] ]
; IND-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; IND-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; IND-NEXT: [[TMP10:%.*]] = shl nsw i64 [[TMP0]], 2
+; IND-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP10]]
; IND-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP1]], align 4
; IND-NEXT: br i1 [[C:%.*]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
; IND: pred.udiv.if:
@@ -2062,7 +2149,8 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; IND-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[IF_END:%.*]] ]
; IND-NEXT: [[SUM:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[VAR4:%.*]], [[IF_END]] ]
; IND-NEXT: [[TMP16:%.*]] = zext nneg i32 [[I]] to i64
-; IND-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[TMP16]]
+; IND-NEXT: [[TMP17:%.*]] = shl nuw nsw i64 [[TMP16]], 2
+; IND-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP17]]
; IND-NEXT: [[VAR1:%.*]] = load i32, ptr [[VAR0]], align 4
; IND-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[IF_END]]
; IND: if.then:
@@ -2091,7 +2179,8 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; UNROLL-NEXT: [[VEC_PHI:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP22:%.*]], [[PRED_UDIV_CONTINUE8]] ]
; UNROLL-NEXT: [[VEC_PHI1:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP23:%.*]], [[PRED_UDIV_CONTINUE8]] ]
; UNROLL-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; UNROLL-NEXT: [[TMP21:%.*]] = shl nsw i64 [[TMP0]], 2
+; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP21]]
; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 8
; UNROLL-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP1]], align 4
; UNROLL-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
@@ -2149,7 +2238,8 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; UNROLL-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[IF_END:%.*]] ]
; UNROLL-NEXT: [[SUM:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[VAR4:%.*]], [[IF_END]] ]
; UNROLL-NEXT: [[TMP26:%.*]] = zext nneg i32 [[I]] to i64
-; UNROLL-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[TMP26]]
+; UNROLL-NEXT: [[TMP27:%.*]] = shl nuw nsw i64 [[TMP26]], 2
+; UNROLL-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP27]]
; UNROLL-NEXT: [[VAR1:%.*]] = load i32, ptr [[VAR0]], align 4
; UNROLL-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[IF_END]]
; UNROLL: if.then:
@@ -2266,7 +2356,8 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; INTERLEAVE-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP42:%.*]], [[PRED_UDIV_CONTINUE16]] ]
; INTERLEAVE-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP43:%.*]], [[PRED_UDIV_CONTINUE16]] ]
; INTERLEAVE-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; INTERLEAVE-NEXT: [[TMP41:%.*]] = shl nsw i64 [[TMP0]], 2
+; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP41]]
; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
; INTERLEAVE-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; INTERLEAVE-NEXT: [[WIDE_LOAD2:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
@@ -2360,7 +2451,8 @@ define i32 @scalarize_induction_variable_05(ptr %a, i32 %x, i1 %c, i32 %n) {
; INTERLEAVE-NEXT: [[I:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[IF_END:%.*]] ]
; INTERLEAVE-NEXT: [[SUM:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[VAR4:%.*]], [[IF_END]] ]
; INTERLEAVE-NEXT: [[TMP46:%.*]] = zext nneg i32 [[I]] to i64
-; INTERLEAVE-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[TMP46]]
+; INTERLEAVE-NEXT: [[TMP47:%.*]] = shl nuw nsw i64 [[TMP46]], 2
+; INTERLEAVE-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP47]]
; INTERLEAVE-NEXT: [[VAR1:%.*]] = load i32, ptr [[VAR0]], align 4
; INTERLEAVE-NEXT: br i1 [[C]], label [[IF_THEN:%.*]], label [[IF_END]]
; INTERLEAVE: if.then:
@@ -2481,9 +2573,11 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; IND-NEXT: [[TMP5:%.*]] = trunc <2 x i32> [[TMP4]] to <2 x i16>
; IND-NEXT: [[TMP8:%.*]] = extractelement <2 x i16> [[TMP5]], i64 0
; IND-NEXT: [[TMP9:%.*]] = extractelement <2 x i16> [[TMP5]], i64 1
-; IND-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I16:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; IND-NEXT: [[TMP17:%.*]] = shl nsw i64 [[INDEX]], 2
+; IND-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP17]]
; IND-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 2
-; IND-NEXT: [[TMP16:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; IND-NEXT: [[TMP18:%.*]] = shl i64 [[INDEX]], 2
+; IND-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP18]]
; IND-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP16]], i64 6
; IND-NEXT: store i16 [[TMP8]], ptr [[TMP6]], align 2
; IND-NEXT: store i16 [[TMP9]], ptr [[TMP7]], align 2
@@ -2502,7 +2596,8 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; IND-NEXT: [[TMP11:%.*]] = trunc i64 [[I]] to i32
; IND-NEXT: [[TMP12:%.*]] = add i32 [[A]], [[TMP11]]
; IND-NEXT: [[TMP13:%.*]] = trunc i32 [[TMP12]] to i16
-; IND-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds nuw [[PAIR_I16]], ptr [[P]], i64 [[I]]
+; IND-NEXT: [[TMP19:%.*]] = shl nsw i64 [[I]], 2
+; IND-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP19]]
; IND-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT2]], i64 2
; IND-NEXT: store i16 [[TMP13]], ptr [[TMP14]], align 2
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -2536,13 +2631,17 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; UNROLL-NEXT: [[TMP9:%.*]] = trunc <2 x i32> [[TMP7]] to <2 x i16>
; UNROLL-NEXT: [[TMP16:%.*]] = extractelement <2 x i16> [[TMP9]], i64 0
; UNROLL-NEXT: [[TMP17:%.*]] = extractelement <2 x i16> [[TMP9]], i64 1
-; UNROLL-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I16:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP27:%.*]] = shl nsw i64 [[INDEX]], 2
+; UNROLL-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP27]]
; UNROLL-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 2
-; UNROLL-NEXT: [[TMP24:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP28:%.*]] = shl i64 [[INDEX]], 2
+; UNROLL-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP28]]
; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP24]], i64 6
-; UNROLL-NEXT: [[TMP25:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP29:%.*]] = shl i64 [[INDEX]], 2
+; UNROLL-NEXT: [[TMP25:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP29]]
; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP25]], i64 10
-; UNROLL-NEXT: [[TMP26:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP30:%.*]] = shl i64 [[INDEX]], 2
+; UNROLL-NEXT: [[TMP26:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP30]]
; UNROLL-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP26]], i64 14
; UNROLL-NEXT: store i16 [[TMP14]], ptr [[TMP10]], align 2
; UNROLL-NEXT: store i16 [[TMP15]], ptr [[TMP11]], align 2
@@ -2563,7 +2662,8 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; UNROLL-NEXT: [[TMP19:%.*]] = trunc i64 [[I]] to i32
; UNROLL-NEXT: [[TMP20:%.*]] = add i32 [[A]], [[TMP19]]
; UNROLL-NEXT: [[TMP21:%.*]] = trunc i32 [[TMP20]] to i16
-; UNROLL-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw [[PAIR_I16]], ptr [[P]], i64 [[I]]
+; UNROLL-NEXT: [[TMP31:%.*]] = shl nsw i64 [[I]], 2
+; UNROLL-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP31]]
; UNROLL-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT4]], i64 2
; UNROLL-NEXT: store i16 [[TMP21]], ptr [[TMP22]], align 2
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -2662,21 +2762,29 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; INTERLEAVE-NEXT: [[TMP27:%.*]] = extractelement <4 x i16> [[TMP13]], i64 1
; INTERLEAVE-NEXT: [[TMP28:%.*]] = extractelement <4 x i16> [[TMP13]], i64 2
; INTERLEAVE-NEXT: [[TMP29:%.*]] = extractelement <4 x i16> [[TMP13]], i64 3
-; INTERLEAVE-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I16:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP43:%.*]] = shl nsw i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP43]]
; INTERLEAVE-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 2
-; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP44:%.*]] = shl i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP44]]
; INTERLEAVE-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[TMP8]], i64 6
-; INTERLEAVE-NEXT: [[TMP36:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP45:%.*]] = shl i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP36:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP45]]
; INTERLEAVE-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[TMP36]], i64 10
-; INTERLEAVE-NEXT: [[TMP37:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP46:%.*]] = shl i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP37:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP46]]
; INTERLEAVE-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP37]], i64 14
-; INTERLEAVE-NEXT: [[TMP38:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP47:%.*]] = shl i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP38:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP47]]
; INTERLEAVE-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[TMP38]], i64 18
-; INTERLEAVE-NEXT: [[TMP39:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP48:%.*]] = shl i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP39:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP48]]
; INTERLEAVE-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[TMP39]], i64 22
-; INTERLEAVE-NEXT: [[TMP40:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP49:%.*]] = shl i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP40:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP49]]
; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP40]], i64 26
-; INTERLEAVE-NEXT: [[TMP41:%.*]] = getelementptr [[PAIR_I16]], ptr [[P]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP50:%.*]] = shl i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP41:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP50]]
; INTERLEAVE-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP41]], i64 30
; INTERLEAVE-NEXT: store i16 [[TMP22]], ptr [[TMP14]], align 2
; INTERLEAVE-NEXT: store i16 [[TMP23]], ptr [[TMP15]], align 2
@@ -2701,7 +2809,8 @@ define void @iv_vector_and_scalar_users(ptr %p, i32 %a, i32 %n) {
; INTERLEAVE-NEXT: [[TMP31:%.*]] = trunc i64 [[I]] to i32
; INTERLEAVE-NEXT: [[TMP32:%.*]] = add i32 [[A]], [[TMP31]]
; INTERLEAVE-NEXT: [[TMP33:%.*]] = trunc i32 [[TMP32]] to i16
-; INTERLEAVE-NEXT: [[DOTSPLIT8:%.*]] = getelementptr inbounds nuw [[PAIR_I16]], ptr [[P]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP42:%.*]] = shl nsw i64 [[I]], 2
+; INTERLEAVE-NEXT: [[DOTSPLIT8:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP42]]
; INTERLEAVE-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT8]], i64 2
; INTERLEAVE-NEXT: store i16 [[TMP33]], ptr [[TMP34]], align 2
; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -3372,7 +3481,8 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; IND-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8
; IND-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]]
; IND-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
-; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
+; IND-NEXT: [[TMP14:%.*]] = shl nsw i64 [[TMP10]], 2
+; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP14]]
; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP11]], align 4
; IND-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add <2 x i32> [[VEC_IND]], splat (i32 2)
@@ -3391,7 +3501,8 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; IND-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
; IND-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; IND-NEXT: [[TMP13:%.*]] = sext i8 [[IDX]] to i64
-; IND-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP13]]
+; IND-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP13]], 2
+; IND-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP17]]
; IND-NEXT: store i32 [[SPHI]], ptr [[PTR]], align 4
; IND-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
; IND-NEXT: [[IDX_INC_EXT]] = zext i8 [[IDX_INC]] to i32
@@ -3439,7 +3550,8 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; UNROLL-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8
; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]]
; UNROLL-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
-; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
+; UNROLL-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP10]], 2
+; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP15]]
; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw 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
@@ -3460,7 +3572,8 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; UNROLL-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[TMP14:%.*]] = sext i8 [[IDX]] to i64
-; UNROLL-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
+; UNROLL-NEXT: [[TMP18:%.*]] = shl nsw i64 [[TMP14]], 2
+; UNROLL-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; UNROLL-NEXT: store i32 [[SPHI]], ptr [[PTR]], align 4
; UNROLL-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
; UNROLL-NEXT: [[IDX_INC_EXT]] = zext i8 [[IDX_INC]] to i32
@@ -3579,7 +3692,8 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; INTERLEAVE-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8
; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]]
; INTERLEAVE-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
-; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
+; INTERLEAVE-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP10]], 2
+; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP15]]
; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw 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
@@ -3600,7 +3714,8 @@ define void @wrappingindvars1(i8 %t, i32 %len, ptr %A) {
; INTERLEAVE-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[SPHI:%.*]] = phi i32 [ [[IDX_INC_EXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[TMP14:%.*]] = sext i8 [[IDX]] to i64
-; INTERLEAVE-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
+; INTERLEAVE-NEXT: [[TMP18:%.*]] = shl nsw i64 [[TMP14]], 2
+; INTERLEAVE-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; INTERLEAVE-NEXT: store i32 [[SPHI]], ptr [[PTR]], align 4
; INTERLEAVE-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
; INTERLEAVE-NEXT: [[IDX_INC_EXT]] = zext i8 [[IDX_INC]] to i32
@@ -3751,7 +3866,8 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; IND-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8
; IND-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]]
; IND-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
-; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
+; IND-NEXT: [[TMP14:%.*]] = shl nsw i64 [[TMP10]], 2
+; IND-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP14]]
; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP11]], align 4
; IND-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add <2 x i32> [[VEC_IND]], splat (i32 8)
@@ -3770,7 +3886,8 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; IND-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; IND-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; IND-NEXT: [[TMP13:%.*]] = sext i8 [[IDX]] to i64
-; IND-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP13]]
+; IND-NEXT: [[TMP17:%.*]] = shl nsw i64 [[TMP13]], 2
+; IND-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP17]]
; IND-NEXT: store i32 [[SPHI]], ptr [[PTR]], align 4
; IND-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
; IND-NEXT: [[IDX_INC_EXT:%.*]] = zext i8 [[IDX_INC]] to i32
@@ -3821,7 +3938,8 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; UNROLL-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8
; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]]
; UNROLL-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
-; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
+; UNROLL-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP10]], 2
+; UNROLL-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP15]]
; UNROLL-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw 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
@@ -3842,7 +3960,8 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; UNROLL-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[TMP14:%.*]] = sext i8 [[IDX]] to i64
-; UNROLL-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
+; UNROLL-NEXT: [[TMP18:%.*]] = shl nsw i64 [[TMP14]], 2
+; UNROLL-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; UNROLL-NEXT: store i32 [[SPHI]], ptr [[PTR]], align 4
; UNROLL-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
; UNROLL-NEXT: [[IDX_INC_EXT:%.*]] = zext i8 [[IDX_INC]] to i32
@@ -3967,7 +4086,8 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; INTERLEAVE-NEXT: [[DOTCAST4:%.*]] = trunc i32 [[INDEX]] to i8
; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i8 [[T]], [[DOTCAST4]]
; INTERLEAVE-NEXT: [[TMP10:%.*]] = sext i8 [[OFFSET_IDX]] to i64
-; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP10]]
+; INTERLEAVE-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP10]], 2
+; INTERLEAVE-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP15]]
; INTERLEAVE-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw 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
@@ -3988,7 +4108,8 @@ define void @wrappingindvars2(i8 %t, i32 %len, ptr %A) {
; INTERLEAVE-NEXT: [[SPHI:%.*]] = phi i32 [ [[MUL:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[IDX_B:%.*]] = phi i32 [ [[IDX_B_INC:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL3]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[TMP14:%.*]] = sext i8 [[IDX]] to i64
-; INTERLEAVE-NEXT: [[PTR:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
+; INTERLEAVE-NEXT: [[TMP18:%.*]] = shl nsw i64 [[TMP14]], 2
+; INTERLEAVE-NEXT: [[PTR:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; INTERLEAVE-NEXT: store i32 [[SPHI]], ptr [[PTR]], align 4
; INTERLEAVE-NEXT: [[IDX_INC]] = add i8 [[IDX]], 1
; INTERLEAVE-NEXT: [[IDX_INC_EXT:%.*]] = zext i8 [[IDX_INC]] to i32
@@ -4078,7 +4199,8 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; IND-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 0, i32 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; IND-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; IND-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP0]], 2
+; IND-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP1]], align 4
; IND-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <2 x i32> [[VEC_IND]], splat (i32 2)
@@ -4093,7 +4215,8 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; IND: for.body:
; IND-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[TMP3:%.*]] = sext i32 [[INDVARS_IV]] to i64
-; IND-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP3]]
+; IND-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP3]], 2
+; IND-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP5]]
; IND-NEXT: store i32 [[INDVARS_IV]], ptr [[ARRAYIDX]], align 4
; IND-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
; IND-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[K]]
@@ -4113,7 +4236,8 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; UNROLL-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 0, i32 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; UNROLL-NEXT: [[STEP_ADD:%.*]] = add nuw <2 x i32> [[VEC_IND]], splat (i32 2)
; UNROLL-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; UNROLL-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP0]], 2
+; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw 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
@@ -4130,7 +4254,8 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; UNROLL: for.body:
; UNROLL-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[TMP4:%.*]] = sext i32 [[INDVARS_IV]] to i64
-; UNROLL-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP4]]
+; UNROLL-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP4]], 2
+; UNROLL-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP6]]
; UNROLL-NEXT: store i32 [[INDVARS_IV]], ptr [[ARRAYIDX]], align 4
; UNROLL-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
; UNROLL-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[K]]
@@ -4186,7 +4311,8 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; INTERLEAVE-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add nuw <4 x i32> [[VEC_IND]], splat (i32 4)
; INTERLEAVE-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP0]]
+; INTERLEAVE-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP0]], 2
+; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw 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
@@ -4203,7 +4329,8 @@ define void @veciv(ptr nocapture %a, i32 %start, i32 %k) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[TMP4:%.*]] = sext i32 [[INDVARS_IV]] to i64
-; INTERLEAVE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP4]]
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP4]], 2
+; INTERLEAVE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP6]]
; INTERLEAVE-NEXT: store i32 [[INDVARS_IV]], ptr [[ARRAYIDX]], align 4
; INTERLEAVE-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
; INTERLEAVE-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[K]]
@@ -4515,7 +4642,8 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; IND-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[I]], [[INDEX]]
; IND-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64
-; IND-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]]
+; IND-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP1]], 2
+; IND-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP2]], align 4
; IND-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <2 x i32> [[VEC_IND]], splat (i32 2)
@@ -4530,7 +4658,8 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; IND: for.body:
; IND-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[TMP4:%.*]] = sext i32 [[INDVARS_IV]] to i64
-; IND-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP4]]
+; IND-NEXT: [[TMP7:%.*]] = shl nsw i64 [[TMP4]], 2
+; IND-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP7]]
; IND-NEXT: store i32 [[INDVARS_IV]], ptr [[ARRAYIDX]], align 4
; IND-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
; IND-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[K]]
@@ -4556,7 +4685,8 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; UNROLL-NEXT: [[STEP_ADD:%.*]] = add <2 x i32> [[VEC_IND]], splat (i32 2)
; UNROLL-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[I]], [[INDEX]]
; UNROLL-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64
-; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]]
+; UNROLL-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP1]], 2
+; UNROLL-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; UNROLL-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw 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
@@ -4573,7 +4703,8 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; UNROLL: for.body:
; UNROLL-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[TMP5:%.*]] = sext i32 [[INDVARS_IV]] to i64
-; UNROLL-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP5]]
+; UNROLL-NEXT: [[TMP8:%.*]] = shl nsw i64 [[TMP5]], 2
+; UNROLL-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; UNROLL-NEXT: store i32 [[INDVARS_IV]], ptr [[ARRAYIDX]], align 4
; UNROLL-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
; UNROLL-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[K]]
@@ -4641,7 +4772,8 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; INTERLEAVE-NEXT: [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], splat (i32 4)
; INTERLEAVE-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[I]], [[INDEX]]
; INTERLEAVE-NEXT: [[TMP1:%.*]] = sext i32 [[OFFSET_IDX]] to i64
-; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP1]]
+; INTERLEAVE-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP1]], 2
+; INTERLEAVE-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP6]]
; INTERLEAVE-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw 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
@@ -4658,7 +4790,8 @@ define void @nonprimary(ptr nocapture %a, i32 %start, i32 %i, i32 %k) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[INDVARS_IV:%.*]] = phi i32 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[TMP5:%.*]] = sext i32 [[INDVARS_IV]] to i64
-; INTERLEAVE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP5]]
+; INTERLEAVE-NEXT: [[TMP8:%.*]] = shl nsw i64 [[TMP5]], 2
+; INTERLEAVE-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
; INTERLEAVE-NEXT: store i32 [[INDVARS_IV]], ptr [[ARRAYIDX]], align 4
; INTERLEAVE-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
; INTERLEAVE-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[K]]
@@ -4733,7 +4866,8 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; IND: vector.body:
; IND-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[VEC_IND:%.*]] = phi <2 x i32> [ <i32 0, i32 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; IND-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; IND-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDEX]], 2
+; IND-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP2]]
; IND-NEXT: store <2 x i32> [[VEC_IND]], ptr [[TMP0]], align 4
; IND-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add <2 x i32> [[VEC_IND]], splat (i32 4)
@@ -4749,7 +4883,8 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; IND: for.body:
; IND-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; IND-NEXT: [[J:%.*]] = phi i64 [ [[J_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; IND-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[I]]
+; IND-NEXT: [[TMP4:%.*]] = shl nsw i64 [[I]], 2
+; IND-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP4]]
; IND-NEXT: [[VAR1:%.*]] = trunc i64 [[J]] to i32
; IND-NEXT: store i32 [[VAR1]], ptr [[VAR0]], align 4
; IND-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -4772,7 +4907,8 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; UNROLL-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], splat (i32 4)
-; UNROLL-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDEX]], 2
+; UNROLL-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP3]]
; UNROLL-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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
@@ -4790,7 +4926,8 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; UNROLL: for.body:
; UNROLL-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; UNROLL-NEXT: [[J:%.*]] = phi i64 [ [[J_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; UNROLL-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[I]]
+; UNROLL-NEXT: [[TMP5:%.*]] = shl nsw i64 [[I]], 2
+; UNROLL-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP5]]
; UNROLL-NEXT: [[VAR1:%.*]] = trunc i64 [[J]] to i32
; UNROLL-NEXT: store i32 [[VAR1]], ptr [[VAR0]], align 4
; UNROLL-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -4855,7 +4992,8 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; INTERLEAVE-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]], splat (i32 8)
-; INTERLEAVE-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP3]]
; INTERLEAVE-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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
@@ -4873,7 +5011,8 @@ define void @non_primary_iv_trunc(ptr %a, i64 %n) {
; INTERLEAVE: for.body:
; INTERLEAVE-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; INTERLEAVE-NEXT: [[J:%.*]] = phi i64 [ [[J_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; INTERLEAVE-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[I]]
+; INTERLEAVE-NEXT: [[TMP5:%.*]] = shl nsw i64 [[I]], 2
+; INTERLEAVE-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP5]]
; INTERLEAVE-NEXT: [[VAR1:%.*]] = trunc i64 [[J]] to i32
; INTERLEAVE-NEXT: store i32 [[VAR1]], ptr [[VAR0]], align 4
; INTERLEAVE-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -5981,7 +6120,8 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; IND-NEXT: [[VECTOR_RECUR:%.*]] = phi <2 x i32> [ <i32 poison, i32 0>, [[VECTOR_PH]] ], [ [[VEC_IND:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[VEC_IND]] = phi <2 x i32> [ [[TMP15]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; IND-NEXT: [[TMP17:%.*]] = shufflevector <2 x i32> [[VECTOR_RECUR]], <2 x i32> [[VEC_IND]], <2 x i32> <i32 1, i32 2>
-; IND-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[INDEX]]
+; IND-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDEX]], 2
+; IND-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 [[TMP20]]
; IND-NEXT: store <2 x i32> [[TMP17]], ptr [[TMP18]], align 4
; IND-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; IND-NEXT: [[VEC_IND_NEXT]] = add nsw <2 x i32> [[VEC_IND]], [[DOTSPLAT3]]
@@ -6002,7 +6142,8 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; IND-NEXT: [[IV_2:%.*]] = phi i32 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV_2_NEXT:%.*]], [[LOOP]] ]
; IND-NEXT: [[IV_2_EXT:%.*]] = shl i32 [[IV_2]], 24
; IND-NEXT: [[IV_2_CONV]] = ashr exact i32 [[IV_2_EXT]], 24
-; IND-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[IV_1]]
+; IND-NEXT: [[TMP22:%.*]] = shl nsw i64 [[IV_1]], 2
+; IND-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP22]]
; IND-NEXT: store i32 [[FOR]], ptr [[GEP]], align 4
; IND-NEXT: [[IV_2_NEXT]] = add nsw i32 [[IV_2_CONV]], [[STEP]]
; IND-NEXT: [[IV_1_NEXT]] = add nuw nsw i64 [[IV_1]], 1
@@ -6052,7 +6193,8 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; UNROLL-NEXT: [[STEP_ADD]] = add <2 x i32> [[VEC_IND]], [[TMP16]]
; UNROLL-NEXT: [[TMP18:%.*]] = shufflevector <2 x i32> [[VECTOR_RECUR]], <2 x i32> [[VEC_IND]], <2 x i32> <i32 1, i32 2>
; UNROLL-NEXT: [[TMP19:%.*]] = shufflevector <2 x i32> [[VEC_IND]], <2 x i32> [[STEP_ADD]], <2 x i32> <i32 1, i32 2>
-; UNROLL-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[INDEX]]
+; UNROLL-NEXT: [[TMP23:%.*]] = shl nsw i64 [[INDEX]], 2
+; UNROLL-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 [[TMP23]]
; UNROLL-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP20]], i64 8
; UNROLL-NEXT: store <2 x i32> [[TMP18]], ptr [[TMP20]], align 4
; UNROLL-NEXT: store <2 x i32> [[TMP19]], ptr [[TMP21]], align 4
@@ -6075,7 +6217,8 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; UNROLL-NEXT: [[IV_2:%.*]] = phi i32 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV_2_NEXT:%.*]], [[LOOP]] ]
; UNROLL-NEXT: [[IV_2_EXT:%.*]] = shl i32 [[IV_2]], 24
; UNROLL-NEXT: [[IV_2_CONV]] = ashr exact i32 [[IV_2_EXT]], 24
-; UNROLL-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[IV_1]]
+; UNROLL-NEXT: [[TMP24:%.*]] = shl nsw i64 [[IV_1]], 2
+; UNROLL-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP24]]
; UNROLL-NEXT: store i32 [[FOR]], ptr [[GEP]], align 4
; UNROLL-NEXT: [[IV_2_NEXT]] = add nsw i32 [[IV_2_CONV]], [[STEP]]
; UNROLL-NEXT: [[IV_1_NEXT]] = add nuw nsw i64 [[IV_1]], 1
@@ -6201,7 +6344,8 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; INTERLEAVE-NEXT: [[STEP_ADD]] = add <4 x i32> [[VEC_IND]], [[TMP16]]
; INTERLEAVE-NEXT: [[TMP18:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[VEC_IND]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
; INTERLEAVE-NEXT: [[TMP19:%.*]] = shufflevector <4 x i32> [[VEC_IND]], <4 x i32> [[STEP_ADD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[PTR:%.*]], i64 [[INDEX]]
+; INTERLEAVE-NEXT: [[TMP23:%.*]] = shl nsw i64 [[INDEX]], 2
+; INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 [[TMP23]]
; INTERLEAVE-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP20]], i64 16
; INTERLEAVE-NEXT: store <4 x i32> [[TMP18]], ptr [[TMP20]], align 4
; INTERLEAVE-NEXT: store <4 x i32> [[TMP19]], ptr [[TMP21]], align 4
@@ -6224,7 +6368,8 @@ define void @test_optimized_cast_induction_feeding_first_order_recurrence(i64 %n
; INTERLEAVE-NEXT: [[IV_2:%.*]] = phi i32 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV_2_NEXT:%.*]], [[LOOP]] ]
; INTERLEAVE-NEXT: [[IV_2_EXT:%.*]] = shl i32 [[IV_2]], 24
; INTERLEAVE-NEXT: [[IV_2_CONV]] = ashr exact i32 [[IV_2_EXT]], 24
-; INTERLEAVE-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[PTR]], i64 [[IV_1]]
+; INTERLEAVE-NEXT: [[TMP24:%.*]] = shl nsw i64 [[IV_1]], 2
+; INTERLEAVE-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i64 [[TMP24]]
; INTERLEAVE-NEXT: store i32 [[FOR]], ptr [[GEP]], align 4
; INTERLEAVE-NEXT: [[IV_2_NEXT]] = add nsw i32 [[IV_2_CONV]], [[STEP]]
; INTERLEAVE-NEXT: [[IV_1_NEXT]] = add nuw nsw i64 [[IV_1]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses-2.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-2.ll
index b8eb5311b4bb2..def1b78c196e9 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses-2.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-2.ll
@@ -43,7 +43,8 @@ define void @_Z4testPfS_m(ptr noalias nocapture %out, ptr noalias nocapture read
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 [[DOTIDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP2]], 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: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[OUT]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[OUT]], i64 [[TMP6]]
; CHECK-NEXT: store <4 x i32> [[STRIDED_VEC]], ptr [[TMP3]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -62,7 +63,8 @@ define void @_Z4testPfS_m(ptr noalias nocapture %out, ptr noalias nocapture read
; CHECK-NEXT: [[ARRAYIDX_IDX:%.*]] = shl i64 [[OUT_OFFSET_08]], 3
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 [[ARRAYIDX_IDX]]
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds float, ptr [[OUT]], i64 [[OUT_OFFSET_08]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[OUT_OFFSET_08]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[OUT]], i64 [[TMP7]]
; CHECK-NEXT: store i32 [[TMP5]], ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[INC]] = add nuw i64 [[OUT_OFFSET_08]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INC]], [[SIZE]]
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses-pred-stores.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-pred-stores.ll
index 16a56f3a38ac9..ec971c5229582 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses-pred-stores.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses-pred-stores.ll
@@ -27,7 +27,8 @@ define void @interleaved_with_cond_store_0(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP12]]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 8
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <4 x i64>, ptr [[TMP0]], align 8
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <4 x i64> [[WIDE_VEC]], <4 x i64> poison, <2 x i32> <i32 0, i32 2>
@@ -35,7 +36,8 @@ define void @interleaved_with_cond_store_0(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x i1> [[TMP1]], i64 0
; CHECK-NEXT: br i1 [[TMP2]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
-; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP14]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT3]], i64 8
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 0
; CHECK-NEXT: store i64 [[TMP4]], ptr [[TMP3]], align 8
@@ -44,7 +46,8 @@ define void @interleaved_with_cond_store_0(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <2 x i1> [[TMP1]], i64 1
; CHECK-NEXT: br i1 [[TMP5]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; CHECK: pred.store.if1:
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr [[PAIR]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP15]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP6]], i64 24
; CHECK-NEXT: [[TMP8:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 2
; CHECK-NEXT: store i64 [[TMP8]], ptr [[TMP7]], align 8
@@ -60,7 +63,8 @@ define void @interleaved_with_cond_store_0(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[IF_MERGE:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[P_1_SPLIT:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[I]], 4
+; CHECK-NEXT: [[P_1_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP13]]
; CHECK-NEXT: [[P_1:%.*]] = getelementptr inbounds nuw i8, ptr [[P_1_SPLIT]], i64 8
; CHECK-NEXT: [[TMP10:%.*]] = load i64, ptr [[P_1]], align 8
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[TMP10]], [[X]]
@@ -123,10 +127,13 @@ define void @interleaved_with_cond_store_1(ptr %p, i64 %x, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; CHECK-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[PAIR:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP17]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP18]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 8
-; CHECK-NEXT: [[DOTSPLIT5:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[TMP0]], 4
+; CHECK-NEXT: [[DOTSPLIT5:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT5]], i64 8
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <4 x i64>, ptr [[TMP2]], align 8
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <4 x i64> [[WIDE_VEC]], <4 x i64> poison, <2 x i32> <i32 0, i32 2>
@@ -134,7 +141,8 @@ define void @interleaved_with_cond_store_1(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <2 x i1> [[TMP4]], i64 0
; CHECK-NEXT: br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP22]]
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 0
; CHECK-NEXT: store i64 [[TMP7]], ptr [[TMP6]], align 8
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE]]
@@ -142,7 +150,8 @@ define void @interleaved_with_cond_store_1(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: [[TMP8:%.*]] = extractelement <2 x i1> [[TMP4]], i64 1
; CHECK-NEXT: br i1 [[TMP8]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; CHECK: pred.store.if1:
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[TMP0]], 4
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP23]]
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 2
; CHECK-NEXT: store i64 [[TMP10]], ptr [[TMP9]], align 8
; CHECK-NEXT: br label [[PRED_STORE_CONTINUE2]]
@@ -162,8 +171,10 @@ define void @interleaved_with_cond_store_1(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[IF_MERGE:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[P_0:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[I]]
-; CHECK-NEXT: [[P_1_SPLIT:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[I]], 4
+; CHECK-NEXT: [[P_0:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP19]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[I]], 4
+; CHECK-NEXT: [[P_1_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP20]]
; CHECK-NEXT: [[P_1:%.*]] = getelementptr inbounds nuw i8, ptr [[P_1_SPLIT]], i64 8
; CHECK-NEXT: [[TMP14:%.*]] = load i64, ptr [[P_1]], align 8
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[TMP14]], [[X]]
@@ -230,9 +241,12 @@ define void @interleaved_with_cond_store_2(ptr %p, i64 %x, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE2:%.*]] ]
; CHECK-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[PAIR:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[TMP0]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[TMP0]], 4
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP16]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 8
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <4 x i64>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <4 x i64> [[WIDE_VEC]], <4 x i64> poison, <2 x i32> <i32 0, i32 2>
@@ -242,7 +256,8 @@ define void @interleaved_with_cond_store_2(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <2 x i1> [[TMP4]], i64 0
; CHECK-NEXT: br i1 [[TMP5]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
-; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT3]], i64 8
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 0
; CHECK-NEXT: store i64 [[TMP7]], ptr [[TMP6]], align 8
@@ -251,7 +266,8 @@ define void @interleaved_with_cond_store_2(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: [[TMP8:%.*]] = extractelement <2 x i1> [[TMP4]], i64 1
; CHECK-NEXT: br i1 [[TMP8]], label [[PRED_STORE_IF1:%.*]], label [[PRED_STORE_CONTINUE2]]
; CHECK: pred.store.if1:
-; CHECK-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[TMP0]], 4
+; CHECK-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT4]], i64 8
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x i64> [[WIDE_VEC]], i64 2
; CHECK-NEXT: store i64 [[TMP10]], ptr [[TMP9]], align 8
@@ -267,8 +283,10 @@ define void @interleaved_with_cond_store_2(ptr %p, i64 %x, i64 %n) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[IF_MERGE:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[P_0:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[I]]
-; CHECK-NEXT: [[P_1_SPLIT:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[I]], 4
+; CHECK-NEXT: [[P_0:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP17]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[I]], 4
+; CHECK-NEXT: [[P_1_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP18]]
; CHECK-NEXT: [[P_1:%.*]] = getelementptr inbounds nuw i8, ptr [[P_1_SPLIT]], i64 8
; CHECK-NEXT: [[TMP12:%.*]] = load i64, ptr [[P_1]], align 8
; CHECK-NEXT: store i64 [[X]], ptr [[P_0]], align 8
diff --git a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
index caaf7bbd687b1..a97020e74c311 100644
--- a/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
+++ b/llvm/test/Transforms/LoopVectorize/interleaved-accesses.ll
@@ -33,14 +33,15 @@ define void @test_array_load2_store2(i32 %C, i32 %D) {
; 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 @AB, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr @AB, i64 [[TMP1]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP0]], 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>
; 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 i32, ptr @CD, i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr @CD, i64 [[DOTIDX]]
; 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 [[TMP4]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -109,7 +110,8 @@ define void @test_struct_array_load3_store3() {
; CHECK-NEXT: [[STRIDED_VEC2:%.*]] = shufflevector <12 x i32> [[WIDE_VEC]], <12 x i32> poison, <4 x i32> <i32 1, i32 4, i32 7, i32 10>
; CHECK-NEXT: [[STRIDED_VEC3:%.*]] = shufflevector <12 x i32> [[WIDE_VEC]], <12 x i32> poison, <4 x i32> <i32 2, i32 5, i32 8, i32 11>
; CHECK-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[STRIDED_VEC]], splat (i32 1)
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ST3:%.*]], ptr @S, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = mul nsw i64 [[INDEX]], 12
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr @S, i64 [[TMP7]]
; CHECK-NEXT: [[TMP1:%.*]] = add nsw <4 x i32> [[STRIDED_VEC2]], splat (i32 2)
; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[STRIDED_VEC3]], splat (i32 3)
; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <4 x i32> [[TMP0]], <4 x i32> [[TMP1]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -184,7 +186,8 @@ define i32 @test_struct_load4(ptr nocapture readonly %S) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ST4:%.*]], ptr [[S:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[S:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <16 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <16 x i32> [[WIDE_VEC]], <16 x i32> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
; CHECK-NEXT: [[STRIDED_VEC1:%.*]] = shufflevector <16 x i32> [[WIDE_VEC]], <16 x i32> poison, <4 x i32> <i32 1, i32 5, i32 9, i32 13>
@@ -255,7 +258,8 @@ define void @test_struct_store4(ptr noalias nocapture readonly %A, ptr noalias n
; CHECK-NEXT: [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[NEXT_GEP]], align 4
; CHECK-NEXT: [[TMP0:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], splat (i32 1)
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_ST4:%.*]], ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP8]]
; CHECK-NEXT: [[TMP1:%.*]] = shl nsw <4 x i32> [[WIDE_LOAD]], splat (i32 1)
; CHECK-NEXT: [[TMP2:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], splat (i32 3)
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_LOAD]], splat (i32 4)
@@ -329,7 +333,8 @@ define void @test_reversed_load2_store2(ptr noalias nocapture readonly %A, ptr n
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; 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]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[OFFSET_IDX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP7]]
; 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>
@@ -338,7 +343,8 @@ define void @test_reversed_load2_store2(ptr noalias nocapture readonly %A, ptr n
; CHECK-NEXT: [[REVERSE2:%.*]] = shufflevector <4 x i32> [[STRIDED_VEC1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
; 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]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[OFFSET_IDX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP8]]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i64 -24
; 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>
@@ -401,8 +407,8 @@ define void @even_load_static_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP0]], 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: [[TMP1:%.*]] = shl nsw <4 x i32> [[STRIDED_VEC]], splat (i32 1)
-; CHECK-NEXT: [[TMP2:%.*]] = and i64 [[INDEX]], 9223372036854775804
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[TMP2]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: store <4 x i32> [[TMP1]], ptr [[TMP3]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 508
@@ -415,11 +421,12 @@ define void @even_load_static_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 1016, [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL:%.*]] = shl nsw i32 [[TMP]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = lshr exact i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP6]]
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 2
; CHECK-NEXT: [[CMP:%.*]] = icmp samesign ult i64 [[INDVARS_IV]], 1022
@@ -477,8 +484,8 @@ define void @even_load_dynamic_tc(ptr noalias nocapture readonly %A, ptr noalias
; 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: [[TMP6:%.*]] = shl nsw <4 x i32> [[STRIDED_VEC]], splat (i32 1)
-; CHECK-NEXT: [[TMP7:%.*]] = and i64 [[INDEX]], 9223372036854775804
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[TMP7]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: store <4 x i32> [[TMP6]], ptr [[TMP8]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -492,11 +499,12 @@ define void @even_load_dynamic_tc(ptr noalias nocapture readonly %A, ptr noalias
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP11]]
; CHECK-NEXT: [[TMP:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL:%.*]] = shl nsw i32 [[TMP]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = lshr exact i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP12]]
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 2
; CHECK-NEXT: [[CMP:%.*]] = icmp ult i64 [[INDVARS_IV_NEXT]], [[N]]
@@ -561,17 +569,25 @@ define void @load_gap_reverse(ptr noalias nocapture %P1, ptr noalias nocapture %
; CHECK-NEXT: [[TMP22:%.*]] = extractelement <4 x i64> [[TMP3]], i64 1
; CHECK-NEXT: [[TMP23:%.*]] = extractelement <4 x i64> [[TMP3]], i64 2
; CHECK-NEXT: [[TMP24:%.*]] = extractelement <4 x i64> [[TMP3]], i64 3
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[PAIR:%.*]], ptr [[P1:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P1]], i64 [[TMP0]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P1]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P1]], i64 [[TMP2]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P2:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl nsw i64 [[OFFSET_IDX]], 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[P1:%.*]], i64 [[TMP30]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nsw i64 [[TMP0]], 4
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP32:%.*]] = shl nsw i64 [[TMP1]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[TMP32]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl nsw i64 [[TMP2]], 4
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[P1]], i64 [[TMP33]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl nsw i64 [[OFFSET_IDX]], 4
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P2:%.*]], i64 [[TMP34]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 8
-; CHECK-NEXT: [[DOTSPLIT1:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P2]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl nsw i64 [[TMP0]], 4
+; CHECK-NEXT: [[DOTSPLIT1:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT1]], i64 8
-; CHECK-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P2]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[TMP1]], 4
+; CHECK-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[TMP36]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT2]], i64 8
-; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds [[PAIR]], ptr [[P2]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl nsw i64 [[TMP2]], 4
+; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds i8, ptr [[P2]], i64 [[TMP37]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT3]], i64 8
; CHECK-NEXT: [[TMP12:%.*]] = load i64, ptr [[TMP8]], align 8
; CHECK-NEXT: [[TMP13:%.*]] = load i64, ptr [[TMP9]], align 8
@@ -641,13 +657,14 @@ define void @mixed_load2_store2(ptr noalias nocapture readonly %A, ptr noalias n
; 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: [[TMP1:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP0]], 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>
; CHECK-NEXT: [[TMP2:%.*]] = mul nsw <4 x i32> [[STRIDED_VEC1]], [[STRIDED_VEC]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[DOTIDX]]
; CHECK-NEXT: [[STRIDED_VEC3:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; 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]]
@@ -793,7 +810,8 @@ define void @int_float_struct(ptr nocapture readonly %A) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x float> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_INTFLOAT:%.*]], ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP5]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP0]], 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_VEC2:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
@@ -870,17 +888,25 @@ define void @PR27626_0(ptr %p, i32 %z, i64 %n) {
; CHECK-NEXT: [[TMP2:%.*]] = or disjoint i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP4:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP4]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP18]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP20]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[TMP4]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP23]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; CHECK-NEXT: [[DOTSPLIT1:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[DOTSPLIT1:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP24]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT1]], i64 4
-; CHECK-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT2]], i64 4
-; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[TMP4]], 3
+; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP19]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT3]], i64 4
; CHECK-NEXT: store i32 [[Z:%.*]], ptr [[TMP5]], align 4
; CHECK-NEXT: store i32 [[Z]], ptr [[TMP6]], align 4
@@ -905,8 +931,10 @@ define void @PR27626_0(ptr %p, i32 %z, 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: [[P_I_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
-; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP26]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP27]]
; CHECK-NEXT: [[P_I_Y:%.*]] = getelementptr inbounds nuw i8, ptr [[P_I_Y_SPLIT]], i64 4
; CHECK-NEXT: store i32 [[Z]], ptr [[P_I_X]], align 4
; CHECK-NEXT: store i32 [[Z]], ptr [[P_I_Y]], align 4
@@ -960,14 +988,19 @@ define i32 @PR27626_1(ptr %p, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP24]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP4]], i64 12
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP19]], i64 20
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[P]], i64 [[TMP26]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP20]], i64 28
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <8 x i32> [[WIDE_VEC]], i64 0
@@ -994,8 +1027,10 @@ define i32 @PR27626_1(ptr %p, i64 %n) {
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[S:%.*]] = phi i32 [ [[TMP18:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
-; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP23]]
; CHECK-NEXT: [[P_I_Y:%.*]] = getelementptr inbounds nuw i8, ptr [[P_I_Y_SPLIT]], i64 4
; CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[P_I_X]], align 4
; CHECK-NEXT: store i32 [[TMP17]], ptr [[P_I_Y]], align 4
@@ -1054,18 +1089,26 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: [[TMP2:%.*]] = or disjoint i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP4:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP23]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl nsw i64 [[TMP4]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP24]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 -8
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; CHECK-NEXT: [[DOTSPLIT1:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[DOTSPLIT1:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP26]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT1]], i64 4
-; CHECK-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[DOTSPLIT2:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP29]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT2]], i64 4
-; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[TMP4]], 3
+; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT3]], i64 4
; CHECK-NEXT: store i32 [[Z:%.*]], ptr [[TMP5]], align 4
; CHECK-NEXT: store i32 [[Z]], ptr [[TMP6]], align 4
@@ -1090,9 +1133,11 @@ define void @PR27626_2(ptr %p, i64 %n, i32 %z) {
; 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: [[P_I_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP27]]
; CHECK-NEXT: [[P_I_MINUS_1_X:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 -8
-; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP28]]
; CHECK-NEXT: [[P_I_Y:%.*]] = getelementptr inbounds nuw i8, ptr [[P_I_Y_SPLIT]], i64 4
; CHECK-NEXT: store i32 [[Z]], ptr [[P_I_X]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = load i32, ptr [[P_I_MINUS_1_X]], align 4
@@ -1154,16 +1199,22 @@ define i32 @PR27626_3(ptr %p, i64 %n, i32 %z) {
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i64> [[TMP2]], i64 1
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i64> [[TMP2]], i64 2
; CHECK-NEXT: [[TMP11:%.*]] = extractelement <4 x i64> [[TMP2]], i64 3
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[PAIR_I32:%.*]], ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[P:%.*]], i64 [[TMP24]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[DOTSPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT]], i64 4
-; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl nsw i64 [[TMP5]], 3
+; CHECK-NEXT: [[DOTSPLIT3:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP28]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT3]], i64 4
-; CHECK-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl nsw i64 [[TMP7]], 3
+; CHECK-NEXT: [[DOTSPLIT4:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP29]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT4]], i64 4
-; CHECK-NEXT: [[DOTSPLIT5:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl nsw i64 [[TMP9]], 3
+; CHECK-NEXT: [[DOTSPLIT5:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP30]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT5]], i64 4
-; CHECK-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds nuw [[PAIR_I32]], ptr [[P]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nsw i64 [[TMP11]], 3
+; CHECK-NEXT: [[DOTSPLIT6:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP31]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[DOTSPLIT6]], i64 4
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = extractelement <8 x i32> [[WIDE_VEC]], i64 0
@@ -1191,10 +1242,13 @@ define i32 @PR27626_3(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: [[S:%.*]] = phi i32 [ [[TMP22:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
-; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_X:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP26]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[P_I_Y_SPLIT:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 [[TMP27]]
; CHECK-NEXT: [[P_I_Y:%.*]] = getelementptr inbounds nuw i8, ptr [[P_I_Y_SPLIT]], i64 4
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr [[PAIR_I32]], ptr [[P]], i64 [[I]]
+; CHECK-NEXT: [[I_PLUS_1:%.*]] = shl i64 [[I]], 3
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[P]], i64 [[I_PLUS_1]]
; CHECK-NEXT: [[P_I_PLUS_1_Y:%.*]] = getelementptr i8, ptr [[TMP23]], i64 12
; CHECK-NEXT: [[TMP20:%.*]] = load i32, ptr [[P_I_X]], align 4
; CHECK-NEXT: store i32 [[TMP20]], ptr [[P_I_PLUS_1_Y]], align 4
@@ -1261,13 +1315,16 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, 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: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[A]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP5]], i64 8
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP16]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP7]], i64 16
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[A]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[A]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[TMP11]], i64 24
; CHECK-NEXT: store i32 [[X:%.*]], ptr [[TMP14]], align 4
; CHECK-NEXT: store i32 [[X]], ptr [[TMP8]], align 4
@@ -1286,8 +1343,10 @@ define void @PR27626_4(ptr %a, i32 %x, i32 %y, i32 %z, 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: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i32, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP17]]
+; CHECK-NEXT: [[I_PLUS_1:%.*]] = shl i64 [[I]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[A]], i64 [[I_PLUS_1]]
; CHECK-NEXT: [[A_I_PLUS_1:%.*]] = getelementptr i8, ptr [[TMP15]], i64 4
; CHECK-NEXT: store i32 [[Y]], ptr [[TMP12]], align 4
; CHECK-NEXT: store i32 [[Z]], ptr [[A_I_PLUS_1]], align 4
@@ -1345,7 +1404,7 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 3, i64 5, i64 7, i64 9>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP5:%.*]] = shl i64 [[INDEX]], 3
; CHECK-NEXT: [[TMP7:%.*]] = add nsw <4 x i64> [[VEC_IND]], splat (i64 -1)
; CHECK-NEXT: [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i64 0
; CHECK-NEXT: [[TMP16:%.*]] = extractelement <4 x i64> [[TMP7]], i64 1
@@ -1356,22 +1415,33 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) {
; CHECK-NEXT: [[TMP24:%.*]] = extractelement <4 x i64> [[TMP8]], i64 1
; CHECK-NEXT: [[TMP26:%.*]] = extractelement <4 x i64> [[TMP8]], i64 2
; CHECK-NEXT: [[TMP28:%.*]] = extractelement <4 x i64> [[TMP8]], i64 3
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[A:%.*]], i64 [[TMP4]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP12]], i64 12
-; CHECK-NEXT: [[TMP33:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP33:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP37]]
; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[TMP33]], i64 20
-; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP39:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP39]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP34]], i64 28
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr i32, ptr [[A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP35]], i64 36
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP16]]
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP18]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP20]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP22]]
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP24]]
-; CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP26]]
-; CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP28]]
+; CHECK-NEXT: [[TMP41:%.*]] = shl nsw i64 [[TMP14]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP41]]
+; CHECK-NEXT: [[TMP42:%.*]] = shl nsw i64 [[TMP16]], 2
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP42]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nsw i64 [[TMP18]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP44]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i64 [[TMP20]], 2
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP45]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i64 [[TMP22]], 2
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP46]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[TMP24]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP36]]
+; CHECK-NEXT: [[TMP38:%.*]] = shl nsw i64 [[TMP26]], 2
+; CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP38]]
+; CHECK-NEXT: [[TMP40:%.*]] = shl nsw i64 [[TMP28]], 2
+; CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP40]]
; CHECK-NEXT: store i32 [[X:%.*]], ptr [[TMP15]], align 4
; CHECK-NEXT: store i32 [[X]], ptr [[TMP17]], align 4
; CHECK-NEXT: store i32 [[X]], ptr [[TMP19]], align 4
@@ -1396,10 +1466,13 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, 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: [[A_I:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[TMP31:%.*]] = getelementptr i32, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP43:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[A_I:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP43]]
+; CHECK-NEXT: [[I_MINUS_1:%.*]] = shl i64 [[I]], 2
+; CHECK-NEXT: [[TMP31:%.*]] = getelementptr i8, ptr [[A]], i64 [[I_MINUS_1]]
; CHECK-NEXT: [[A_I_MINUS_1:%.*]] = getelementptr i8, ptr [[TMP31]], i64 -4
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr i32, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[I_MINUS_3:%.*]] = shl i64 [[I]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr i8, ptr [[A]], i64 [[I_MINUS_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
@@ -1481,8 +1554,9 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[TMP11:%.*]] = sext <4 x i16> [[STRIDED_VEC4]] to <4 x i32>
; CHECK-NEXT: [[TMP12:%.*]] = mul nsw <4 x i32> [[TMP9]], [[TMP10]]
; CHECK-NEXT: [[TMP13:%.*]] = mul nsw <4 x i32> [[TMP12]], [[TMP11]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: store <4 x i32> [[TMP13]], ptr [[TMP14]], align 4, !alias.scope [[META30:![0-9]+]], !noalias [[META27]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP19]]
+; CHECK-NEXT: store <4 x i32> [[TMP13]], ptr [[TMP22]], align 4, !alias.scope [[META30:![0-9]+]], !noalias [[META27]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
@@ -1502,16 +1576,19 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP16]] to i32
; CHECK-NEXT: [[I1]] = add nuw nsw i64 [[I]], 1
; CHECK-NEXT: [[IV2]] = add nuw nsw i64 [[IV]], 2
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i16, ptr [[A]], i64 [[IV]]
+; CHECK-NEXT: [[IV1:%.*]] = shl i64 [[IV]], 1
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[A]], i64 [[IV1]]
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i8, ptr [[TMP18]], i64 2
; CHECK-NEXT: [[LOAD1:%.*]] = load i16, ptr [[GEP1]], align 4
; CHECK-NEXT: [[CONV1:%.*]] = sext i16 [[LOAD1]] to i32
-; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i16, ptr [[A]], i64 [[IV2]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[IV2]], 1
+; CHECK-NEXT: [[GEP2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP20]]
; CHECK-NEXT: [[LOAD2]] = load i16, ptr [[GEP2]], align 4
; CHECK-NEXT: [[CONV2:%.*]] = sext i16 [[LOAD2]] to i32
; CHECK-NEXT: [[MUL01:%.*]] = mul nsw i32 [[CONV]], [[CONV1]]
; CHECK-NEXT: [[MUL012:%.*]] = mul nsw i32 [[MUL01]], [[CONV2]]
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP21]]
; CHECK-NEXT: store i32 [[MUL012]], ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[IV]], [[N]]
; CHECK-NEXT: br i1 [[EXITCOND]], label [[END]], label [[LOOP]], !llvm.loop [[LOOP33:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll
index a02fddc4cf72d..72bfeed042cf7 100644
--- a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll
+++ b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization-2.ll
@@ -41,10 +41,11 @@ define void @inv_val_store_to_inv_address_conditional_diff_values_ic(ptr %a, i64
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; 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 <4 x i32>, ptr [[TMP1]], align 8, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP4]], align 8, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
; CHECK-NEXT: [[DOTNOT:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT4]]
-; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP1]], align 4, !alias.scope [[META0]], !noalias [[META3]]
+; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP4]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[DOTNOT]], <4 x i32> [[BROADCAST_SPLAT]], <4 x i32> [[BROADCAST_SPLAT4]]
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i32> [[PREDPHI]], i64 3
; CHECK-NEXT: store i32 [[TMP2]], ptr [[A]], align 4, !alias.scope [[META3]]
@@ -59,7 +60,8 @@ define void @inv_val_store_to_inv_address_conditional_diff_values_ic(ptr %a, i64
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[LATCH:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP6]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[I2]], [[K]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
@@ -139,8 +141,9 @@ define void @inv_val_store_to_inv_address_conditional_inv(ptr %a, i64 %n, ptr %b
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP3]], align 4, !alias.scope [[META9:![0-9]+]], !noalias [[META12:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP2]], align 4, !alias.scope [[META9:![0-9]+]], !noalias [[META12:![0-9]+]]
; CHECK-NEXT: store i32 [[K]], ptr [[A]], align 4, !alias.scope [[META12]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -153,7 +156,8 @@ define void @inv_val_store_to_inv_address_conditional_inv(ptr %a, i64 %n, ptr %b
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[LATCH:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP5]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
; CHECK-NEXT: br i1 [[CMP]], label [[COND_STORE:%.*]], label [[COND_STORE_K:%.*]]
; CHECK: cond_store:
@@ -223,8 +227,9 @@ define i32 @variant_val_store_to_inv_address(ptr %a, i64 %n, ptr %b, i32 %k) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 8, !alias.scope [[META16:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP6]], align 8, !alias.scope [[META16:![0-9]+]]
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i32> [[WIDE_LOAD]], i64 3
; CHECK-NEXT: store i32 [[TMP2]], ptr [[A]], align 4, !alias.scope [[META19:![0-9]+]], !noalias [[META16]]
; CHECK-NEXT: [[TMP3]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
@@ -243,7 +248,8 @@ define i32 @variant_val_store_to_inv_address(ptr %a, i64 %n, ptr %b, i32 %k) {
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[I0:%.*]] = phi i32 [ [[I3:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP7]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: store i32 [[I2]], ptr [[A]], align 4
; CHECK-NEXT: [[I3]] = add i32 [[I0]], [[I2]]
diff --git a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll
index eea22374ade30..8e6f029512beb 100644
--- a/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll
+++ b/llvm/test/Transforms/LoopVectorize/invariant-store-vectorization.ll
@@ -37,8 +37,9 @@ define i32 @inv_val_store_to_inv_address_with_reduction(ptr %a, i64 %n, ptr %b)
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 8, !alias.scope [[META0:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP5]], align 8, !alias.scope [[META0:![0-9]+]]
; CHECK-NEXT: [[TMP2]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -56,7 +57,8 @@ define i32 @inv_val_store_to_inv_address_with_reduction(ptr %a, i64 %n, ptr %b)
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[I0:%.*]] = phi i32 [ [[I3:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP6]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: [[I3]] = add i32 [[I0]], [[I2]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4
@@ -112,9 +114,10 @@ define void @inv_val_store_to_inv_address(ptr %a, i64 %n, ptr %b) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; 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: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP1]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4, !alias.scope [[META9:![0-9]+]], !noalias [[META12:![0-9]+]]
-; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP1]], align 4, !alias.scope [[META12]]
+; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP3]], align 4, !alias.scope [[META12]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
@@ -126,7 +129,8 @@ define void @inv_val_store_to_inv_address(ptr %a, i64 %n, ptr %b) {
; 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: [[I1:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP4]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[A]], align 4
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
@@ -189,10 +193,11 @@ define void @inv_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b, i3
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE10:%.*]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 8, !alias.scope [[META16:![0-9]+]], !noalias [[META19:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP8]], align 8, !alias.scope [[META16:![0-9]+]], !noalias [[META19:![0-9]+]]
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT4]]
-; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP1]], align 4, !alias.scope [[META16]], !noalias [[META19]]
+; CHECK-NEXT: store <4 x i32> [[BROADCAST_SPLAT]], ptr [[TMP8]], align 4, !alias.scope [[META16]], !noalias [[META19]]
; CHECK-NEXT: [[TMP3:%.*]] = extractelement <4 x i1> [[TMP2]], i64 0
; CHECK-NEXT: br i1 [[TMP3]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
; CHECK: pred.store.if:
@@ -228,7 +233,8 @@ define void @inv_val_store_to_inv_address_conditional(ptr %a, i64 %n, ptr %b, i3
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[LATCH:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i32, ptr [[B]], i64 [[I]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP9]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[I2]], [[K]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
@@ -282,7 +288,8 @@ define void @inv_val_store_to_inv_address_conditional_diff_values(ptr %a, i64 %n
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[LATCH:%.*]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i32, ptr [[B:%.*]], i64 [[I]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[I1:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[I2:%.*]] = load i32, ptr [[I1]], align 8
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[I2]], [[K:%.*]]
; CHECK-NEXT: store i32 [[NTRUNC]], ptr [[I1]], align 4
@@ -356,7 +363,8 @@ define void @multiple_uniform_stores(ptr nocapture %var1, ptr nocapture readonly
; CHECK-NEXT: [[CMP218:%.*]] = icmp ult i32 [[J_022]], [[ITR]]
; CHECK-NEXT: br i1 [[CMP218]], label [[FOR_BODY3_LR_PH:%.*]], label [[FOR_INC8]]
; CHECK: for.body3.lr.ph:
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i32, ptr [[VAR1]], i64 [[INDVARS_IV23]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nsw i64 [[INDVARS_IV23]], 2
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR1]], i64 [[TMP23]]
; CHECK-NEXT: [[TMP4:%.*]] = zext i32 [[J_022]] to i64
; CHECK-NEXT: [[ARRAYIDX5_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = xor i32 [[J_022]], -1
@@ -382,13 +390,14 @@ define void @multiple_uniform_stores(ptr nocapture %var1, ptr nocapture readonly
; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP8]], 8589934588
; CHECK-NEXT: [[IND_END:%.*]] = add nuw nsw i64 [[N_VEC]], [[TMP4]]
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x i32> <i32 poison, i32 0, i32 0, i32 0>, i32 [[ARRAYIDX5_PROMOTED]], i64 0
-; CHECK-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr i32, ptr [[VAR2]], i64 [[TMP4]]
; 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 <4 x i32> [ [[TMP15]], [[VECTOR_PH]] ], [ [[TMP17:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[INVARIANT_GEP]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[GEP]], align 4, !alias.scope [[META23:![0-9]+]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[INDEX]], [[TMP4]]
+; CHECK-NEXT: [[TMP24:%.*]] = shl nsw i64 [[OFFSET_IDX]], 2
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[VAR2]], i64 [[TMP24]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP26]], align 4, !alias.scope [[META23:![0-9]+]]
; CHECK-NEXT: [[TMP16:%.*]] = add <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP17]] = add <4 x i32> [[TMP16]], splat (i32 1)
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -407,7 +416,8 @@ define void @multiple_uniform_stores(ptr nocapture %var1, ptr nocapture readonly
; CHECK: for.body3:
; CHECK-NEXT: [[TMP20:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[TMP22:%.*]], [[FOR_BODY3]] ]
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY3]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VAR2]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[VAR2]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
; CHECK-NEXT: [[TMP22]] = add nsw i32 [[ADD]], 1
@@ -487,14 +497,16 @@ define void @multiple_uniform_stores_conditional(ptr nocapture %var1, ptr nocapt
; CHECK-NEXT: [[CMP218:%.*]] = icmp ult i32 [[J_022]], [[ITR]]
; CHECK-NEXT: br i1 [[CMP218]], label [[FOR_BODY3_LR_PH:%.*]], label [[FOR_INC8]]
; CHECK: for.body3.lr.ph:
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i32, ptr [[VAR1:%.*]], i64 [[INDVARS_IV23]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDVARS_IV23]], 2
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR1:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP0:%.*]] = zext i32 [[J_022]] to i64
; CHECK-NEXT: [[ARRAYIDX5_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: br label [[FOR_BODY3:%.*]]
; CHECK: for.body3:
; CHECK-NEXT: [[TMP1:%.*]] = phi i32 [ [[ARRAYIDX5_PROMOTED]], [[FOR_BODY3_LR_PH]] ], [ [[TMP5:%.*]], [[LATCH:%.*]] ]
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[TMP0]], [[FOR_BODY3_LR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[LATCH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[VAR2:%.*]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR2:%.*]], i64 [[TMP7]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP1]], [[TMP2]]
; CHECK-NEXT: [[TMP3:%.*]] = icmp ugt i32 [[ADD]], 42
@@ -578,7 +590,8 @@ for.end10: ; preds = %for.inc8, %entry
define void @unsafe_dep_uniform_load_store(i32 %arg, i32 %arg1, i64 %arg2, ptr %arg3, i32 %arg4, i64 %arg5) {
; CHECK-LABEL: @unsafe_dep_uniform_load_store(
; CHECK-NEXT: bb:
-; CHECK-NEXT: [[I6:%.*]] = getelementptr inbounds i16, ptr [[ARG3:%.*]], i64 [[ARG5:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[ARG5:%.*]], 1
+; CHECK-NEXT: [[I6:%.*]] = getelementptr inbounds i8, ptr [[ARG3:%.*]], i64 [[TMP0]]
; CHECK-NEXT: br label [[BB7:%.*]]
; CHECK: bb7:
; CHECK-NEXT: [[I121:%.*]] = phi i32 [ [[ARG4:%.*]], [[BB:%.*]] ], [ [[I12:%.*]], [[BB7]] ]
@@ -591,12 +604,14 @@ define void @unsafe_dep_uniform_load_store(i32 %arg, i32 %arg1, i64 %arg2, ptr %
; CHECK-NEXT: [[I15:%.*]] = trunc i64 [[I8]] to i32
; CHECK-NEXT: [[I16:%.*]] = add i32 [[ARG:%.*]], [[I15]]
; CHECK-NEXT: [[I17:%.*]] = zext i32 [[I16]] to i64
-; CHECK-NEXT: [[I18:%.*]] = getelementptr inbounds nuw i16, ptr [[I6]], i64 [[I17]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[I17]], 1
+; CHECK-NEXT: [[I18:%.*]] = getelementptr inbounds nuw i8, ptr [[I6]], i64 [[TMP1]]
; CHECK-NEXT: store i16 [[I14]], ptr [[I18]], align 2
; CHECK-NEXT: [[I19:%.*]] = add i32 [[I13]], [[I9]]
; CHECK-NEXT: [[I20:%.*]] = trunc i32 [[I19]] to i16
; CHECK-NEXT: [[I21:%.*]] = and i16 [[I20]], 255
-; CHECK-NEXT: [[I22:%.*]] = getelementptr inbounds nuw i16, ptr [[ARG3]], i64 [[I17]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[I17]], 1
+; CHECK-NEXT: [[I22:%.*]] = getelementptr inbounds nuw i8, ptr [[ARG3]], i64 [[TMP2]]
; CHECK-NEXT: store i16 [[I21]], ptr [[I22]], align 2
; CHECK-NEXT: [[I23]] = add nsw i32 [[I9]], 1
; CHECK-NEXT: [[I24]] = add nuw nsw i64 [[I8]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/loop-scalars.ll b/llvm/test/Transforms/LoopVectorize/loop-scalars.ll
index 010890e30f4eb..0de6ae346c2e9 100644
--- a/llvm/test/Transforms/LoopVectorize/loop-scalars.ll
+++ b/llvm/test/Transforms/LoopVectorize/loop-scalars.ll
@@ -17,8 +17,10 @@ define void @vector_gep(ptr %a, ptr %b, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], <2 x i64> [[VEC_IND]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds ptr, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw <2 x i64> [[VEC_IND]], splat (i64 2)
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], <2 x i64> [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; CHECK-NEXT: store <2 x ptr> [[TMP0]], ptr [[TMP1]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <2 x i64> [[VEC_IND]], splat (i64 2)
@@ -32,8 +34,10 @@ define void @vector_gep(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 nuw i32, ptr [[B]], i64 [[I]]
-; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw ptr, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP6]]
; 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]]
@@ -75,10 +79,14 @@ define void @scalar_store(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds ptr, ptr [[A]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[TMP3]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP11]]
; CHECK-NEXT: store ptr [[TMP4]], ptr [[TMP6]], align 8
; CHECK-NEXT: store ptr [[TMP5]], ptr [[TMP7]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
@@ -92,8 +100,10 @@ define void @scalar_store(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 [[I]]
-; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds ptr, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP15]]
; CHECK-NEXT: store ptr [[VAR0]], ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 2
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
@@ -135,10 +145,14 @@ define void @expansion(ptr %a, ptr %b, i64 %n) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[A:%.*]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds ptr, ptr [[A]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP11]]
; CHECK-NEXT: store ptr [[TMP4]], ptr [[TMP6]], align 8
; CHECK-NEXT: store ptr [[TMP5]], ptr [[TMP7]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
@@ -152,8 +166,10 @@ define void @expansion(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 i64, ptr [[B]], i64 [[I]]
-; CHECK-NEXT: [[VAR3:%.*]] = getelementptr inbounds ptr, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[VAR3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP15]]
; CHECK-NEXT: store ptr [[VAR0]], ptr [[VAR3]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 2
; CHECK-NEXT: [[COND:%.*]] = icmp slt i64 [[I_NEXT]], [[N]]
@@ -189,7 +205,8 @@ define void @no_gep_or_bitcast(ptr noalias %a, 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 ptr, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x ptr>, ptr [[TMP0]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x ptr> [[WIDE_LOAD]], i64 0
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x ptr> [[WIDE_LOAD]], i64 1
@@ -206,7 +223,8 @@ define void @no_gep_or_bitcast(ptr noalias %a, 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 nuw ptr, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT: [[VAR1:%.*]] = load ptr, ptr [[VAR0]], align 8
; CHECK-NEXT: store i32 0, ptr [[VAR1]], align 8
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/non-const-n.ll b/llvm/test/Transforms/LoopVectorize/non-const-n.ll
index 30f76f8734145..93fe42e3388dd 100644
--- a/llvm/test/Transforms/LoopVectorize/non-const-n.ll
+++ b/llvm/test/Transforms/LoopVectorize/non-const-n.ll
@@ -19,12 +19,15 @@ define void @example1(i32 %n) nounwind uwtable ssp {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr @b, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr @b, i64 [[TMP8]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr @c, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr @c, i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr @a, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr @a, i64 [[TMP7]]
; CHECK-NEXT: store <4 x i32> [[TMP4]], ptr [[TMP5]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX]], [[TMP1]]
diff --git a/llvm/test/Transforms/LoopVectorize/phi-cost.ll b/llvm/test/Transforms/LoopVectorize/phi-cost.ll
index 7b5d0b69639fa..39b5a11f087ab 100644
--- a/llvm/test/Transforms/LoopVectorize/phi-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/phi-cost.ll
@@ -20,9 +20,11 @@ define void @phi_two_incoming_values(ptr noalias %a, ptr noalias %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: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP1]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP2]], align 4
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP7]]
; CHECK-NEXT: [[TMP4:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP5:%.*]] = zext <2 x i1> [[TMP4]] to <2 x i32>
; CHECK-NEXT: [[PREDPHI:%.*]] = add <2 x i32> [[WIDE_LOAD]], [[TMP5]]
@@ -38,17 +40,19 @@ define void @phi_two_incoming_values(ptr noalias %a, ptr noalias %b, i64 %n) {
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[IF_END:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[I]]
-; CHECK-NEXT: [[TMP4:%.*]] = icmp sgt i32 [[TMP2]], 0
-; CHECK-NEXT: br i1 [[TMP4]], label %[[IF_THEN:.*]], label %[[IF_END]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP7:%.*]] = icmp sgt i32 [[TMP1]], 0
+; CHECK-NEXT: br i1 [[TMP7]], label %[[IF_THEN:.*]], label %[[IF_END]]
; CHECK: [[IF_THEN]]:
-; CHECK-NEXT: [[TMP5:%.*]] = add nuw i32 [[TMP2]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = add nuw i32 [[TMP1]], 1
; CHECK-NEXT: br label %[[IF_END]]
; CHECK: [[IF_END]]:
-; CHECK-NEXT: [[TMP6:%.*]] = phi i32 [ [[TMP2]], %[[FOR_BODY]] ], [ [[TMP5]], %[[IF_THEN]] ]
-; CHECK-NEXT: store i32 [[TMP6]], ptr [[TMP3]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = phi i32 [ [[TMP1]], %[[FOR_BODY]] ], [ [[TMP3]], %[[IF_THEN]] ]
+; CHECK-NEXT: store i32 [[TMP4]], ptr [[TMP2]], align 4
; CHECK-NEXT: [[I_NEXT]] = add i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp eq i64 [[I]], [[N]]
; CHECK-NEXT: br i1 [[COND]], label %[[FOR_END]], label %[[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
@@ -97,9 +101,11 @@ define void @phi_three_incoming_values(ptr noalias %a, ptr noalias %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: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP1]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP3]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <2 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD]], [[WIDE_LOAD2]]
; CHECK-NEXT: [[TMP6:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD]], splat (i32 19)
@@ -119,22 +125,24 @@ define void @phi_three_incoming_values(ptr noalias %a, ptr noalias %b, i64 %n) {
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[IF_END:.*]] ], [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I]]
-; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[I]]
-; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
-; CHECK-NEXT: [[TMP6:%.*]] = icmp sgt i32 [[TMP3]], [[TMP5]]
-; CHECK-NEXT: br i1 [[TMP6]], label %[[IF_THEN:.*]], label %[[IF_END]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP3]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = icmp sgt i32 [[TMP1]], [[TMP12]]
+; CHECK-NEXT: br i1 [[TMP4]], label %[[IF_THEN:.*]], label %[[IF_END]]
; CHECK: [[IF_THEN]]:
-; CHECK-NEXT: [[TMP7:%.*]] = icmp sgt i32 [[TMP3]], 19
-; CHECK-NEXT: br i1 [[TMP7]], label %[[IF_END]], label %[[IF_ELSE:.*]]
+; CHECK-NEXT: [[TMP5:%.*]] = icmp sgt i32 [[TMP1]], 19
+; CHECK-NEXT: br i1 [[TMP5]], label %[[IF_END]], label %[[IF_ELSE:.*]]
; CHECK: [[IF_ELSE]]:
-; CHECK-NEXT: [[TMP8:%.*]] = icmp slt i32 [[TMP5]], 4
-; CHECK-NEXT: [[TMP9:%.*]] = select i1 [[TMP8]], i32 4, i32 5
+; CHECK-NEXT: [[TMP6:%.*]] = icmp slt i32 [[TMP12]], 4
+; CHECK-NEXT: [[TMP7:%.*]] = select i1 [[TMP6]], i32 4, i32 5
; CHECK-NEXT: br label %[[IF_END]]
; CHECK: [[IF_END]]:
-; CHECK-NEXT: [[TMP10:%.*]] = phi i32 [ 9, %[[FOR_BODY]] ], [ 3, %[[IF_THEN]] ], [ [[TMP9]], %[[IF_ELSE]] ]
-; CHECK-NEXT: store i32 [[TMP10]], ptr [[TMP0]], align 4
+; CHECK-NEXT: [[TMP8:%.*]] = phi i32 [ 9, %[[FOR_BODY]] ], [ 3, %[[IF_THEN]] ], [ [[TMP7]], %[[IF_ELSE]] ]
+; CHECK-NEXT: store i32 [[TMP8]], ptr [[TMP0]], align 4
; CHECK-NEXT: [[I_NEXT]] = add i64 [[I]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp eq i64 [[I]], [[N]]
; CHECK-NEXT: br i1 [[COND]], label %[[FOR_END]], label %[[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/reduction-inloop-cond.ll b/llvm/test/Transforms/LoopVectorize/reduction-inloop-cond.ll
index 481095225af6a..3185762dcfe58 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction-inloop-cond.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction-inloop-cond.ll
@@ -13,13 +13,15 @@ define float @cond_fadd(ptr noalias nocapture readonly %a, ptr noalias nocapture
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE6:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi float [ 1.000000e+00, [[VECTOR_PH]] ], [ [[TMP26:%.*]], [[PRED_LOAD_CONTINUE6]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP30]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = fcmp une <4 x float> [[WIDE_LOAD]], splat (float 5.000000e+00)
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; CHECK-NEXT: br i1 [[TMP2]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP31]]
; CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x float> poison, float [[TMP4]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -28,7 +30,8 @@ define float @cond_fadd(ptr noalias nocapture readonly %a, ptr noalias nocapture
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
; CHECK-NEXT: br i1 [[TMP7]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP32:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP32]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 4
; CHECK-NEXT: [[TMP10:%.*]] = load float, ptr [[TMP9]], align 4
; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x float> [[TMP6]], float [[TMP10]], i64 1
@@ -38,7 +41,8 @@ define float @cond_fadd(ptr noalias nocapture readonly %a, ptr noalias nocapture
; CHECK-NEXT: [[TMP13:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
; CHECK-NEXT: br i1 [[TMP13]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP34]]
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[TMP14]], i64 8
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x float> [[TMP12]], float [[TMP16]], i64 2
@@ -48,7 +52,8 @@ define float @cond_fadd(ptr noalias nocapture readonly %a, ptr noalias nocapture
; CHECK-NEXT: [[TMP19:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
; CHECK-NEXT: br i1 [[TMP19]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP36]]
; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP20]], i64 12
; CHECK-NEXT: [[TMP22:%.*]] = load float, ptr [[TMP21]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x float> [[TMP18]], float [[TMP22]], i64 3
@@ -70,12 +75,14 @@ define float @cond_fadd(ptr noalias nocapture readonly %a, ptr noalias nocapture
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-NEXT: [[RDX:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[RES:%.*]], [[FOR_INC]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP33]]
; CHECK-NEXT: [[TMP28:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP28]], 5.000000e+00
; CHECK-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP29:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[FADD:%.*]] = fadd fast float [[RDX]], [[TMP29]]
; CHECK-NEXT: br label [[FOR_INC]]
@@ -127,13 +134,15 @@ define float @cond_cmp_sel(ptr noalias %a, ptr noalias %cond, i64 %N) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE6:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi float [ 1.000000e+00, [[VECTOR_PH]] ], [ [[RDX_MINMAX_SELECT:%.*]], [[PRED_LOAD_CONTINUE6]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds float, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP30]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = fcmp une <4 x float> [[WIDE_LOAD]], splat (float 3.000000e+00)
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; CHECK-NEXT: br i1 [[TMP2]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP31]]
; CHECK-NEXT: [[TMP4:%.*]] = load float, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x float> poison, float [[TMP4]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -142,7 +151,8 @@ define float @cond_cmp_sel(ptr noalias %a, ptr noalias %cond, i64 %N) {
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
; CHECK-NEXT: br i1 [[TMP7]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP32:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP32]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 4
; CHECK-NEXT: [[TMP10:%.*]] = load float, ptr [[TMP9]], align 4
; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x float> [[TMP6]], float [[TMP10]], i64 1
@@ -152,7 +162,8 @@ define float @cond_cmp_sel(ptr noalias %a, ptr noalias %cond, i64 %N) {
; CHECK-NEXT: [[TMP13:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
; CHECK-NEXT: br i1 [[TMP13]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP34]]
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[TMP14]], i64 8
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x float> [[TMP12]], float [[TMP16]], i64 2
@@ -162,7 +173,8 @@ define float @cond_cmp_sel(ptr noalias %a, ptr noalias %cond, i64 %N) {
; CHECK-NEXT: [[TMP19:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
; CHECK-NEXT: br i1 [[TMP19]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP36]]
; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP20]], i64 12
; CHECK-NEXT: [[TMP22:%.*]] = load float, ptr [[TMP21]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x float> [[TMP18]], float [[TMP22]], i64 3
@@ -185,12 +197,14 @@ define float @cond_cmp_sel(ptr noalias %a, ptr noalias %cond, i64 %N) {
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-NEXT: [[RDX:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[RES:%.*]], [[FOR_INC]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP33]]
; CHECK-NEXT: [[TMP28:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP28]], 3.000000e+00
; CHECK-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP29:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[FSEL:%.*]] = call fast float @llvm.minnum.f32(float [[RDX]], float [[TMP29]])
; CHECK-NEXT: br label [[FOR_INC]]
@@ -245,13 +259,15 @@ define i32 @conditional_and(ptr noalias %A, ptr noalias %B, i32 %cond, i64 nound
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE6:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 7, [[VECTOR_PH]] ], [ [[TMP27:%.*]], [[PRED_LOAD_CONTINUE6]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP31]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; CHECK-NEXT: br i1 [[TMP2]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP32:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP32]]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -260,7 +276,8 @@ define i32 @conditional_and(ptr noalias %A, ptr noalias %B, i32 %cond, i64 nound
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
; CHECK-NEXT: br i1 [[TMP7]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP33]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 4
; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> [[TMP6]], i32 [[TMP10]], i64 1
@@ -270,7 +287,8 @@ define i32 @conditional_and(ptr noalias %A, ptr noalias %B, i32 %cond, i64 nound
; CHECK-NEXT: [[TMP13:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
; CHECK-NEXT: br i1 [[TMP13]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[TMP14]], i64 8
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP12]], i32 [[TMP16]], i64 2
@@ -280,7 +298,8 @@ define i32 @conditional_and(ptr noalias %A, ptr noalias %B, i32 %cond, i64 nound
; CHECK-NEXT: [[TMP19:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
; CHECK-NEXT: br i1 [[TMP19]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP37]]
; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP20]], i64 12
; CHECK-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP22]], i64 3
@@ -303,12 +322,14 @@ define i32 @conditional_and(ptr noalias %A, ptr noalias %B, i32 %cond, i64 nound
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-NEXT: [[RDX:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[RES:%.*]], [[FOR_INC]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP34]]
; CHECK-NEXT: [[TMP29:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TOBOOL:%.*]] = icmp eq i32 [[TMP29]], [[COND]]
; CHECK-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP36]]
; CHECK-NEXT: [[TMP30:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[AND:%.*]] = and i32 [[TMP30]], [[RDX]]
; CHECK-NEXT: br label [[FOR_INC]]
@@ -360,16 +381,19 @@ define i32 @simple_chained_rdx(ptr noalias %a, ptr noalias %b, ptr noalias %cond
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE6:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 5, [[VECTOR_PH]] ], [ [[TMP46:%.*]], [[PRED_LOAD_CONTINUE6]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP51]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ne <4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; CHECK-NEXT: br i1 [[TMP2]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP53]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <4 x i32> poison, i32 [[TMP7]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -380,10 +404,12 @@ define i32 @simple_chained_rdx(ptr noalias %a, ptr noalias %b, ptr noalias %cond
; CHECK-NEXT: br i1 [[TMP11]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP12:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP12]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP14]], i64 1
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP55:%.*]] = shl nsw i64 [[TMP12]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP55]]
; CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
; CHECK-NEXT: [[TMP18:%.*]] = insertelement <4 x i32> [[TMP10]], i32 [[TMP17]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -394,10 +420,12 @@ define i32 @simple_chained_rdx(ptr noalias %a, ptr noalias %b, ptr noalias %cond
; CHECK-NEXT: br i1 [[TMP21]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP22:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl nsw i64 [[TMP22]], 2
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP56]]
; CHECK-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP24]], i64 2
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP58:%.*]] = shl nsw i64 [[TMP22]], 2
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP58]]
; CHECK-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4
; CHECK-NEXT: [[TMP28:%.*]] = insertelement <4 x i32> [[TMP20]], i32 [[TMP27]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -408,10 +436,12 @@ define i32 @simple_chained_rdx(ptr noalias %a, ptr noalias %b, ptr noalias %cond
; CHECK-NEXT: br i1 [[TMP31]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP32:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP32]]
+; CHECK-NEXT: [[TMP60:%.*]] = shl nsw i64 [[TMP32]], 2
+; CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP60]]
; CHECK-NEXT: [[TMP34:%.*]] = load i32, ptr [[TMP33]], align 4
; CHECK-NEXT: [[TMP35:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP34]], i64 3
-; CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP32]]
+; CHECK-NEXT: [[TMP62:%.*]] = shl nsw i64 [[TMP32]], 2
+; CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP62]]
; CHECK-NEXT: [[TMP37:%.*]] = load i32, ptr [[TMP36]], align 4
; CHECK-NEXT: [[TMP38:%.*]] = insertelement <4 x i32> [[TMP30]], i32 [[TMP37]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -437,15 +467,18 @@ define i32 @simple_chained_rdx(ptr noalias %a, ptr noalias %b, ptr noalias %cond
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[RDX:%.*]] = phi i32 [ [[RES:%.*]], [[FOR_INC]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP57:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP57]]
; CHECK-NEXT: [[TMP48:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP48]], 0
; CHECK-NEXT: br i1 [[TOBOOL_NOT]], label [[FOR_INC]], label [[IF_THEN:%.*]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; CHECK-NEXT: [[TMP59:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP59]]
; CHECK-NEXT: [[TMP49:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP49]], [[RDX]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; CHECK-NEXT: [[TMP61:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP61]]
; CHECK-NEXT: [[TMP50:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD3:%.*]] = add nsw i32 [[ADD]], [[TMP50]]
; CHECK-NEXT: br label [[FOR_INC]]
@@ -510,13 +543,15 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP1:%.*]] = or disjoint i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP2:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP47]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x i64> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <4 x i1> [[TMP4]], i64 0
; CHECK-NEXT: br i1 [[TMP5]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP48]]
; CHECK-NEXT: [[TMP7:%.*]] = load i64, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <4 x i64> poison, i64 [[TMP7]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -525,7 +560,8 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x i1> [[TMP4]], i64 1
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP0]], 3
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP12:%.*]] = load i64, ptr [[TMP11]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i64> [[TMP9]], i64 [[TMP12]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -534,7 +570,8 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP4]], i64 2
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP55:%.*]] = shl nsw i64 [[TMP1]], 3
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP55]]
; CHECK-NEXT: [[TMP17:%.*]] = load i64, ptr [[TMP16]], align 4
; CHECK-NEXT: [[TMP18:%.*]] = insertelement <4 x i64> [[TMP14]], i64 [[TMP17]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -543,7 +580,8 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP20:%.*]] = extractelement <4 x i1> [[TMP4]], i64 3
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6:%.*]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP56]]
; CHECK-NEXT: [[TMP22:%.*]] = load i64, ptr [[TMP21]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP22]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -554,7 +592,8 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP27:%.*]] = extractelement <4 x i1> [[TMP26]], i64 0
; CHECK-NEXT: br i1 [[TMP27]], label [[PRED_LOAD_IF7:%.*]], label [[PRED_LOAD_CONTINUE8:%.*]]
; CHECK: pred.load.if7:
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP57:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP57]]
; CHECK-NEXT: [[TMP29:%.*]] = load i64, ptr [[TMP28]], align 4
; CHECK-NEXT: [[TMP30:%.*]] = insertelement <4 x i64> poison, i64 [[TMP29]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE8]]
@@ -563,7 +602,8 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP32:%.*]] = extractelement <4 x i1> [[TMP26]], i64 1
; CHECK-NEXT: br i1 [[TMP32]], label [[PRED_LOAD_IF9:%.*]], label [[PRED_LOAD_CONTINUE10:%.*]]
; CHECK: pred.load.if9:
-; CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP59:%.*]] = shl nsw i64 [[TMP0]], 3
+; CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP59]]
; CHECK-NEXT: [[TMP34:%.*]] = load i64, ptr [[TMP33]], align 4
; CHECK-NEXT: [[TMP35:%.*]] = insertelement <4 x i64> [[TMP31]], i64 [[TMP34]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE10]]
@@ -572,7 +612,8 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP37:%.*]] = extractelement <4 x i1> [[TMP26]], i64 2
; CHECK-NEXT: br i1 [[TMP37]], label [[PRED_LOAD_IF11:%.*]], label [[PRED_LOAD_CONTINUE12:%.*]]
; CHECK: pred.load.if11:
-; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP61:%.*]] = shl nsw i64 [[TMP1]], 3
+; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP61]]
; CHECK-NEXT: [[TMP39:%.*]] = load i64, ptr [[TMP38]], align 4
; CHECK-NEXT: [[TMP40:%.*]] = insertelement <4 x i64> [[TMP36]], i64 [[TMP39]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE12]]
@@ -581,7 +622,8 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK-NEXT: [[TMP42:%.*]] = extractelement <4 x i1> [[TMP26]], i64 3
; CHECK-NEXT: br i1 [[TMP42]], label [[PRED_LOAD_IF13:%.*]], label [[PRED_LOAD_CONTINUE14]]
; CHECK: pred.load.if13:
-; CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP63:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP63]]
; CHECK-NEXT: [[TMP44:%.*]] = load i64, ptr [[TMP43]], align 4
; CHECK-NEXT: [[TMP45:%.*]] = insertelement <4 x i64> [[TMP41]], i64 [[TMP44]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE14]]
@@ -604,18 +646,21 @@ define i64 @nested_cond_and(ptr noalias nocapture readonly %a, ptr noalias nocap
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-NEXT: [[RDX:%.*]] = phi i64 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[RES:%.*]], [[FOR_INC]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP58:%.*]] = shl nsw i64 [[IV]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP58]]
; CHECK-NEXT: [[TMP51:%.*]] = load i64, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TOBOOL:%.*]] = icmp eq i64 [[TMP51]], 0
; CHECK-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[IV]]
+; CHECK-NEXT: [[TMP60:%.*]] = shl nsw i64 [[IV]], 3
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP60]]
; CHECK-NEXT: [[TMP52:%.*]] = load i64, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[AND1:%.*]] = and i64 [[RDX]], [[TMP52]]
; CHECK-NEXT: [[TOBOOL2:%.*]] = icmp eq i64 [[TMP52]], 3
; CHECK-NEXT: br i1 [[TOBOOL2]], label [[IF_THEN_2:%.*]], label [[FOR_INC]]
; CHECK: if.then.2:
-; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; CHECK-NEXT: [[TMP62:%.*]] = shl nsw i64 [[IV]], 3
+; CHECK-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP62]]
; CHECK-NEXT: [[TMP53:%.*]] = load i64, ptr [[ARRAYIDX3]], align 4
; CHECK-NEXT: [[AND2:%.*]] = and i64 [[RDX]], [[TMP53]]
; CHECK-NEXT: br label [[FOR_INC]]
@@ -678,13 +723,15 @@ define i32 @cond-uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE6:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP27:%.*]], [[PRED_LOAD_CONTINUE6]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP33]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = icmp ne <4 x i32> [[WIDE_LOAD]], zeroinitializer
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; CHECK-NEXT: br i1 [[TMP2]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP34]]
; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -693,7 +740,8 @@ define i32 @cond-uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP7:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
; CHECK-NEXT: br i1 [[TMP7]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 4
; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> [[TMP6]], i32 [[TMP10]], i64 1
@@ -703,7 +751,8 @@ define i32 @cond-uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP13:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
; CHECK-NEXT: br i1 [[TMP13]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP37]]
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[TMP14]], i64 8
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP12]], i32 [[TMP16]], i64 2
@@ -713,7 +762,8 @@ define i32 @cond-uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP19:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
; CHECK-NEXT: br i1 [[TMP19]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP39:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP39]]
; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP20]], i64 12
; CHECK-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP22]], i64 3
@@ -722,7 +772,8 @@ define i32 @cond-uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP24:%.*]] = phi <4 x i32> [ [[TMP18]], [[PRED_LOAD_CONTINUE4]] ], [ [[TMP23]], [[PRED_LOAD_IF5]] ]
; CHECK-NEXT: [[TMP25:%.*]] = select <4 x i1> [[TMP1]], <4 x i32> [[TMP24]], <4 x i32> zeroinitializer
; CHECK-NEXT: [[PREDPHI:%.*]] = add <4 x i32> [[VEC_PHI]], [[TMP25]]
-; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP41:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP41]]
; CHECK-NEXT: [[WIDE_LOAD7:%.*]] = load <4 x i32>, ptr [[TMP26]], align 4
; CHECK-NEXT: [[TMP27]] = add <4 x i32> [[WIDE_LOAD7]], [[PREDPHI]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -739,18 +790,21 @@ define i32 @cond-uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK: for.body:
; CHECK-NEXT: [[RDX1:%.*]] = phi i32 [ [[ADD2:%.*]], [[IF_END:%.*]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[IF_END]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP36]]
; CHECK-NEXT: [[TMP30:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP30]], 0
; CHECK-NEXT: br i1 [[TOBOOL_NOT]], label [[IF_END]], label [[IF_THEN:%.*]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[SRC2]], i64 [[IV]]
+; CHECK-NEXT: [[TMP38:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP38]]
; CHECK-NEXT: [[TMP31:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP31]], [[RDX1]]
; CHECK-NEXT: br label [[IF_END]]
; CHECK: if.end:
; CHECK-NEXT: [[RES:%.*]] = phi i32 [ [[ADD]], [[IF_THEN]] ], [ [[RDX1]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 [[IV]]
+; CHECK-NEXT: [[TMP40:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP40]]
; CHECK-NEXT: [[TMP32:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD2]] = add nsw i32 [[TMP32]], [[RES]]
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
@@ -809,13 +863,15 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 1
; CHECK-NEXT: [[TMP1:%.*]] = or disjoint i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP2:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds float, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP53]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = fcmp fast oeq <4 x float> [[WIDE_LOAD]], splat (float 3.000000e+00)
; CHECK-NEXT: [[TMP5:%.*]] = extractelement <4 x i1> [[TMP4]], i64 0
; CHECK-NEXT: br i1 [[TMP5]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[SRC1]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP7:%.*]] = load float, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <4 x float> poison, float [[TMP7]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -824,7 +880,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP10:%.*]] = extractelement <4 x i1> [[TMP4]], i64 1
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[SRC1]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP55:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP55]]
; CHECK-NEXT: [[TMP12:%.*]] = load float, ptr [[TMP11]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x float> [[TMP9]], float [[TMP12]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -833,7 +890,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP4]], i64 2
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[SRC1]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP56]]
; CHECK-NEXT: [[TMP17:%.*]] = load float, ptr [[TMP16]], align 4
; CHECK-NEXT: [[TMP18:%.*]] = insertelement <4 x float> [[TMP14]], float [[TMP17]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -842,7 +900,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP20:%.*]] = extractelement <4 x i1> [[TMP4]], i64 3
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6:%.*]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds float, ptr [[SRC1]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP57:%.*]] = shl nsw i64 [[TMP2]], 2
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP57]]
; CHECK-NEXT: [[TMP22:%.*]] = load float, ptr [[TMP21]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x float> [[TMP19]], float [[TMP22]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -854,7 +913,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP27:%.*]] = extractelement <4 x i1> [[TMP26]], i64 0
; CHECK-NEXT: br i1 [[TMP27]], label [[PRED_LOAD_IF7:%.*]], label [[PRED_LOAD_CONTINUE8:%.*]]
; CHECK: pred.load.if7:
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP58:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP58]]
; CHECK-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP28]], align 4
; CHECK-NEXT: [[TMP30:%.*]] = insertelement <4 x float> poison, float [[TMP29]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE8]]
@@ -863,7 +923,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP32:%.*]] = extractelement <4 x i1> [[TMP26]], i64 1
; CHECK-NEXT: br i1 [[TMP32]], label [[PRED_LOAD_IF9:%.*]], label [[PRED_LOAD_CONTINUE10:%.*]]
; CHECK: pred.load.if9:
-; CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds float, ptr [[SRC2]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP60:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP60]]
; CHECK-NEXT: [[TMP34:%.*]] = load float, ptr [[TMP33]], align 4
; CHECK-NEXT: [[TMP35:%.*]] = insertelement <4 x float> [[TMP31]], float [[TMP34]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE10]]
@@ -872,7 +933,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP37:%.*]] = extractelement <4 x i1> [[TMP26]], i64 2
; CHECK-NEXT: br i1 [[TMP37]], label [[PRED_LOAD_IF11:%.*]], label [[PRED_LOAD_CONTINUE12:%.*]]
; CHECK: pred.load.if11:
-; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds float, ptr [[SRC2]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP62:%.*]] = shl nsw i64 [[TMP1]], 2
+; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP62]]
; CHECK-NEXT: [[TMP39:%.*]] = load float, ptr [[TMP38]], align 4
; CHECK-NEXT: [[TMP40:%.*]] = insertelement <4 x float> [[TMP36]], float [[TMP39]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE12]]
@@ -881,7 +943,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP42:%.*]] = extractelement <4 x i1> [[TMP26]], i64 3
; CHECK-NEXT: br i1 [[TMP42]], label [[PRED_LOAD_IF13:%.*]], label [[PRED_LOAD_CONTINUE14]]
; CHECK: pred.load.if13:
-; CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds float, ptr [[SRC2]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP64:%.*]] = shl nsw i64 [[TMP2]], 2
+; CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP64]]
; CHECK-NEXT: [[TMP44:%.*]] = load float, ptr [[TMP43]], align 4
; CHECK-NEXT: [[TMP45:%.*]] = insertelement <4 x float> [[TMP41]], float [[TMP44]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE14]]
@@ -903,12 +966,14 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK: for.body:
; CHECK-NEXT: [[RDX1:%.*]] = phi float [ [[RES:%.*]], [[FOR_INC:%.*]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP59:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP59]]
; CHECK-NEXT: [[TMP50:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[CMP1:%.*]] = fcmp fast oeq float [[TMP50]], 3.000000e+00
; CHECK-NEXT: br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[SRC1]], i64 [[IV]]
+; CHECK-NEXT: [[TMP61:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP61]]
; CHECK-NEXT: [[TMP51:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD:%.*]] = fadd fast float [[TMP51]], [[RDX1]]
; CHECK-NEXT: br label [[IF_END]]
@@ -917,7 +982,8 @@ define float @cond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[CMP5:%.*]] = fcmp fast oeq float [[TMP50]], 7.000000e+00
; CHECK-NEXT: br i1 [[CMP5]], label [[IF_THEN6:%.*]], label [[FOR_INC]]
; CHECK: if.then6:
-; CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds float, ptr [[SRC2]], i64 [[IV]]
+; CHECK-NEXT: [[TMP63:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP63]]
; CHECK-NEXT: [[TMP52:%.*]] = load float, ptr [[ARRAYIDX7]], align 4
; CHECK-NEXT: [[ADD2:%.*]] = fadd fast float [[TMP52]], [[RDX2]]
; CHECK-NEXT: br label [[FOR_INC]]
@@ -984,16 +1050,19 @@ define i32 @uncond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE7:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PREDPHI:%.*]], [[PRED_LOAD_CONTINUE7]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP33]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = add <4 x i32> [[WIDE_LOAD]], [[VEC_PHI]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP34]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne <4 x i32> [[WIDE_LOAD1]], zeroinitializer
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i1> [[TMP3]], i64 0
; CHECK-NEXT: br i1 [[TMP4]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -1002,7 +1071,8 @@ define i32 @uncond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP3]], i64 1
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_LOAD_IF2:%.*]], label [[PRED_LOAD_CONTINUE3:%.*]]
; CHECK: pred.load.if2:
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP37]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP10]], i64 4
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP12]], i64 1
@@ -1012,7 +1082,8 @@ define i32 @uncond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP3]], i64 2
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_LOAD_IF4:%.*]], label [[PRED_LOAD_CONTINUE5:%.*]]
; CHECK: pred.load.if4:
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP39:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP39]]
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP16]], i64 8
; CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x i32> [[TMP14]], i32 [[TMP18]], i64 2
@@ -1022,7 +1093,8 @@ define i32 @uncond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK-NEXT: [[TMP21:%.*]] = extractelement <4 x i1> [[TMP3]], i64 3
; CHECK-NEXT: br i1 [[TMP21]], label [[PRED_LOAD_IF6:%.*]], label [[PRED_LOAD_CONTINUE7]]
; CHECK: pred.load.if6:
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP41:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP41]]
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[TMP22]], i64 12
; CHECK-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x i32> [[TMP20]], i32 [[TMP24]], i64 3
@@ -1045,15 +1117,18 @@ define i32 @uncond_cond(ptr noalias %src1, ptr noalias %src2, ptr noalias %cond,
; CHECK: for.body:
; CHECK-NEXT: [[RDX:%.*]] = phi i32 [ [[RES:%.*]], [[FOR_INC:%.*]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[FOR_INC]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 [[IV]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP36]]
; CHECK-NEXT: [[TMP30:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP30]], [[RDX]]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP38:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP38]]
; CHECK-NEXT: [[TMP31:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP31]], 0
; CHECK-NEXT: br i1 [[TOBOOL_NOT]], label [[FOR_INC]], label [[IF_THEN:%.*]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[SRC2]], i64 [[IV]]
+; CHECK-NEXT: [[TMP40:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP40]]
; CHECK-NEXT: [[TMP32:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP32]], [[ADD1]]
; CHECK-NEXT: br label [[FOR_INC]]
@@ -1112,16 +1187,19 @@ define i32 @uncond_cond_uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_LOAD_CONTINUE7:%.*]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP28:%.*]], [[PRED_LOAD_CONTINUE7]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP34]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[TMP1:%.*]] = add <4 x i32> [[WIDE_LOAD]], [[VEC_PHI]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP35]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = icmp ne <4 x i32> [[WIDE_LOAD1]], zeroinitializer
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i1> [[TMP3]], i64 0
; CHECK-NEXT: br i1 [[TMP4]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP36:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP36]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -1130,7 +1208,8 @@ define i32 @uncond_cond_uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP3]], i64 1
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_LOAD_IF2:%.*]], label [[PRED_LOAD_CONTINUE3:%.*]]
; CHECK: pred.load.if2:
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP38:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP38]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP10]], i64 4
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP12]], i64 1
@@ -1140,7 +1219,8 @@ define i32 @uncond_cond_uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP3]], i64 2
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_LOAD_IF4:%.*]], label [[PRED_LOAD_CONTINUE5:%.*]]
; CHECK: pred.load.if4:
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP40:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP40]]
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP16]], i64 8
; CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x i32> [[TMP14]], i32 [[TMP18]], i64 2
@@ -1150,7 +1230,8 @@ define i32 @uncond_cond_uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias
; CHECK-NEXT: [[TMP21:%.*]] = extractelement <4 x i1> [[TMP3]], i64 3
; CHECK-NEXT: br i1 [[TMP21]], label [[PRED_LOAD_IF6:%.*]], label [[PRED_LOAD_CONTINUE7]]
; CHECK: pred.load.if6:
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i32, ptr [[SRC2]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP42:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i8, ptr [[SRC2]], i64 [[TMP42]]
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[TMP22]], i64 12
; CHECK-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x i32> [[TMP20]], i32 [[TMP24]], i64 3
@@ -1174,15 +1255,18 @@ define i32 @uncond_cond_uncond(ptr noalias %src1, ptr noalias %src2, ptr noalias
; CHECK: for.body:
; CHECK-NEXT: [[RDX:%.*]] = phi i32 [ [[ADD3:%.*]], [[IF_END:%.*]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[IF_END]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[SRC1]], i64 [[IV]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[SRC1]], i64 [[TMP37]]
; CHECK-NEXT: [[TMP31:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP31]], [[RDX]]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[COND]], i64 [[IV]]
+; CHECK-NEXT: [[TMP39:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[COND]], i64 [[TMP39]]
; CHECK-NEXT: [[TMP32:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[TMP32]], 0
; CHECK-NEXT: br i1 [[TOBOOL_NOT]], label [[IF_END]], label [[IF_THEN:%.*]]
; CHECK: if.then:
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[SRC2]], i64 [[IV]]
+; CHECK-NEXT: [[TMP41:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[SRC2]], i64 [[TMP41]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP33]], [[ADD1]]
; CHECK-NEXT: br label [[IF_END]]
diff --git a/llvm/test/Transforms/LoopVectorize/reduction-inloop-min-max.ll b/llvm/test/Transforms/LoopVectorize/reduction-inloop-min-max.ll
index 73ddddc69a7c7..f962b5abf780b 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction-inloop-min-max.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction-inloop-min-max.ll
@@ -11,13 +11,14 @@ define i32 @reduction_smin(ptr nocapture %A, ptr nocapture %B) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 1000, [[VECTOR_PH]] ], [ [[RDX_MINMAX:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[WIDE_LOAD]])
-; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.smin.i32(i32 [[TMP1]], i32 [[VEC_PHI]])
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[WIDE_LOAD]])
+; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.smin.i32(i32 [[TMP2]], i32 [[VEC_PHI]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end:
@@ -53,13 +54,14 @@ define i32 @reduction_smin_select_ops_flipped(ptr nocapture %A, ptr nocapture %B
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 1000, [[VECTOR_PH]] ], [ [[RDX_MINMAX:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> [[WIDE_LOAD]])
-; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.smax.i32(i32 [[TMP1]], i32 [[VEC_PHI]])
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> [[WIDE_LOAD]])
+; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.smax.i32(i32 [[TMP2]], i32 [[VEC_PHI]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end:
@@ -94,18 +96,19 @@ define i32 @reduction_smin_intrinsic(ptr nocapture %A, ptr nocapture %B) {
; 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 <4 x i32> [ splat (i32 1000), [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[WIDE_LOAD]])
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1000), [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[WIDE_LOAD]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end:
-; CHECK-NEXT: [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[TMP1]])
-; CHECK-NEXT: ret i32 [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[TMP2]])
+; CHECK-NEXT: ret i32 [[TMP4]]
;
entry:
br label %for.body
@@ -138,13 +141,14 @@ define i32 @reduction_umax(ptr nocapture %A, ptr nocapture %B) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 1000, [[VECTOR_PH]] ], [ [[RDX_MINMAX:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> [[WIDE_LOAD]])
-; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.umax.i32(i32 [[TMP1]], i32 [[VEC_PHI]])
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> [[WIDE_LOAD]])
+; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.umax.i32(i32 [[TMP2]], i32 [[VEC_PHI]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end:
@@ -180,13 +184,14 @@ define i32 @reduction_umax_select_ops_flipped(ptr nocapture %A, ptr nocapture %B
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi i32 [ 1000, [[VECTOR_PH]] ], [ [[RDX_MINMAX:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> [[WIDE_LOAD]])
-; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.umin.i32(i32 [[TMP1]], i32 [[VEC_PHI]])
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> [[WIDE_LOAD]])
+; CHECK-NEXT: [[RDX_MINMAX]] = call i32 @llvm.umin.i32(i32 [[TMP2]], i32 [[VEC_PHI]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end:
@@ -221,18 +226,19 @@ define i32 @reduction_umax_intrinsic(ptr nocapture %A, ptr nocapture %B) {
; 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 <4 x i32> [ splat (i32 1000), [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[WIDE_LOAD]])
+; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1000), [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[WIDE_LOAD]])
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
-; CHECK-NEXT: br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], 256
+; CHECK-NEXT: br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: br label [[FOR_END:%.*]]
; CHECK: for.end:
-; CHECK-NEXT: [[TMP3:%.*]] = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> [[TMP1]])
-; CHECK-NEXT: ret i32 [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> [[TMP2]])
+; CHECK-NEXT: ret i32 [[TMP4]]
;
entry:
br label %for.body
diff --git a/llvm/test/Transforms/LoopVectorize/reduction-inloop-pred.ll b/llvm/test/Transforms/LoopVectorize/reduction-inloop-pred.ll
index 1b9dcadbbfc39..3877b54e2a070 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction-inloop-pred.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction-inloop-pred.ll
@@ -17,7 +17,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP28]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -26,7 +27,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP29]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP9]], i64 1
@@ -36,7 +38,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP30]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 8
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP15]], i64 2
@@ -46,7 +49,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP18]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP31]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i64 12
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP21]], i64 3
@@ -99,10 +103,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -113,10 +119,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if2:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP53]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -127,10 +135,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6:%.*]]
; CHECK: pred.load.if4:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP55:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP55]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -141,10 +151,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF7:%.*]], label [[PRED_LOAD_CONTINUE8]]
; CHECK: pred.load.if6:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP56]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP57:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP57]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE8]]
@@ -208,7 +220,8 @@ define i32 @reduction_sum_const(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP31]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -217,7 +230,8 @@ define i32 @reduction_sum_const(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP32:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP32]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP9]], i64 1
@@ -227,7 +241,8 @@ define i32 @reduction_sum_const(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP33]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 8
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP15]], i64 2
@@ -237,7 +252,8 @@ define i32 @reduction_sum_const(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP18]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP34]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i64 12
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP21]], i64 3
@@ -294,10 +310,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -308,10 +326,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if2:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP53]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -322,10 +342,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6:%.*]]
; CHECK: pred.load.if4:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP55:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP55]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -336,10 +358,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF7:%.*]], label [[PRED_LOAD_CONTINUE8]]
; CHECK: pred.load.if6:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP56]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP57:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP57]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE8]]
@@ -404,10 +428,12 @@ define i32 @reduction_mix(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[B1:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i32> poison, i32 [[TMP12]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -418,10 +444,12 @@ define i32 @reduction_mix(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if2:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i32, ptr [[B1]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP53]]
; CHECK-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP18]], align 4
; CHECK-NEXT: [[TMP23:%.*]] = insertelement <4 x i32> [[TMP14]], i32 [[TMP22]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -432,10 +460,12 @@ define i32 @reduction_mix(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6:%.*]]
; CHECK: pred.load.if4:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i32, ptr [[B1]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP55:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP55]]
; CHECK-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP28]], align 4
; CHECK-NEXT: [[TMP33:%.*]] = insertelement <4 x i32> [[TMP24]], i32 [[TMP32]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -446,10 +476,12 @@ define i32 @reduction_mix(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF7:%.*]], label [[PRED_LOAD_CONTINUE8]]
; CHECK: pred.load.if6:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP56]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
-; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds i32, ptr [[B1]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP57:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP57]]
; CHECK-NEXT: [[TMP48:%.*]] = load i32, ptr [[TMP38]], align 4
; CHECK-NEXT: [[TMP49:%.*]] = insertelement <4 x i32> [[TMP34]], i32 [[TMP48]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE8]]
@@ -511,10 +543,12 @@ define i32 @reduction_mul(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP47]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP48]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -525,10 +559,12 @@ define i32 @reduction_mul(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP49]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -539,10 +575,12 @@ define i32 @reduction_mul(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -553,10 +591,12 @@ define i32 @reduction_mul(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP53]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -614,10 +654,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP47]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP48]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -628,10 +670,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP49]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -642,10 +686,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -656,10 +702,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP53]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -717,10 +765,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP45]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP46]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -731,10 +781,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP47]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP48]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -745,10 +797,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP49]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -759,10 +813,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -818,10 +874,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP45]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP46]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -832,10 +890,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP47]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP48]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -846,10 +906,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP49]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -860,10 +922,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -919,10 +983,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP45]]
; CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x float> poison, float [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP46]]
; CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x float> poison, float [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -933,10 +999,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP47]]
; CHECK-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x float> [[TMP8]], float [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP48]]
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x float> [[TMP9]], float [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -947,10 +1015,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP49]]
; CHECK-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x float> [[TMP18]], float [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP26:%.*]] = load float, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x float> [[TMP19]], float [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -961,10 +1031,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x float> [[TMP28]], float [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP36:%.*]] = load float, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x float> [[TMP29]], float [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -1020,10 +1092,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP47]]
; CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x float> poison, float [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP48]]
; CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x float> poison, float [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -1034,10 +1108,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP49]]
; CHECK-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x float> [[TMP8]], float [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i64 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP50]]
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x float> [[TMP9]], float [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -1048,10 +1124,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP51]]
; CHECK-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x float> [[TMP18]], float [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i64 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP52]]
; CHECK-NEXT: [[TMP26:%.*]] = load float, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x float> [[TMP19]], float [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -1062,10 +1140,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP53:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP53]]
; CHECK-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x float> [[TMP28]], float [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP54:%.*]] = shl nsw i64 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP54]]
; CHECK-NEXT: [[TMP36:%.*]] = load float, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x float> [[TMP29]], float [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -1123,7 +1203,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP27]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -1132,7 +1213,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP28]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP9]], i64 1
@@ -1142,7 +1224,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP29]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 8
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP15]], i64 2
@@ -1152,7 +1235,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP18]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP30]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i64 12
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP21]], i64 3
@@ -1205,7 +1289,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP27]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -1214,7 +1299,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP28]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i64 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP9]], i64 1
@@ -1224,7 +1310,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP29]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 8
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP15]], i64 2
@@ -1234,7 +1321,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP18]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP30]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i64 12
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP21]], i64 3
@@ -1284,9 +1372,11 @@ define float @reduction_conditional(ptr %A, ptr %B, ptr %C, float %S) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x float> [ [[TMP0]], [[VECTOR_PH]] ], [ [[PREDPHI3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP13]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP14]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = fcmp ogt <4 x float> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-NEXT: [[TMP4:%.*]] = fcmp ogt <4 x float> [[WIDE_LOAD1]], splat (float 1.000000e+00)
diff --git a/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll b/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll
index 1c94e259fc71b..d0a6aae81bf16 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction-inloop-uf4.ll
@@ -15,7 +15,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; 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: [[TMP13:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP13]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 16
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 32
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 48
@@ -74,8 +75,10 @@ define i64 @reduction_sum_chain(ptr noalias %p, ptr noalias %q) {
; CHECK-NEXT: [[VEC_PHI1:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP19:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI2:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP21:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI3:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[TMP23:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i64, ptr [[P:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i64, ptr [[Q:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[P:%.*]], i64 [[TMP25]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[Q:%.*]], i64 [[TMP26]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP0]], i64 32
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP0]], i64 64
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr [[TMP0]], i64 96
@@ -161,7 +164,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP4:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP4]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP112:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP112]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -170,7 +174,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_LOAD_IF7:%.*]], label [[PRED_LOAD_CONTINUE8:%.*]]
; CHECK: pred.load.if4:
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP113:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP113]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP10]], i64 4
; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP12]], i64 1
@@ -180,7 +185,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP15]], label [[PRED_LOAD_IF9:%.*]], label [[PRED_LOAD_CONTINUE10:%.*]]
; CHECK: pred.load.if6:
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP114:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP114]]
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP16]], i64 8
; CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP17]], align 4
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <4 x i32> [[TMP14]], i32 [[TMP18]], i64 2
@@ -190,7 +196,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP21:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP21]], label [[PRED_LOAD_IF11:%.*]], label [[PRED_LOAD_CONTINUE12:%.*]]
; CHECK: pred.load.if8:
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP115:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP115]]
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[TMP22]], i64 12
; CHECK-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
; CHECK-NEXT: [[TMP25:%.*]] = insertelement <4 x i32> [[TMP20]], i32 [[TMP24]], i64 3
@@ -200,7 +207,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP27:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
; CHECK-NEXT: br i1 [[TMP27]], label [[PRED_LOAD_IF13:%.*]], label [[PRED_LOAD_CONTINUE14:%.*]]
; CHECK: pred.load.if10:
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP116:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP116]]
; CHECK-NEXT: [[TMP29:%.*]] = getelementptr i8, ptr [[TMP28]], i64 16
; CHECK-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
; CHECK-NEXT: [[TMP31:%.*]] = insertelement <4 x i32> poison, i32 [[TMP30]], i64 0
@@ -210,7 +218,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP33:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
; CHECK-NEXT: br i1 [[TMP33]], label [[PRED_LOAD_IF15:%.*]], label [[PRED_LOAD_CONTINUE16:%.*]]
; CHECK: pred.load.if12:
-; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP117:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP117]]
; CHECK-NEXT: [[TMP35:%.*]] = getelementptr i8, ptr [[TMP34]], i64 20
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP32]], i32 [[TMP36]], i64 1
@@ -220,7 +229,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP39:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
; CHECK-NEXT: br i1 [[TMP39]], label [[PRED_LOAD_IF17:%.*]], label [[PRED_LOAD_CONTINUE18:%.*]]
; CHECK: pred.load.if14:
-; CHECK-NEXT: [[TMP40:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP118:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP40:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP118]]
; CHECK-NEXT: [[TMP41:%.*]] = getelementptr i8, ptr [[TMP40]], i64 24
; CHECK-NEXT: [[TMP42:%.*]] = load i32, ptr [[TMP41]], align 4
; CHECK-NEXT: [[TMP43:%.*]] = insertelement <4 x i32> [[TMP38]], i32 [[TMP42]], i64 2
@@ -230,7 +240,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP45:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
; CHECK-NEXT: br i1 [[TMP45]], label [[PRED_LOAD_IF19:%.*]], label [[PRED_LOAD_CONTINUE20:%.*]]
; CHECK: pred.load.if16:
-; CHECK-NEXT: [[TMP46:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP119:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP46:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP119]]
; CHECK-NEXT: [[TMP47:%.*]] = getelementptr i8, ptr [[TMP46]], i64 28
; CHECK-NEXT: [[TMP48:%.*]] = load i32, ptr [[TMP47]], align 4
; CHECK-NEXT: [[TMP49:%.*]] = insertelement <4 x i32> [[TMP44]], i32 [[TMP48]], i64 3
@@ -240,7 +251,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP51:%.*]] = extractelement <4 x i1> [[TMP2]], i64 0
; CHECK-NEXT: br i1 [[TMP51]], label [[PRED_LOAD_IF21:%.*]], label [[PRED_LOAD_CONTINUE22:%.*]]
; CHECK: pred.load.if18:
-; CHECK-NEXT: [[TMP52:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP120:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP52:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP120]]
; CHECK-NEXT: [[TMP53:%.*]] = getelementptr i8, ptr [[TMP52]], i64 32
; CHECK-NEXT: [[TMP54:%.*]] = load i32, ptr [[TMP53]], align 4
; CHECK-NEXT: [[TMP55:%.*]] = insertelement <4 x i32> poison, i32 [[TMP54]], i64 0
@@ -250,7 +262,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP57:%.*]] = extractelement <4 x i1> [[TMP2]], i64 1
; CHECK-NEXT: br i1 [[TMP57]], label [[PRED_LOAD_IF23:%.*]], label [[PRED_LOAD_CONTINUE24:%.*]]
; CHECK: pred.load.if20:
-; CHECK-NEXT: [[TMP58:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP121:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP58:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP121]]
; CHECK-NEXT: [[TMP59:%.*]] = getelementptr i8, ptr [[TMP58]], i64 36
; CHECK-NEXT: [[TMP60:%.*]] = load i32, ptr [[TMP59]], align 4
; CHECK-NEXT: [[TMP61:%.*]] = insertelement <4 x i32> [[TMP56]], i32 [[TMP60]], i64 1
@@ -260,7 +273,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP63:%.*]] = extractelement <4 x i1> [[TMP2]], i64 2
; CHECK-NEXT: br i1 [[TMP63]], label [[PRED_LOAD_IF25:%.*]], label [[PRED_LOAD_CONTINUE26:%.*]]
; CHECK: pred.load.if22:
-; CHECK-NEXT: [[TMP64:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP122:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP64:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP122]]
; CHECK-NEXT: [[TMP65:%.*]] = getelementptr i8, ptr [[TMP64]], i64 40
; CHECK-NEXT: [[TMP66:%.*]] = load i32, ptr [[TMP65]], align 4
; CHECK-NEXT: [[TMP67:%.*]] = insertelement <4 x i32> [[TMP62]], i32 [[TMP66]], i64 2
@@ -270,7 +284,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP69:%.*]] = extractelement <4 x i1> [[TMP2]], i64 3
; CHECK-NEXT: br i1 [[TMP69]], label [[PRED_LOAD_IF27:%.*]], label [[PRED_LOAD_CONTINUE28:%.*]]
; CHECK: pred.load.if24:
-; CHECK-NEXT: [[TMP70:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP123:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP70:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP123]]
; CHECK-NEXT: [[TMP71:%.*]] = getelementptr i8, ptr [[TMP70]], i64 44
; CHECK-NEXT: [[TMP72:%.*]] = load i32, ptr [[TMP71]], align 4
; CHECK-NEXT: [[TMP73:%.*]] = insertelement <4 x i32> [[TMP68]], i32 [[TMP72]], i64 3
@@ -280,7 +295,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP75:%.*]] = extractelement <4 x i1> [[TMP3]], i64 0
; CHECK-NEXT: br i1 [[TMP75]], label [[PRED_LOAD_IF29:%.*]], label [[PRED_LOAD_CONTINUE30:%.*]]
; CHECK: pred.load.if26:
-; CHECK-NEXT: [[TMP76:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP124:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP76:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP124]]
; CHECK-NEXT: [[TMP77:%.*]] = getelementptr i8, ptr [[TMP76]], i64 48
; CHECK-NEXT: [[TMP78:%.*]] = load i32, ptr [[TMP77]], align 4
; CHECK-NEXT: [[TMP79:%.*]] = insertelement <4 x i32> poison, i32 [[TMP78]], i64 0
@@ -290,7 +306,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP81:%.*]] = extractelement <4 x i1> [[TMP3]], i64 1
; CHECK-NEXT: br i1 [[TMP81]], label [[PRED_LOAD_IF31:%.*]], label [[PRED_LOAD_CONTINUE32:%.*]]
; CHECK: pred.load.if28:
-; CHECK-NEXT: [[TMP82:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP125:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP82:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP125]]
; CHECK-NEXT: [[TMP83:%.*]] = getelementptr i8, ptr [[TMP82]], i64 52
; CHECK-NEXT: [[TMP84:%.*]] = load i32, ptr [[TMP83]], align 4
; CHECK-NEXT: [[TMP85:%.*]] = insertelement <4 x i32> [[TMP80]], i32 [[TMP84]], i64 1
@@ -300,7 +317,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP87:%.*]] = extractelement <4 x i1> [[TMP3]], i64 2
; CHECK-NEXT: br i1 [[TMP87]], label [[PRED_LOAD_IF33:%.*]], label [[PRED_LOAD_CONTINUE34:%.*]]
; CHECK: pred.load.if30:
-; CHECK-NEXT: [[TMP88:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP126:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP88:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP126]]
; CHECK-NEXT: [[TMP89:%.*]] = getelementptr i8, ptr [[TMP88]], i64 56
; CHECK-NEXT: [[TMP90:%.*]] = load i32, ptr [[TMP89]], align 4
; CHECK-NEXT: [[TMP91:%.*]] = insertelement <4 x i32> [[TMP86]], i32 [[TMP90]], i64 2
@@ -310,7 +328,8 @@ define i32 @predicated(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP93:%.*]] = extractelement <4 x i1> [[TMP3]], i64 3
; CHECK-NEXT: br i1 [[TMP93]], label [[PRED_LOAD_IF35:%.*]], label [[PRED_LOAD_CONTINUE36]]
; CHECK: pred.load.if32:
-; CHECK-NEXT: [[TMP94:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP127:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP94:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP127]]
; CHECK-NEXT: [[TMP95:%.*]] = getelementptr i8, ptr [[TMP94]], i64 60
; CHECK-NEXT: [[TMP96:%.*]] = load i32, ptr [[TMP95]], align 4
; CHECK-NEXT: [[TMP97:%.*]] = insertelement <4 x i32> [[TMP92]], i32 [[TMP96]], i64 3
@@ -395,7 +414,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP8]], i64 0
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP120:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP120]]
; CHECK-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x i32> poison, i32 [[TMP14]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -404,7 +424,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP17:%.*]] = extractelement <4 x i1> [[TMP8]], i64 1
; CHECK-NEXT: br i1 [[TMP17]], label [[PRED_LOAD_IF9:%.*]], label [[PRED_LOAD_CONTINUE10:%.*]]
; CHECK: pred.load.if6:
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP121:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP121]]
; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[TMP18]], i64 4
; CHECK-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
; CHECK-NEXT: [[TMP21:%.*]] = insertelement <4 x i32> [[TMP16]], i32 [[TMP20]], i64 1
@@ -414,7 +435,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP23:%.*]] = extractelement <4 x i1> [[TMP8]], i64 2
; CHECK-NEXT: br i1 [[TMP23]], label [[PRED_LOAD_IF11:%.*]], label [[PRED_LOAD_CONTINUE12:%.*]]
; CHECK: pred.load.if8:
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP122:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP122]]
; CHECK-NEXT: [[TMP25:%.*]] = getelementptr i8, ptr [[TMP24]], i64 8
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP22]], i32 [[TMP26]], i64 2
@@ -424,7 +446,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP29:%.*]] = extractelement <4 x i1> [[TMP8]], i64 3
; CHECK-NEXT: br i1 [[TMP29]], label [[PRED_LOAD_IF13:%.*]], label [[PRED_LOAD_CONTINUE14:%.*]]
; CHECK: pred.load.if10:
-; CHECK-NEXT: [[TMP30:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP123:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP30:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP123]]
; CHECK-NEXT: [[TMP31:%.*]] = getelementptr i8, ptr [[TMP30]], i64 12
; CHECK-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
; CHECK-NEXT: [[TMP33:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP32]], i64 3
@@ -434,7 +457,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP35:%.*]] = extractelement <4 x i1> [[TMP9]], i64 0
; CHECK-NEXT: br i1 [[TMP35]], label [[PRED_LOAD_IF15:%.*]], label [[PRED_LOAD_CONTINUE16:%.*]]
; CHECK: pred.load.if12:
-; CHECK-NEXT: [[TMP36:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP124:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP36:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP124]]
; CHECK-NEXT: [[TMP37:%.*]] = getelementptr i8, ptr [[TMP36]], i64 16
; CHECK-NEXT: [[TMP38:%.*]] = load i32, ptr [[TMP37]], align 4
; CHECK-NEXT: [[TMP39:%.*]] = insertelement <4 x i32> poison, i32 [[TMP38]], i64 0
@@ -444,7 +468,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP41:%.*]] = extractelement <4 x i1> [[TMP9]], i64 1
; CHECK-NEXT: br i1 [[TMP41]], label [[PRED_LOAD_IF17:%.*]], label [[PRED_LOAD_CONTINUE18:%.*]]
; CHECK: pred.load.if14:
-; CHECK-NEXT: [[TMP42:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP125:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP42:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP125]]
; CHECK-NEXT: [[TMP43:%.*]] = getelementptr i8, ptr [[TMP42]], i64 20
; CHECK-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
; CHECK-NEXT: [[TMP45:%.*]] = insertelement <4 x i32> [[TMP40]], i32 [[TMP44]], i64 1
@@ -454,7 +479,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP47:%.*]] = extractelement <4 x i1> [[TMP9]], i64 2
; CHECK-NEXT: br i1 [[TMP47]], label [[PRED_LOAD_IF19:%.*]], label [[PRED_LOAD_CONTINUE20:%.*]]
; CHECK: pred.load.if16:
-; CHECK-NEXT: [[TMP48:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP126:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP48:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP126]]
; CHECK-NEXT: [[TMP49:%.*]] = getelementptr i8, ptr [[TMP48]], i64 24
; CHECK-NEXT: [[TMP50:%.*]] = load i32, ptr [[TMP49]], align 4
; CHECK-NEXT: [[TMP51:%.*]] = insertelement <4 x i32> [[TMP46]], i32 [[TMP50]], i64 2
@@ -464,7 +490,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP53:%.*]] = extractelement <4 x i1> [[TMP9]], i64 3
; CHECK-NEXT: br i1 [[TMP53]], label [[PRED_LOAD_IF21:%.*]], label [[PRED_LOAD_CONTINUE22:%.*]]
; CHECK: pred.load.if18:
-; CHECK-NEXT: [[TMP54:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP127:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP54:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP127]]
; CHECK-NEXT: [[TMP55:%.*]] = getelementptr i8, ptr [[TMP54]], i64 28
; CHECK-NEXT: [[TMP56:%.*]] = load i32, ptr [[TMP55]], align 4
; CHECK-NEXT: [[TMP57:%.*]] = insertelement <4 x i32> [[TMP52]], i32 [[TMP56]], i64 3
@@ -474,7 +501,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP59:%.*]] = extractelement <4 x i1> [[TMP10]], i64 0
; CHECK-NEXT: br i1 [[TMP59]], label [[PRED_LOAD_IF23:%.*]], label [[PRED_LOAD_CONTINUE24:%.*]]
; CHECK: pred.load.if20:
-; CHECK-NEXT: [[TMP60:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP128:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP60:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP128]]
; CHECK-NEXT: [[TMP61:%.*]] = getelementptr i8, ptr [[TMP60]], i64 32
; CHECK-NEXT: [[TMP62:%.*]] = load i32, ptr [[TMP61]], align 4
; CHECK-NEXT: [[TMP63:%.*]] = insertelement <4 x i32> poison, i32 [[TMP62]], i64 0
@@ -484,7 +512,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP65:%.*]] = extractelement <4 x i1> [[TMP10]], i64 1
; CHECK-NEXT: br i1 [[TMP65]], label [[PRED_LOAD_IF25:%.*]], label [[PRED_LOAD_CONTINUE26:%.*]]
; CHECK: pred.load.if22:
-; CHECK-NEXT: [[TMP66:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP129:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP66:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP129]]
; CHECK-NEXT: [[TMP67:%.*]] = getelementptr i8, ptr [[TMP66]], i64 36
; CHECK-NEXT: [[TMP68:%.*]] = load i32, ptr [[TMP67]], align 4
; CHECK-NEXT: [[TMP69:%.*]] = insertelement <4 x i32> [[TMP64]], i32 [[TMP68]], i64 1
@@ -494,7 +523,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP71:%.*]] = extractelement <4 x i1> [[TMP10]], i64 2
; CHECK-NEXT: br i1 [[TMP71]], label [[PRED_LOAD_IF27:%.*]], label [[PRED_LOAD_CONTINUE28:%.*]]
; CHECK: pred.load.if24:
-; CHECK-NEXT: [[TMP72:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP130:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP72:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP130]]
; CHECK-NEXT: [[TMP73:%.*]] = getelementptr i8, ptr [[TMP72]], i64 40
; CHECK-NEXT: [[TMP74:%.*]] = load i32, ptr [[TMP73]], align 4
; CHECK-NEXT: [[TMP75:%.*]] = insertelement <4 x i32> [[TMP70]], i32 [[TMP74]], i64 2
@@ -504,7 +534,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP77:%.*]] = extractelement <4 x i1> [[TMP10]], i64 3
; CHECK-NEXT: br i1 [[TMP77]], label [[PRED_LOAD_IF29:%.*]], label [[PRED_LOAD_CONTINUE30:%.*]]
; CHECK: pred.load.if26:
-; CHECK-NEXT: [[TMP78:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP131:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP78:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP131]]
; CHECK-NEXT: [[TMP79:%.*]] = getelementptr i8, ptr [[TMP78]], i64 44
; CHECK-NEXT: [[TMP80:%.*]] = load i32, ptr [[TMP79]], align 4
; CHECK-NEXT: [[TMP81:%.*]] = insertelement <4 x i32> [[TMP76]], i32 [[TMP80]], i64 3
@@ -514,7 +545,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP83:%.*]] = extractelement <4 x i1> [[TMP11]], i64 0
; CHECK-NEXT: br i1 [[TMP83]], label [[PRED_LOAD_IF31:%.*]], label [[PRED_LOAD_CONTINUE32:%.*]]
; CHECK: pred.load.if28:
-; CHECK-NEXT: [[TMP84:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP132:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP84:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP132]]
; CHECK-NEXT: [[TMP85:%.*]] = getelementptr i8, ptr [[TMP84]], i64 48
; CHECK-NEXT: [[TMP86:%.*]] = load i32, ptr [[TMP85]], align 4
; CHECK-NEXT: [[TMP87:%.*]] = insertelement <4 x i32> poison, i32 [[TMP86]], i64 0
@@ -524,7 +556,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP89:%.*]] = extractelement <4 x i1> [[TMP11]], i64 1
; CHECK-NEXT: br i1 [[TMP89]], label [[PRED_LOAD_IF33:%.*]], label [[PRED_LOAD_CONTINUE34:%.*]]
; CHECK: pred.load.if30:
-; CHECK-NEXT: [[TMP90:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP133:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP90:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP133]]
; CHECK-NEXT: [[TMP91:%.*]] = getelementptr i8, ptr [[TMP90]], i64 52
; CHECK-NEXT: [[TMP92:%.*]] = load i32, ptr [[TMP91]], align 4
; CHECK-NEXT: [[TMP93:%.*]] = insertelement <4 x i32> [[TMP88]], i32 [[TMP92]], i64 1
@@ -534,7 +567,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP95:%.*]] = extractelement <4 x i1> [[TMP11]], i64 2
; CHECK-NEXT: br i1 [[TMP95]], label [[PRED_LOAD_IF35:%.*]], label [[PRED_LOAD_CONTINUE36:%.*]]
; CHECK: pred.load.if32:
-; CHECK-NEXT: [[TMP96:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP134:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP96:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP134]]
; CHECK-NEXT: [[TMP97:%.*]] = getelementptr i8, ptr [[TMP96]], i64 56
; CHECK-NEXT: [[TMP98:%.*]] = load i32, ptr [[TMP97]], align 4
; CHECK-NEXT: [[TMP99:%.*]] = insertelement <4 x i32> [[TMP94]], i32 [[TMP98]], i64 2
@@ -544,7 +578,8 @@ define i32 @cond_rdx_pred(i32 %cond, ptr noalias %a, i64 %N) {
; CHECK-NEXT: [[TMP101:%.*]] = extractelement <4 x i1> [[TMP11]], i64 3
; CHECK-NEXT: br i1 [[TMP101]], label [[PRED_LOAD_IF37:%.*]], label [[PRED_LOAD_CONTINUE38]]
; CHECK: pred.load.if34:
-; CHECK-NEXT: [[TMP102:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP135:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP102:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP135]]
; CHECK-NEXT: [[TMP103:%.*]] = getelementptr i8, ptr [[TMP102]], i64 60
; CHECK-NEXT: [[TMP104:%.*]] = load i32, ptr [[TMP103]], align 4
; CHECK-NEXT: [[TMP105:%.*]] = insertelement <4 x i32> [[TMP100]], i32 [[TMP104]], i64 3
diff --git a/llvm/test/Transforms/LoopVectorize/reduction-predselect.ll b/llvm/test/Transforms/LoopVectorize/reduction-predselect.ll
index 855a0ce56f2c7..9a6e0bd4d3301 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction-predselect.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction-predselect.ll
@@ -17,7 +17,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP27]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -26,7 +27,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP28]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i32 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP9]], i64 1
@@ -36,7 +38,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP29]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i32 8
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP15]], i64 2
@@ -46,7 +49,8 @@ define i32 @reduction_sum_single(ptr noalias nocapture %A) {
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP18]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP30]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i32 12
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP21]], i64 3
@@ -97,10 +101,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP45]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP46]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -111,10 +117,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP47]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP48]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -125,10 +133,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP49]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP50]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -139,10 +149,12 @@ define i32 @reduction_sum(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i32 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP51]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP52]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -199,10 +211,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP44]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP45]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -213,10 +227,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP46]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP47]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -227,10 +243,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP48]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP49]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -241,10 +259,12 @@ define i32 @reduction_prod(ptr noalias nocapture %A, ptr noalias nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i32 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP50]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP51]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -299,10 +319,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP44]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP45]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -313,10 +335,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP46]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP47]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -327,10 +351,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP48]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP49]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -341,10 +367,12 @@ define i32 @reduction_and(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i32 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP50]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP51]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -399,10 +427,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP44]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP45]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -413,10 +443,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP46]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP47]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -427,10 +459,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP48]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP49]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -441,10 +475,12 @@ define i32 @reduction_or(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i32 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP50]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP51]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -499,10 +535,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP44]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP45]]
; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -513,10 +551,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP46]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP47]]
; CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x i32> [[TMP9]], i32 [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -527,10 +567,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP48]]
; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x i32> [[TMP18]], i32 [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP49]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x i32> [[TMP19]], i32 [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -541,10 +583,12 @@ define i32 @reduction_xor(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i32 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i32, ptr [[A]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP50]]
; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x i32> [[TMP28]], i32 [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i32, ptr [[B]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP51]]
; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -599,10 +643,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP44]]
; CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x float> poison, float [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP45]]
; CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x float> poison, float [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -613,10 +659,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[A]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP46]]
; CHECK-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x float> [[TMP8]], float [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds float, ptr [[B]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP47]]
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x float> [[TMP9]], float [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -627,10 +675,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP48]]
; CHECK-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x float> [[TMP18]], float [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds float, ptr [[B]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP49]]
; CHECK-NEXT: [[TMP26:%.*]] = load float, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x float> [[TMP19]], float [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -641,10 +691,12 @@ define float @reduction_fadd(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i32 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP50]]
; CHECK-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x float> [[TMP28]], float [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[B]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP51]]
; CHECK-NEXT: [[TMP36:%.*]] = load float, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x float> [[TMP29]], float [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -699,10 +751,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP44]]
; CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x float> poison, float [[TMP3]], i64 0
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[B:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i32 [[TMP45]]
; CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[TMP5]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = insertelement <4 x float> poison, float [[TMP6]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -713,10 +767,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP10]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
; CHECK-NEXT: [[TMP11:%.*]] = or disjoint i32 [[INDEX]], 1
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[A]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP46]]
; CHECK-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x float> [[TMP8]], float [[TMP13]], i64 1
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds float, ptr [[B]], i32 [[TMP11]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl nsw i32 [[TMP11]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP47]]
; CHECK-NEXT: [[TMP16:%.*]] = load float, ptr [[TMP15]], align 4
; CHECK-NEXT: [[TMP17:%.*]] = insertelement <4 x float> [[TMP9]], float [[TMP16]], i64 1
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE2]]
@@ -727,10 +783,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP20]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
; CHECK-NEXT: [[TMP21:%.*]] = or disjoint i32 [[INDEX]], 2
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds float, ptr [[A]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP48:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP48]]
; CHECK-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4
; CHECK-NEXT: [[TMP24:%.*]] = insertelement <4 x float> [[TMP18]], float [[TMP23]], i64 2
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds float, ptr [[B]], i32 [[TMP21]]
+; CHECK-NEXT: [[TMP49:%.*]] = shl nsw i32 [[TMP21]], 2
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP49]]
; CHECK-NEXT: [[TMP26:%.*]] = load float, ptr [[TMP25]], align 4
; CHECK-NEXT: [[TMP27:%.*]] = insertelement <4 x float> [[TMP19]], float [[TMP26]], i64 2
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE4]]
@@ -741,10 +799,12 @@ define float @reduction_fmul(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: br i1 [[TMP30]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
; CHECK-NEXT: [[TMP31:%.*]] = or disjoint i32 [[INDEX]], 3
-; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds float, ptr [[A]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[A]], i32 [[TMP50]]
; CHECK-NEXT: [[TMP33:%.*]] = load float, ptr [[TMP32]], align 4
; CHECK-NEXT: [[TMP34:%.*]] = insertelement <4 x float> [[TMP28]], float [[TMP33]], i64 3
-; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds float, ptr [[B]], i32 [[TMP31]]
+; CHECK-NEXT: [[TMP51:%.*]] = shl nsw i32 [[TMP31]], 2
+; CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[B]], i32 [[TMP51]]
; CHECK-NEXT: [[TMP36:%.*]] = load float, ptr [[TMP35]], align 4
; CHECK-NEXT: [[TMP37:%.*]] = insertelement <4 x float> [[TMP29]], float [[TMP36]], i64 3
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE6]]
@@ -799,7 +859,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP27]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -808,7 +869,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP28]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i32 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP9]], i64 1
@@ -818,7 +880,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP29]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i32 8
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP15]], i64 2
@@ -828,7 +891,8 @@ define i32 @reduction_min(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP18]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP30]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i32 12
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP21]], i64 3
@@ -880,7 +944,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <4 x i1> [[TMP0]], i64 0
; CHECK-NEXT: br i1 [[TMP1]], label [[PRED_LOAD_IF:%.*]], label [[PRED_LOAD_CONTINUE:%.*]]
; CHECK: pred.load.if:
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP27:%.*]] = shl nsw i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i32 [[TMP27]]
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
; CHECK-NEXT: br label [[PRED_LOAD_CONTINUE]]
@@ -889,7 +954,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i1> [[TMP0]], i64 1
; CHECK-NEXT: br i1 [[TMP6]], label [[PRED_LOAD_IF1:%.*]], label [[PRED_LOAD_CONTINUE2:%.*]]
; CHECK: pred.load.if1:
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP28]]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP7]], i32 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP9]], i64 1
@@ -899,7 +965,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP0]], i64 2
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_LOAD_IF3:%.*]], label [[PRED_LOAD_CONTINUE4:%.*]]
; CHECK: pred.load.if3:
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP29:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP29]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i32 8
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP14]], align 4
; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP15]], i64 2
@@ -909,7 +976,8 @@ define i32 @reduction_max(ptr nocapture %A, ptr nocapture %B) {
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP0]], i64 3
; CHECK-NEXT: br i1 [[TMP18]], label [[PRED_LOAD_IF5:%.*]], label [[PRED_LOAD_CONTINUE6]]
; CHECK: pred.load.if5:
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[A]], i32 [[INDEX]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl i32 [[INDEX]], 2
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[A]], i32 [[TMP30]]
; CHECK-NEXT: [[TMP20:%.*]] = getelementptr i8, ptr [[TMP19]], i32 12
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
; CHECK-NEXT: [[TMP22:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP21]], i64 3
diff --git a/llvm/test/Transforms/LoopVectorize/reduction.ll b/llvm/test/Transforms/LoopVectorize/reduction.ll
index 65d57015b0140..e8d6bba52cf10 100644
--- a/llvm/test/Transforms/LoopVectorize/reduction.ll
+++ b/llvm/test/Transforms/LoopVectorize/reduction.ll
@@ -17,9 +17,11 @@ define i32 @reduction_sum(i32 %n, ptr %A, ptr %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP19]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = add <4 x i32> [[VEC_PHI]], [[VEC_IND]]
; CHECK-NEXT: [[TMP6:%.*]] = add <4 x i32> [[TMP5]], [[WIDE_LOAD]]
@@ -39,9 +41,11 @@ define i32 @reduction_sum(i32 %n, ptr %A, ptr %B) {
; CHECK: .lr.ph:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[DOTLR_PH]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[SUM_02:%.*]] = phi i32 [ [[TMP17:%.*]], [[DOTLR_PH]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = trunc i64 [[INDVARS_IV]] to i32
; CHECK-NEXT: [[TMP15:%.*]] = add i32 [[SUM_02]], [[TMP14]]
@@ -98,9 +102,11 @@ define i32 @reduction_prod(i32 %n, ptr %A, ptr %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 1), [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP19]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = mul <4 x i32> [[VEC_PHI]], [[VEC_IND]]
; CHECK-NEXT: [[TMP6:%.*]] = mul <4 x i32> [[TMP5]], [[WIDE_LOAD]]
@@ -120,9 +126,11 @@ define i32 @reduction_prod(i32 %n, ptr %A, ptr %B) {
; CHECK: .lr.ph:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[DOTLR_PH]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[PROD_02:%.*]] = phi i32 [ [[TMP17:%.*]], [[DOTLR_PH]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = trunc i64 [[INDVARS_IV]] to i32
; CHECK-NEXT: [[TMP15:%.*]] = mul i32 [[PROD_02]], [[TMP14]]
@@ -179,9 +187,11 @@ define i32 @reduction_mix(i32 %n, ptr %A, ptr %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP19]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = mul nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP6:%.*]] = add <4 x i32> [[VEC_PHI]], [[VEC_IND]]
@@ -201,9 +211,11 @@ define i32 @reduction_mix(i32 %n, ptr %A, ptr %B) {
; CHECK: .lr.ph:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[DOTLR_PH]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[SUM_02:%.*]] = phi i32 [ [[TMP17:%.*]], [[DOTLR_PH]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = mul nsw i32 [[TMP13]], [[TMP11]]
; CHECK-NEXT: [[TMP15:%.*]] = trunc i64 [[INDVARS_IV]] to i32
@@ -260,9 +272,11 @@ define i32 @reduction_mul(i32 %n, ptr %A, ptr %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 19, i32 1, i32 1, i32 1>, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP19]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = add <4 x i32> [[WIDE_LOAD]], [[VEC_IND]]
; CHECK-NEXT: [[TMP6:%.*]] = add <4 x i32> [[TMP5]], [[WIDE_LOAD1]]
@@ -282,9 +296,11 @@ define i32 @reduction_mul(i32 %n, ptr %A, ptr %B) {
; CHECK: .lr.ph:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[DOTLR_PH]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[SUM_02:%.*]] = phi i32 [ [[TMP17:%.*]], [[DOTLR_PH]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = trunc i64 [[INDVARS_IV]] to i32
; CHECK-NEXT: [[TMP15:%.*]] = add i32 [[TMP11]], [[TMP14]]
@@ -341,9 +357,11 @@ define i32 @start_at_non_zero(ptr %in, ptr %coeff, ptr %out, i32 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ <i32 120, i32 0, i32 0, i32 0>, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[COEFF]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[COEFF]], i64 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = mul nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP4]] = add <4 x i32> [[TMP3]], [[VEC_PHI]]
@@ -361,9 +379,11 @@ define i32 @start_at_non_zero(ptr %in, ptr %coeff, ptr %out, i32 %n) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[SUM_09:%.*]] = phi i32 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[IN]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[IN]], i64 [[TMP9]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[COEFF]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[COEFF]], i64 [[TMP11]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], [[TMP7]]
; CHECK-NEXT: [[ADD]] = add nsw i32 [[MUL]], [[SUM_09]]
@@ -417,9 +437,11 @@ define i32 @reduction_and(i32 %n, ptr %A, ptr %B) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ splat (i32 -1), [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP4]] = and <4 x i32> [[TMP3]], [[VEC_PHI]]
@@ -437,9 +459,11 @@ define i32 @reduction_and(i32 %n, ptr %A, ptr %B) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[RESULT_08:%.*]] = phi i32 [ [[AND:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP11]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP7]]
; CHECK-NEXT: [[AND]] = and i32 [[ADD]], [[RESULT_08]]
@@ -493,9 +517,11 @@ define i32 @reduction_or(i32 %n, ptr %A, ptr %B) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP4]] = or <4 x i32> [[TMP3]], [[VEC_PHI]]
@@ -513,9 +539,11 @@ define i32 @reduction_or(i32 %n, ptr %A, ptr %B) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[RESULT_08:%.*]] = phi i32 [ [[OR:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP11]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP7]]
; CHECK-NEXT: [[OR]] = or i32 [[ADD]], [[RESULT_08]]
@@ -569,9 +597,11 @@ define i32 @reduction_xor(i32 %n, ptr %A, ptr %B) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP12]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <4 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
; CHECK-NEXT: [[TMP4]] = xor <4 x i32> [[TMP3]], [[VEC_PHI]]
@@ -589,9 +619,11 @@ define i32 @reduction_xor(i32 %n, ptr %A, ptr %B) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[RESULT_08:%.*]] = phi i32 [ [[XOR:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP11]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP7]]
; CHECK-NEXT: [[XOR]] = xor i32 [[ADD]], [[RESULT_08]]
@@ -641,7 +673,8 @@ define i32 @reduction_sub_rhs(i32 %n, ptr %A) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; CHECK-NEXT: [[X_05:%.*]] = phi i32 [ [[SUB:%.*]], [[FOR_BODY]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[SUB]] = sub nsw i32 [[TMP0]], [[X_05]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -692,7 +725,8 @@ define i32 @reduction_sub_lhs(i32 %n, ptr %A) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[TMP2]] = sub <4 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
@@ -709,7 +743,8 @@ define i32 @reduction_sub_lhs(i32 %n, ptr %A) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[X_05:%.*]] = phi i32 [ [[SUB:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[SUB]] = sub nsw i32 [[X_05]], [[TMP5]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1
@@ -755,9 +790,11 @@ define float @reduction_conditional(ptr %A, ptr %B, ptr %C, float %S) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x float> [ [[TMP0]], [[VECTOR_PH]] ], [ [[PREDPHI3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP13]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP14]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = fcmp ogt <4 x float> [[WIDE_LOAD]], [[WIDE_LOAD1]]
; CHECK-NEXT: [[TMP4:%.*]] = fcmp ogt <4 x float> [[WIDE_LOAD1]], splat (float 1.000000e+00)
@@ -830,9 +867,11 @@ define float @noreduction_phi(ptr %A, ptr %B, ptr %C, float %S) {
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
; CHECK-NEXT: [[SUM_033:%.*]] = phi float [ [[S]], [[ENTRY]] ], [ [[SUM_1:%.*]], [[FOR_INC]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
; CHECK-NEXT: [[CMP3:%.*]] = fcmp ogt float [[TMP0]], [[TMP1]]
; CHECK-NEXT: br i1 [[CMP3]], label [[IF_THEN:%.*]], label [[FOR_INC]]
@@ -908,7 +947,8 @@ define float @noredux_header_phi(ptr %A, ptr %B, ptr %C, float %S) {
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[SUM2_09:%.*]] = phi float [ 0.000000e+00, [[ENTRY]] ], [ [[ADD1:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[SUM_08:%.*]] = phi float [ [[S]], [[ENTRY]] ], [ [[ADD:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD]] = fadd fast float [[SUM_08]], [[TMP0]]
; CHECK-NEXT: [[ADD1]] = fadd fast float [[SUM2_09]], [[ADD]]
@@ -1036,9 +1076,11 @@ define i32 @reduction_sum_multiuse(i32 %n, ptr %A, ptr %B) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP7:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP20]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <4 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = add <4 x i32> [[VEC_PHI]], [[VEC_IND]]
; CHECK-NEXT: [[TMP6:%.*]] = add <4 x i32> [[TMP5]], [[WIDE_LOAD]]
@@ -1058,9 +1100,11 @@ define i32 @reduction_sum_multiuse(i32 %n, ptr %A, ptr %B) {
; CHECK: .lr.ph:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[DOTLR_PH]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[SUM_02:%.*]] = phi i32 [ [[TMP17:%.*]], [[DOTLR_PH]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP22:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP22]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
; CHECK-NEXT: [[TMP14:%.*]] = trunc i64 [[INDVARS_IV]] to i32
; CHECK-NEXT: [[TMP15:%.*]] = add i32 [[SUM_02]], [[TMP14]]
@@ -1126,7 +1170,8 @@ define void @reduction_reset(i32 %N, ptr %arrayA, ptr %arrayB) {
; CHECK: .lr.ph:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[DOTLR_PH_PREHEADER]] ], [ [[INDVARS_IV_NEXT:%.*]], [[DOTLR_PH]] ]
; CHECK-NEXT: [[DOT017:%.*]] = phi i32 [ 100, [[DOTLR_PH_PREHEADER]] ], [ [[CSEL:%.*]], [[DOTLR_PH]] ]
-; CHECK-NEXT: [[C6:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAYA]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[C6:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYA]], i64 [[TMP0]]
; CHECK-NEXT: [[C7:%.*]] = load i32, ptr [[C6]], align 4
; CHECK-NEXT: [[C8:%.*]] = icmp sgt i32 [[C7]], 0
; CHECK-NEXT: [[C9:%.*]] = add nsw i32 [[C7]], [[DOT017]]
@@ -1136,11 +1181,12 @@ define void @reduction_reset(i32 %N, ptr %arrayA, ptr %arrayB) {
; CHECK-NEXT: br i1 [[EXITCOND]], label [[DOT_CRIT_EDGE_LOOPEXIT:%.*]], label [[DOTLR_PH]]
; CHECK: ._crit_edge.loopexit:
; CHECK-NEXT: [[PHITMP19:%.*]] = sext i32 [[C5]] to i64
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[PHITMP19]], 2
; CHECK-NEXT: br label [[DOT_CRIT_EDGE]]
; CHECK: ._crit_edge:
-; CHECK-NEXT: [[DOT015_LCSSA:%.*]] = phi i64 [ -1, [[ENTRY:%.*]] ], [ [[PHITMP19]], [[DOT_CRIT_EDGE_LOOPEXIT]] ]
+; CHECK-NEXT: [[DOT015_LCSSA:%.*]] = phi i64 [ -4, [[ENTRY:%.*]] ], [ [[TMP1]], [[DOT_CRIT_EDGE_LOOPEXIT]] ]
; CHECK-NEXT: [[DOT0_LCSSA:%.*]] = phi i32 [ 100, [[ENTRY]] ], [ [[CSEL]], [[DOT_CRIT_EDGE_LOOPEXIT]] ]
-; CHECK-NEXT: [[C10:%.*]] = getelementptr inbounds i32, ptr [[ARRAYB]], i64 [[DOT015_LCSSA]]
+; CHECK-NEXT: [[C10:%.*]] = getelementptr inbounds i8, ptr [[ARRAYB]], i64 [[DOT015_LCSSA]]
; CHECK-NEXT: store i32 [[DOT0_LCSSA]], ptr [[C10]], align 4
; CHECK-NEXT: ret void
;
@@ -1197,9 +1243,10 @@ define i64 @reduction_with_phi_with_one_incoming_on_backedge(i16 %n, ptr %A) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[DOTCAST1:%.*]] = trunc i32 [[INDEX]] to i16
-; CHECK-NEXT: [[TMP2:%.*]] = sext i16 [[DOTCAST1]] to i64
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i64, ptr [[A]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = or disjoint i16 [[DOTCAST1]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = sext i16 [[OFFSET_IDX]] to i64
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP4]] = add <4 x i64> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -1217,7 +1264,8 @@ define i64 @reduction_with_phi_with_one_incoming_on_backedge(i16 %n, ptr %A) {
; CHECK-NEXT: [[IV:%.*]] = phi i16 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT: [[SUM:%.*]] = phi i64 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[SUM_NEXT:%.*]], [[LOOP_LATCH]] ]
; CHECK-NEXT: [[TMP7:%.*]] = sext i16 [[IV]] to i64
-; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i64, ptr [[A]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[TMP7]], 3
+; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP10]]
; CHECK-NEXT: [[LV_A:%.*]] = load i64, ptr [[GEP_A]], align 4
; CHECK-NEXT: [[SUM_NEXT]] = add nsw i64 [[SUM]], [[LV_A]]
; CHECK-NEXT: br label [[LOOP_BB:%.*]]
@@ -1275,9 +1323,10 @@ define i64 @reduction_with_phi_with_two_incoming_on_backedge(i16 %n, ptr %A) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[DOTCAST1:%.*]] = trunc i32 [[INDEX]] to i16
-; CHECK-NEXT: [[TMP2:%.*]] = sext i16 [[DOTCAST1]] to i64
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i64, ptr [[A]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP3]], i64 8
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = or disjoint i16 [[DOTCAST1]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = sext i16 [[OFFSET_IDX]] to i64
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[TMP3]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP9]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[TMP4]] = add <4 x i64> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -1295,7 +1344,8 @@ define i64 @reduction_with_phi_with_two_incoming_on_backedge(i16 %n, ptr %A) {
; CHECK-NEXT: [[IV:%.*]] = phi i16 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT: [[SUM:%.*]] = phi i64 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[SUM_NEXT:%.*]], [[LOOP_LATCH]] ]
; CHECK-NEXT: [[TMP7:%.*]] = sext i16 [[IV]] to i64
-; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i64, ptr [[A]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[TMP7]], 3
+; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP10]]
; CHECK-NEXT: [[LV_A:%.*]] = load i64, ptr [[GEP_A]], align 4
; CHECK-NEXT: [[SUM_NEXT]] = add nsw i64 [[SUM]], [[LV_A]]
; CHECK-NEXT: [[CMP_0:%.*]] = icmp eq i64 [[LV_A]], 29
diff --git a/llvm/test/Transforms/LoopVectorize/runtime-check.ll b/llvm/test/Transforms/LoopVectorize/runtime-check.ll
index 6a9e6089c5499..eeff83f960a18 100644
--- a/llvm/test/Transforms/LoopVectorize/runtime-check.ll
+++ b/llvm/test/Transforms/LoopVectorize/runtime-check.ll
@@ -30,14 +30,16 @@ define void @foo(ptr nocapture %a, ptr nocapture %b, i32 %n) nounwind uwtable ss
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]], !dbg [[DBG9]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ], !dbg [[DBG9]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDEX]], !dbg [[DBG9]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP2]], align 4, !dbg [[DBG9]]
-; CHECK-NEXT: [[TMP3:%.*]] = fmul <4 x float> [[WIDE_LOAD]], splat (float 3.000000e+00), !dbg [[DBG9]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDEX]], !dbg [[DBG9]]
-; CHECK-NEXT: store <4 x float> [[TMP3]], ptr [[TMP4]], align 4, !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDEX]], 2, !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP2]], !dbg [[DBG9]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP3]], align 4, !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP4:%.*]] = fmul <4 x float> [[WIDE_LOAD]], splat (float 3.000000e+00), !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 2, !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP5]], !dbg [[DBG9]]
+; CHECK-NEXT: store <4 x float> [[TMP4]], ptr [[TMP6]], align 4, !dbg [[DBG9]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4, !dbg [[DBG9]]
-; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]], !dbg [[DBG9]]
-; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !dbg [[DBG9]], !llvm.loop [[LOOP10:![0-9]+]]
+; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]], !dbg [[DBG9]]
+; CHECK-NEXT: br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !dbg [[DBG9]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[TMP0]], !dbg [[DBG9]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]], !dbg [[DBG9]]
@@ -46,10 +48,12 @@ define void @foo(ptr nocapture %a, ptr nocapture %b, i32 %n) nounwind uwtable ss
; CHECK-NEXT: br label [[FOR_BODY:%.*]], !dbg [[DBG9]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], !dbg [[DBG9]]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[INDVARS_IV]], !dbg [[DBG9]]
-; CHECK-NEXT: [[TMP6:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !dbg [[DBG9]]
-; CHECK-NEXT: [[MUL:%.*]] = fmul float [[TMP6]], 3.000000e+00, !dbg [[DBG9]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[INDVARS_IV]], !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDVARS_IV]], 2, !dbg [[DBG9]]
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP8]], !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP9:%.*]] = load float, ptr [[ARRAYIDX]], align 4, !dbg [[DBG9]]
+; CHECK-NEXT: [[MUL:%.*]] = fmul float [[TMP9]], 3.000000e+00, !dbg [[DBG9]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDVARS_IV]], 2, !dbg [[DBG9]]
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP10]], !dbg [[DBG9]]
; CHECK-NEXT: store float [[MUL]], ptr [[ARRAYIDX2]], align 4, !dbg [[DBG9]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add i64 [[INDVARS_IV]], 1, !dbg [[DBG9]]
; CHECK-NEXT: [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32, !dbg [[DBG9]]
@@ -131,15 +135,17 @@ define void @test_runtime_check(ptr %a, float %b, i64 %offset, i64 %offset2, i64
; 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 float, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[TMP5]], i64 [[OFFSET]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP6]], align 4, !alias.scope [[META15:![0-9]+]], !noalias [[META18:![0-9]+]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr float, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP7]], i64 [[OFFSET2]]
-; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x float>, ptr [[TMP8]], align 4, !alias.scope [[META18]]
+; CHECK-NEXT: [[TMP5:%.*]] = add i64 [[INDEX]], [[OFFSET]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[TMP5]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP6]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP7]], align 4, !alias.scope [[META15:![0-9]+]], !noalias [[META18:![0-9]+]]
+; CHECK-NEXT: [[TMP8:%.*]] = add i64 [[INDEX]], [[OFFSET2]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[TMP8]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP12]]
+; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <4 x float>, ptr [[TMP13]], align 4, !alias.scope [[META18]]
; CHECK-NEXT: [[TMP9:%.*]] = fmul fast <4 x float> [[BROADCAST_SPLAT]], [[WIDE_LOAD4]]
; CHECK-NEXT: [[TMP10:%.*]] = fadd fast <4 x float> [[WIDE_LOAD]], [[TMP9]]
-; CHECK-NEXT: store <4 x float> [[TMP10]], ptr [[TMP6]], align 4, !alias.scope [[META15]], !noalias [[META18]]
+; CHECK-NEXT: store <4 x float> [[TMP10]], ptr [[TMP7]], align 4, !alias.scope [[META15]], !noalias [[META18]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
@@ -151,11 +157,13 @@ define void @test_runtime_check(ptr %a, float %b, i64 %offset, i64 %offset2, i64
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; CHECK-NEXT: [[ARR_IDX:%.*]] = getelementptr float, ptr [[TMP12]], i64 [[OFFSET]]
+; CHECK-NEXT: [[IND_SUM:%.*]] = add i64 [[IV]], [[OFFSET]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[IND_SUM]], 2
+; CHECK-NEXT: [[ARR_IDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP14]]
; CHECK-NEXT: [[L1:%.*]] = load float, ptr [[ARR_IDX]], align 4
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; CHECK-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP13]], i64 [[OFFSET2]]
+; CHECK-NEXT: [[IND_SUM2:%.*]] = add i64 [[IV]], [[OFFSET2]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[IND_SUM2]], 2
+; CHECK-NEXT: [[ARR_IDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP15]]
; CHECK-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4
; CHECK-NEXT: [[M:%.*]] = fmul fast float [[B]], [[L2]]
; CHECK-NEXT: [[AD:%.*]] = fadd fast float [[L1]], [[M]]
@@ -225,20 +233,24 @@ define void @test_runtime_check2(ptr %a, float %b, i64 %offset, i64 %offset2, i6
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr float, ptr [[A:%.*]], i64 [[IV]]
-; CHECK-NEXT: [[ARR_IDX:%.*]] = getelementptr float, ptr [[TMP0]], i64 [[OFFSET:%.*]]
+; CHECK-NEXT: [[IND_SUM:%.*]] = add i64 [[IV]], [[OFFSET:%.*]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[IND_SUM]], 2
+; CHECK-NEXT: [[ARR_IDX:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[L1:%.*]] = load float, ptr [[ARR_IDX]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr float, ptr [[A]], i64 [[IV]]
-; CHECK-NEXT: [[ARR_IDX2:%.*]] = getelementptr float, ptr [[TMP1]], i64 [[OFFSET2:%.*]]
+; CHECK-NEXT: [[IND_SUM2:%.*]] = add i64 [[IV]], [[OFFSET2:%.*]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[IND_SUM2]], 2
+; CHECK-NEXT: [[ARR_IDX2:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP1]]
; CHECK-NEXT: [[L2:%.*]] = load float, ptr [[ARR_IDX2]], align 4
; CHECK-NEXT: [[M:%.*]] = fmul fast float [[B:%.*]], [[L2]]
; 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_IND:%.*]] = shl i64 [[IV]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 [[C_IND]]
; 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 nuw float, ptr [[C]], i64 [[IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[C_IDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP3]]
; CHECK-NEXT: store float [[VC]], ptr [[C_IDX2]], align 4
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[IV_NEXT]], [[N:%.*]]
@@ -337,14 +349,18 @@ define void @different_load_store_pairs(ptr %src.1, ptr %src.2, ptr %dst.1, ptr
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[SRC_1]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !alias.scope [[META22:![0-9]+]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i64, ptr [[SRC_2]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD20:%.*]] = load <4 x i64>, ptr [[TMP3]], align 8, !alias.scope [[META25:![0-9]+]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr nusw i32, ptr [[DST_1]], i64 [[INDEX]]
-; CHECK-NEXT: store <4 x i32> [[WIDE_LOAD]], ptr [[TMP4]], align 4, !alias.scope [[META27:![0-9]+]], !noalias [[META29:![0-9]+]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr nusw i64, ptr [[DST_2]], i64 [[INDEX]]
-; CHECK-NEXT: store <4 x i64> [[WIDE_LOAD20]], ptr [[TMP5]], align 8, !alias.scope [[META31:![0-9]+]], !noalias [[META32:![0-9]+]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[SRC_1]], i64 [[TMP2]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4, !alias.scope [[META22:![0-9]+]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[SRC_2]], i64 [[TMP4]]
+; CHECK-NEXT: [[WIDE_LOAD20:%.*]] = load <4 x i64>, ptr [[TMP5]], align 8, !alias.scope [[META25:![0-9]+]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr nusw i8, ptr [[DST_1]], i64 [[TMP10]]
+; CHECK-NEXT: store <4 x i32> [[WIDE_LOAD]], ptr [[TMP7]], align 4, !alias.scope [[META27:![0-9]+]], !noalias [[META29:![0-9]+]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr nusw i8, ptr [[DST_2]], i64 [[TMP8]]
+; CHECK-NEXT: store <4 x i64> [[WIDE_LOAD20]], ptr [[TMP9]], align 8, !alias.scope [[META31:![0-9]+]], !noalias [[META32:![0-9]+]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP6]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP33:![0-9]+]]
@@ -356,13 +372,17 @@ define void @different_load_store_pairs(ptr %src.1, ptr %src.2, ptr %dst.1, ptr
; CHECK-NEXT: br label [[LOOP:%.*]]
; CHECK: loop:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT: [[GEP_SRC_1:%.*]] = getelementptr i32, ptr [[SRC_1]], i64 [[IV]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl i64 [[IV]], 2
+; CHECK-NEXT: [[GEP_SRC_1:%.*]] = getelementptr i8, ptr [[SRC_1]], i64 [[TMP11]]
; CHECK-NEXT: [[LD_SRC_1:%.*]] = load i32, ptr [[GEP_SRC_1]], align 4
-; CHECK-NEXT: [[GEP_SRC_2:%.*]] = getelementptr i64, ptr [[SRC_2]], i64 [[IV]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl i64 [[IV]], 3
+; CHECK-NEXT: [[GEP_SRC_2:%.*]] = getelementptr i8, ptr [[SRC_2]], i64 [[TMP12]]
; CHECK-NEXT: [[LD_SRC_2:%.*]] = load i64, ptr [[GEP_SRC_2]], align 8
-; CHECK-NEXT: [[GEP_DST_1:%.*]] = getelementptr nusw i32, ptr [[DST_1]], i64 [[IV]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[IV]], 2
+; CHECK-NEXT: [[GEP_DST_1:%.*]] = getelementptr nusw i8, ptr [[DST_1]], i64 [[TMP13]]
; CHECK-NEXT: store i32 [[LD_SRC_1]], ptr [[GEP_DST_1]], align 4
-; CHECK-NEXT: [[GEP_DST_2:%.*]] = getelementptr nusw i64, ptr [[DST_2]], i64 [[IV]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[IV]], 3
+; CHECK-NEXT: [[GEP_DST_2:%.*]] = getelementptr nusw i8, ptr [[DST_2]], i64 [[TMP14]]
; CHECK-NEXT: store i64 [[LD_SRC_2]], ptr [[GEP_DST_2]], align 8
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
; CHECK-NEXT: [[COND:%.*]] = icmp ult i64 [[IV_NEXT]], [[UMAX19]]
@@ -418,12 +438,15 @@ define dso_local void @forced_optsize(ptr noalias nocapture readonly %x_p, ptr 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 i64, ptr [[X_P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[X_P:%.*]], i64 [[TMP6]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP0]], align 8
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i64, ptr [[Y_P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[Y_P:%.*]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD1:%.*]] = load <2 x i64>, ptr [[TMP1]], align 8
; CHECK-NEXT: [[TMP2:%.*]] = add nsw <2 x i64> [[WIDE_LOAD1]], [[WIDE_LOAD]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[Z_P:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[Z_P:%.*]], i64 [[TMP5]]
; CHECK-NEXT: store <2 x i64> [[TMP2]], ptr [[TMP3]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 128
@@ -504,8 +527,9 @@ define void @test_scev_check_mul_add_expansion(ptr %out, ptr %in, i32 %len, i32
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i32 [[INDEX]], 6
; CHECK-NEXT: [[TMP6:%.*]] = sext i32 [[OFFSET_IDX]] to i64
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i16, ptr [[OUT]], i64 [[TMP6]]
-; CHECK-NEXT: store <4 x i16> zeroinitializer, ptr [[TMP7]], align 2, !alias.scope [[META36:![0-9]+]], !noalias [[META39:![0-9]+]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[TMP6]], 1
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[OUT]], i64 [[TMP7]]
+; CHECK-NEXT: store <4 x i16> zeroinitializer, ptr [[TMP10]], align 2, !alias.scope [[META36:![0-9]+]], !noalias [[META39:![0-9]+]]
; CHECK-NEXT: store i32 0, ptr [[IN]], align 4, !alias.scope [[META39]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
@@ -519,7 +543,8 @@ define void @test_scev_check_mul_add_expansion(ptr %out, ptr %in, i32 %len, i32
; CHECK: loop:
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[TMP9:%.*]] = sext i32 [[IV]] to i64
-; CHECK-NEXT: [[ARRAYIDX80:%.*]] = getelementptr i16, ptr [[OUT]], i64 [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[TMP9]], 1
+; CHECK-NEXT: [[ARRAYIDX80:%.*]] = getelementptr i8, ptr [[OUT]], i64 [[TMP11]]
; CHECK-NEXT: store i16 0, ptr [[ARRAYIDX80]], align 2
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i32 [[IV]], 1
; CHECK-NEXT: store i32 0, ptr [[IN]], align 4
diff --git a/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll b/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll
index 606051f5f9a8a..2011460ed56bb 100644
--- a/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/scalable-inductions.ll
@@ -28,14 +28,18 @@ define void @add_ind64_unrolled(ptr noalias nocapture %a, ptr noalias nocapture
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 2 x i64> [ [[TMP6]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[STEP_ADD:%.*]] = add nuw <vscale x 2 x i64> [[VEC_IND]], [[DOTSPLAT]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[TMP9]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl i64 [[TMP4]], 4
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[TMP9]], i64 [[TMP10]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 2 x i64>, ptr [[TMP9]], align 8
; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 2 x i64>, ptr [[TMP11]], align 8
; CHECK-NEXT: [[TMP12:%.*]] = add nsw <vscale x 2 x i64> [[WIDE_LOAD]], [[VEC_IND]]
; CHECK-NEXT: [[TMP13:%.*]] = add nsw <vscale x 2 x i64> [[WIDE_LOAD2]], [[STEP_ADD]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[TMP14]], i64 [[TMP8]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl i64 [[TMP4]], 4
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP14]], i64 [[TMP20]]
; CHECK-NEXT: store <vscale x 2 x i64> [[TMP12]], ptr [[TMP14]], align 8
; CHECK-NEXT: store <vscale x 2 x i64> [[TMP13]], ptr [[TMP16]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP5]]
@@ -50,10 +54,12 @@ define void @add_ind64_unrolled(ptr noalias nocapture %a, ptr noalias nocapture
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[I_08]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP18:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP18]], [[I_08]]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[I_08]], 3
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP19]]
; CHECK-NEXT: store i64 [[ADD]], ptr [[ARRAYIDX1]], align 8
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[N]]
@@ -102,14 +108,18 @@ define void @add_ind64_unrolled_nxv1i64(ptr noalias nocapture %a, ptr noalias no
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 1 x i64> [ [[TMP6]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[STEP_ADD:%.*]] = add nuw <vscale x 1 x i64> [[VEC_IND]], [[DOTSPLAT]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[TMP8]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[TMP8]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <vscale x 1 x i64>, ptr [[TMP8]], align 8
; CHECK-NEXT: [[WIDE_LOAD2:%.*]] = load <vscale x 1 x i64>, ptr [[TMP10]], align 8
; CHECK-NEXT: [[TMP11:%.*]] = add nsw <vscale x 1 x i64> [[WIDE_LOAD]], [[VEC_IND]]
; CHECK-NEXT: [[TMP12:%.*]] = add nsw <vscale x 1 x i64> [[WIDE_LOAD2]], [[STEP_ADD]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl nsw i64 [[TMP2]], 3
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP13]], i64 [[TMP19]]
; CHECK-NEXT: store <vscale x 1 x i64> [[TMP11]], ptr [[TMP13]], align 8
; CHECK-NEXT: store <vscale x 1 x i64> [[TMP12]], ptr [[TMP15]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
@@ -124,10 +134,12 @@ define void @add_ind64_unrolled_nxv1i64(ptr noalias nocapture %a, ptr noalias no
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[I_08]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP20]]
; CHECK-NEXT: [[TMP17:%.*]] = load i64, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP17]], [[I_08]]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nsw i64 [[I_08]], 3
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP18]]
; CHECK-NEXT: store i64 [[ADD]], ptr [[ARRAYIDX1]], align 8
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_08]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[N]]
@@ -187,7 +199,8 @@ define void @add_unique_ind32(ptr noalias nocapture %a, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x i32> [ [[TMP7]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP13]]
; CHECK-NEXT: store <vscale x 4 x i32> [[VEC_IND]], ptr [[TMP10]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <vscale x 4 x i32> [[VEC_IND]], [[BROADCAST_SPLAT]]
@@ -203,7 +216,8 @@ define void @add_unique_ind32(ptr noalias nocapture %a, i64 %n) {
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[R_07:%.*]] = phi i32 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nsw i64 [[I_08]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP12]]
; CHECK-NEXT: store i32 [[R_07]], ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD]] = add nuw nsw i32 [[R_07]], 2
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_08]], 1
@@ -264,7 +278,8 @@ define void @add_unique_indf32(ptr noalias nocapture %a, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <vscale x 4 x float> [ [[TMP9]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP11]]
; CHECK-NEXT: store <vscale x 4 x float> [[VEC_IND]], ptr [[TMP14]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP3]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = fadd <vscale x 4 x float> [[VEC_IND]], [[DOTSPLAT]]
@@ -280,7 +295,8 @@ define void @add_unique_indf32(ptr noalias nocapture %a, i64 %n) {
; CHECK: for.body:
; CHECK-NEXT: [[I_08:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[R_07:%.*]] = phi float [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[I_08]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nsw i64 [[I_08]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP16]]
; CHECK-NEXT: store float [[R_07]], ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD]] = fadd float [[R_07]], 2.000000e+00
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_08]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/trunc-reductions.ll b/llvm/test/Transforms/LoopVectorize/trunc-reductions.ll
index 2aebb73081364..f04b40e4cad7d 100644
--- a/llvm/test/Transforms/LoopVectorize/trunc-reductions.ll
+++ b/llvm/test/Transforms/LoopVectorize/trunc-reductions.ll
@@ -53,7 +53,8 @@ define i16 @reduction_or_trunc(ptr noalias nocapture %ptr) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <8 x i16> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[PTR:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP0]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 [[TMP4]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2]] = or <8 x i16> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -95,7 +96,8 @@ define i16 @reduction_xor_trunc(ptr noalias nocapture %ptr) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <8 x i16> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i16, ptr [[PTR:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP0]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 [[TMP4]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2]] = xor <8 x i16> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -218,7 +220,8 @@ define i16 @reduction_smax_trunc(ptr noalias nocapture %ptr) {
; CHECK-NEXT: [[SUM_02P:%.*]] = phi i32 [ [[MIN:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT: [[SUM_02:%.*]] = and i32 [[SUM_02P]], 65535
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[IV]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i16, ptr [[PTR:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[TMP0]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
; CHECK-NEXT: [[EXT:%.*]] = sext i16 [[LOAD]] to i32
; CHECK-NEXT: [[MIN]] = call i32 @llvm.smax.i32(i32 [[SUM_02]], i32 [[EXT]])
@@ -259,7 +262,8 @@ define i16 @reduction_umax_trunc(ptr noalias nocapture %ptr) {
; CHECK-NEXT: [[SUM_02P:%.*]] = phi i32 [ [[MIN:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT: [[SUM_02:%.*]] = and i32 [[SUM_02P]], 65535
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[IV]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i16, ptr [[PTR:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nsw i64 [[TMP0]], 1
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[PTR:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[LOAD:%.*]] = load i16, ptr [[GEP]], align 2
; CHECK-NEXT: [[EXT:%.*]] = zext i16 [[LOAD]] to i32
; CHECK-NEXT: [[MIN]] = call i32 @llvm.umax.i32(i32 [[SUM_02]], i32 [[EXT]])
@@ -302,7 +306,8 @@ define i32 @reduction_and_or(i16 %a, i32 %b, ptr %src) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <8 x i32> [ <i32 10, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>, [[VECTOR_PH]] ], [ [[TMP2:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[INDEX]] to i64
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[SRC:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[SRC:%.*]], i64 [[TMP4]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[TMP2]] = or <8 x i32> [[VEC_PHI]], [[WIDE_LOAD]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
@@ -317,7 +322,8 @@ define i32 @reduction_and_or(i16 %a, i32 %b, ptr %src) {
; CHECK-NEXT: [[IV:%.*]] = phi i32 [ 992, [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[OR67:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[OR:%.*]], [[LOOP]] ]
; CHECK-NEXT: [[TMP5:%.*]] = zext nneg i32 [[IV]] to i64
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[SRC]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nuw nsw i64 [[TMP5]], 2
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[SRC]], i64 [[TMP7]]
; CHECK-NEXT: [[L:%.*]] = load i32, ptr [[GEP]], align 4
; CHECK-NEXT: [[OR]] = or i32 [[OR67]], [[L]]
; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i32 [[IV]], 1
diff --git a/llvm/test/Transforms/LoopVectorize/uniform-args-call-variants.ll b/llvm/test/Transforms/LoopVectorize/uniform-args-call-variants.ll
index abdd5e9562590..238b5364beb07 100644
--- a/llvm/test/Transforms/LoopVectorize/uniform-args-call-variants.ll
+++ b/llvm/test/Transforms/LoopVectorize/uniform-args-call-variants.ll
@@ -14,14 +14,16 @@ define void @test_uniform(ptr noalias %dst, ptr readonly %src, i64 %uniform , i6
; 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 double, ptr [[SRC]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP0]], align 8
-; CHECK-NEXT: [[TMP1:%.*]] = call <2 x double> @foo_uniform(<2 x double> [[WIDE_LOAD]], i64 [[UNIFORM]])
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
-; CHECK-NEXT: store <2 x double> [[TMP1]], ptr [[TMP2]], align 8
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP1]], align 8
+; CHECK-NEXT: [[TMP2:%.*]] = call <2 x double> @foo_uniform(<2 x double> [[WIDE_LOAD]], i64 [[UNIFORM]])
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP3]]
+; CHECK-NEXT: store <2 x double> [[TMP2]], ptr [[TMP4]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; 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: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
@@ -30,10 +32,12 @@ define void @test_uniform(ptr noalias %dst, ptr readonly %src, i64 %uniform , i6
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[GEPSRC:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[GEPSRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP6]]
; CHECK-NEXT: [[DATA:%.*]] = load double, ptr [[GEPSRC]], align 8
; CHECK-NEXT: [[CALL:%.*]] = call double @foo(double [[DATA]], i64 [[UNIFORM]]) #[[ATTR0:[0-9]+]]
-; CHECK-NEXT: [[GEPDST:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[GEPDST:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP7]]
; CHECK-NEXT: store double [[CALL]], ptr [[GEPDST]], align 8
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
@@ -73,19 +77,21 @@ define void @test_uniform_not_invariant(ptr noalias %dst, ptr readonly %src, i64
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP0:%.*]] = or disjoint i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP1]], align 8
-; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x double> [[WIDE_LOAD]], i64 0
-; CHECK-NEXT: [[TMP3:%.*]] = extractelement <2 x double> [[WIDE_LOAD]], i64 1
-; CHECK-NEXT: [[TMP4:%.*]] = call double @foo(double [[TMP2]], i64 [[INDEX]]) #[[ATTR0]]
-; CHECK-NEXT: [[TMP5:%.*]] = call double @foo(double [[TMP3]], i64 [[TMP0]]) #[[ATTR0]]
-; CHECK-NEXT: [[TMP6:%.*]] = insertelement <2 x double> poison, double [[TMP4]], i64 0
-; CHECK-NEXT: [[TMP7:%.*]] = insertelement <2 x double> [[TMP6]], double [[TMP5]], i64 1
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]]
-; CHECK-NEXT: store <2 x double> [[TMP7]], ptr [[TMP8]], align 8
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP1]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP2]], align 8
+; CHECK-NEXT: [[TMP3:%.*]] = extractelement <2 x double> [[WIDE_LOAD]], i64 0
+; CHECK-NEXT: [[TMP4:%.*]] = extractelement <2 x double> [[WIDE_LOAD]], i64 1
+; CHECK-NEXT: [[TMP5:%.*]] = call double @foo(double [[TMP3]], i64 [[INDEX]]) #[[ATTR0]]
+; CHECK-NEXT: [[TMP6:%.*]] = call double @foo(double [[TMP4]], i64 [[TMP0]]) #[[ATTR0]]
+; CHECK-NEXT: [[TMP7:%.*]] = insertelement <2 x double> poison, double [[TMP5]], i64 0
+; CHECK-NEXT: [[TMP8:%.*]] = insertelement <2 x double> [[TMP7]], double [[TMP6]], i64 1
+; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP9]]
+; CHECK-NEXT: store <2 x double> [[TMP8]], ptr [[TMP10]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK: middle.block:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
@@ -94,10 +100,12 @@ define void @test_uniform_not_invariant(ptr noalias %dst, ptr readonly %src, i64
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[GEPSRC:%.*]] = getelementptr double, ptr [[SRC]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[GEPSRC:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP12]]
; CHECK-NEXT: [[DATA:%.*]] = load double, ptr [[GEPSRC]], align 8
; CHECK-NEXT: [[CALL:%.*]] = call double @foo(double [[DATA]], i64 [[INDVARS_IV]]) #[[ATTR0]]
-; CHECK-NEXT: [[GEPDST:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[GEPDST:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 [[TMP13]]
; CHECK-NEXT: store double [[CALL]], ptr [[GEPDST]], align 8
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
diff --git a/llvm/test/Transforms/LoopVectorize/vector-geps.ll b/llvm/test/Transforms/LoopVectorize/vector-geps.ll
index 90b4be0f96dce..91eb9504957c0 100644
--- a/llvm/test/Transforms/LoopVectorize/vector-geps.ll
+++ b/llvm/test/Transforms/LoopVectorize/vector-geps.ll
@@ -16,8 +16,10 @@ define void @vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], <4 x i64> [[VEC_IND]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds ptr, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw <4 x i64> [[VEC_IND]], splat (i64 2)
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i8, ptr [[B:%.*]], <4 x i64> [[TMP3]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP4]]
; 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 nuw nsw <4 x i64> [[VEC_IND]], splat (i64 4)
@@ -31,8 +33,10 @@ define void @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 nuw i32, ptr [[B]], i64 [[I]]
-; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw ptr, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[I]], 2
+; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP6]]
; 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]]
@@ -71,7 +75,8 @@ 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: [[TMP1:%.*]] = getelementptr inbounds ptr, ptr [[A:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A:%.*]], i64 [[TMP3]]
; CHECK-NEXT: store <4 x ptr> [[DOTSPLAT]], ptr [[TMP1]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
@@ -85,7 +90,8 @@ define void @uniform_vector_gep_stored(ptr %a, ptr %b, i64 %n) {
; CHECK: for.body:
; CHECK-NEXT: [[I:%.*]] = phi i64 [ [[I_NEXT:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ]
; CHECK-NEXT: [[VAR0:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 4
-; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw ptr, ptr [[A]], i64 [[I]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[I]], 3
+; CHECK-NEXT: [[VAR1:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP4]]
; 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]]
diff --git a/llvm/test/Transforms/LoopVersioningLICM/loopversioningLICM1.ll b/llvm/test/Transforms/LoopVersioningLICM/loopversioningLICM1.ll
index b4f8abbf83a53..524d0a2ad0ba3 100644
--- a/llvm/test/Transforms/LoopVersioningLICM/loopversioningLICM1.ll
+++ b/llvm/test/Transforms/LoopVersioningLICM/loopversioningLICM1.ll
@@ -30,7 +30,8 @@ define i32 @foo(ptr nocapture %var1, ptr nocapture readnone %var2, ptr nocapture
; CHECK: for.body3.lver.check:
; CHECK-NEXT: [[ADD:%.*]] = add i32 [[I_015]], [[ITR]]
; CHECK-NEXT: [[IDXPROM6:%.*]] = zext i32 [[I_015]] to i64
-; CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds nuw i32, ptr [[VAR3]], i64 [[IDXPROM6]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[IDXPROM6]], 2
+; CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR3]], i64 [[TMP11]]
; CHECK-NEXT: [[TMP3:%.*]] = zext i32 [[J_016]] to i64
; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[TMP3]], 2
; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[VAR1]], i64 [[TMP4]]
@@ -48,7 +49,8 @@ define i32 @foo(ptr nocapture %var1, ptr nocapture readnone %var2, ptr nocapture
; CHECK: for.body3.lver.orig:
; CHECK-NEXT: [[J_113_LVER_ORIG:%.*]] = phi i32 [ [[J_016]], [[FOR_BODY3_PH_LVER_ORIG]] ], [ [[INC_LVER_ORIG:%.*]], [[FOR_BODY3_LVER_ORIG]] ]
; CHECK-NEXT: [[IDXPROM_LVER_ORIG:%.*]] = zext i32 [[J_113_LVER_ORIG]] to i64
-; CHECK-NEXT: [[ARRAYIDX_LVER_ORIG:%.*]] = getelementptr inbounds nuw i32, ptr [[VAR1]], i64 [[IDXPROM_LVER_ORIG]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw i64 [[IDXPROM_LVER_ORIG]], 2
+; CHECK-NEXT: [[ARRAYIDX_LVER_ORIG:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR1]], i64 [[TMP10]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX_LVER_ORIG]], align 4
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX7]], align 4
; CHECK-NEXT: [[ADD8_LVER_ORIG:%.*]] = add nsw i32 [[TMP9]], [[ADD]]
@@ -63,7 +65,8 @@ define i32 @foo(ptr nocapture %var1, ptr nocapture readnone %var2, ptr nocapture
; CHECK-NEXT: [[ADD86:%.*]] = phi i32 [ [[ARRAYIDX7_PROMOTED]], [[FOR_BODY3_PH]] ], [ [[ADD8:%.*]], [[FOR_BODY3]] ]
; CHECK-NEXT: [[J_113:%.*]] = phi i32 [ [[J_016]], [[FOR_BODY3_PH]] ], [ [[INC:%.*]], [[FOR_BODY3]] ]
; CHECK-NEXT: [[IDXPROM:%.*]] = zext i32 [[J_113]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[VAR1]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nuw nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[VAR1]], i64 [[TMP13]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX]], align 4, !alias.scope [[META5:![0-9]+]], !noalias [[META2]]
; CHECK-NEXT: [[ADD8]] = add nsw i32 [[ADD86]], [[ADD]]
; CHECK-NEXT: [[INC]] = add nuw i32 [[J_113]], 1
diff --git a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll
index c448bb547bd27..425420742929f 100644
--- a/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll
+++ b/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-loops.ll
@@ -27,14 +27,16 @@ define void @multiply_noalias_4x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: br label [[INNER_BODY:%.*]]
; CHECK: inner.body:
; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[INNER_IV]], 5
+; CHECK-NEXT: [[TMP10:%.*]] = shl i64 [[ROWS_IV]], 3
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[DOTIDX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr double, ptr [[TMP0]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[TMP10]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x double>, ptr [[TMP1]], align 8
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP1]], i64 32
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x double>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[DOTIDX17:%.*]] = shl i64 [[COLS_IV]], 5
+; CHECK-NEXT: [[TMP11:%.*]] = shl i64 [[INNER_IV]], 3
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 [[DOTIDX17]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr double, ptr [[TMP2]], i64 [[INNER_IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP2]], i64 [[TMP11]]
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x double>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[TMP3]], i64 32
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x double>, ptr [[VEC_GEP3]], align 8
@@ -55,8 +57,9 @@ define void @multiply_noalias_4x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: [[ROWS_STEP]] = add i64 [[ROWS_IV]], 2
; CHECK-NEXT: [[ROWS_COND_NOT:%.*]] = icmp eq i64 [[ROWS_STEP]], 4
; CHECK-NEXT: [[DOTIDX18:%.*]] = shl i64 [[COLS_IV]], 5
+; CHECK-NEXT: [[TMP13:%.*]] = shl i64 [[ROWS_IV]], 3
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 [[DOTIDX18]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr double, ptr [[TMP8]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 [[TMP13]]
; CHECK-NEXT: store <2 x double> [[TMP5]], ptr [[TMP9]], align 8
; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i8, ptr [[TMP9]], i64 32
; CHECK-NEXT: store <2 x double> [[TMP7]], ptr [[VEC_GEP16]], align 8
@@ -103,14 +106,16 @@ define void @multiply_noalias_2x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: br label [[INNER_BODY:%.*]]
; CHECK: inner.body:
; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[INNER_IV]], 4
+; CHECK-NEXT: [[TMP14:%.*]] = shl i64 [[ROWS_IV]], 3
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[DOTIDX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i64, ptr [[TMP0]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[TMP14]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i64>, ptr [[TMP1]], align 8
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP1]], i64 16
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i64>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[DOTIDX17:%.*]] = shl i64 [[COLS_IV]], 5
+; CHECK-NEXT: [[TMP15:%.*]] = shl i64 [[INNER_IV]], 3
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 [[DOTIDX17]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i64, ptr [[TMP2]], i64 [[INNER_IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP2]], i64 [[TMP15]]
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x i64>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[TMP3]], i64 32
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x i64>, ptr [[VEC_GEP3]], align 8
@@ -135,8 +140,9 @@ define void @multiply_noalias_2x4(ptr noalias %A, ptr noalias %B, ptr noalias %C
; CHECK-NEXT: [[ROWS_STEP]] = add i64 [[ROWS_IV]], 2
; CHECK-NEXT: [[ROWS_COND_NOT:%.*]] = icmp eq i64 [[ROWS_IV]], 0
; CHECK-NEXT: [[DOTIDX18:%.*]] = shl i64 [[COLS_IV]], 4
+; CHECK-NEXT: [[TMP17:%.*]] = shl i64 [[ROWS_IV]], 3
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 [[DOTIDX18]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i64, ptr [[TMP12]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP12]], i64 [[TMP17]]
; CHECK-NEXT: store <2 x i64> [[TMP7]], ptr [[TMP13]], align 8
; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i8, ptr [[TMP13]], i64 16
; CHECK-NEXT: store <2 x i64> [[TMP11]], ptr [[VEC_GEP16]], align 8
@@ -189,14 +195,16 @@ define void @multiply_noalias_4x2_2x8(ptr noalias %A, ptr noalias %B, ptr noalia
; CHECK-NEXT: br label [[INNER_BODY:%.*]]
; CHECK: inner.body:
; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[INNER_IV]], 5
+; CHECK-NEXT: [[TMP14:%.*]] = shl i64 [[ROWS_IV]], 3
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[DOTIDX]]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i64, ptr [[TMP0]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i64 [[TMP14]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x i64>, ptr [[TMP1]], align 8
; CHECK-NEXT: [[VEC_GEP:%.*]] = getelementptr i8, ptr [[TMP1]], i64 32
; CHECK-NEXT: [[COL_LOAD1:%.*]] = load <2 x i64>, ptr [[VEC_GEP]], align 8
; CHECK-NEXT: [[DOTIDX17:%.*]] = shl i64 [[COLS_IV]], 4
+; CHECK-NEXT: [[TMP15:%.*]] = shl i64 [[INNER_IV]], 3
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[B:%.*]], i64 [[DOTIDX17]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i64, ptr [[TMP2]], i64 [[INNER_IV]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP2]], i64 [[TMP15]]
; CHECK-NEXT: [[COL_LOAD2:%.*]] = load <2 x i64>, ptr [[TMP3]], align 8
; CHECK-NEXT: [[VEC_GEP3:%.*]] = getelementptr i8, ptr [[TMP3]], i64 16
; CHECK-NEXT: [[COL_LOAD4:%.*]] = load <2 x i64>, ptr [[VEC_GEP3]], align 8
@@ -221,8 +229,9 @@ define void @multiply_noalias_4x2_2x8(ptr noalias %A, ptr noalias %B, ptr noalia
; CHECK-NEXT: [[ROWS_STEP]] = add i64 [[ROWS_IV]], 2
; CHECK-NEXT: [[ROWS_COND_NOT:%.*]] = icmp eq i64 [[ROWS_STEP]], 4
; CHECK-NEXT: [[DOTIDX18:%.*]] = shl i64 [[COLS_IV]], 5
+; CHECK-NEXT: [[TMP17:%.*]] = shl i64 [[ROWS_IV]], 3
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[C:%.*]], i64 [[DOTIDX18]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i64, ptr [[TMP12]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP12]], i64 [[TMP17]]
; CHECK-NEXT: store <2 x i64> [[TMP7]], ptr [[TMP13]], align 8
; CHECK-NEXT: [[VEC_GEP16:%.*]] = getelementptr i8, ptr [[TMP13]], i64 32
; CHECK-NEXT: store <2 x i64> [[TMP11]], ptr [[VEC_GEP16]], align 8
@@ -302,14 +311,16 @@ define void @multiply_alias_2x2(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: br label [[INNER_BODY:%.*]]
; CHECK: inner.body:
; CHECK-NEXT: [[DOTIDX1:%.*]] = shl i64 [[INNER_IV]], 3
+; CHECK-NEXT: [[TMP9:%.*]] = shl i64 [[ROWS_IV]], 2
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[TMP3]], i64 [[DOTIDX1]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr float, ptr [[TMP8]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[TMP8]], i64 [[TMP9]]
; CHECK-NEXT: [[COL_LOAD:%.*]] = load <2 x float>, ptr [[TMP10]], align 4
; 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: [[DOTIDX24:%.*]] = shl i64 [[COLS_IV]], 3
+; CHECK-NEXT: [[TMP20:%.*]] = shl i64 [[INNER_IV]], 2
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP7]], i64 [[DOTIDX24]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr float, ptr [[TMP11]], i64 [[INNER_IV]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[TMP11]], i64 [[TMP20]]
; CHECK-NEXT: [[COL_LOAD9:%.*]] = load <2 x float>, ptr [[TMP13]], align 4
; CHECK-NEXT: [[VEC_GEP10:%.*]] = getelementptr i8, ptr [[TMP13]], i64 8
; CHECK-NEXT: [[COL_LOAD11:%.*]] = load <2 x float>, ptr [[VEC_GEP10]], align 4
@@ -330,8 +341,9 @@ define void @multiply_alias_2x2(ptr %A, ptr %B, ptr %C) {
; CHECK-NEXT: [[ROWS_STEP]] = add i64 [[ROWS_IV]], 2
; CHECK-NEXT: [[ROWS_COND_NOT:%.*]] = icmp eq i64 [[ROWS_IV]], 0
; CHECK-NEXT: [[DOTIDX:%.*]] = shl i64 [[COLS_IV]], 3
+; CHECK-NEXT: [[TMP21:%.*]] = shl i64 [[ROWS_IV]], 2
; CHECK-NEXT: [[TMP18:%.*]] = getelementptr i8, ptr [[C]], i64 [[DOTIDX]]
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr float, ptr [[TMP18]], i64 [[ROWS_IV]]
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i8, ptr [[TMP18]], i64 [[TMP21]]
; CHECK-NEXT: store <2 x float> [[TMP15]], ptr [[TMP19]], align 8
; CHECK-NEXT: [[VEC_GEP23:%.*]] = getelementptr i8, ptr [[TMP19]], i64 8
; CHECK-NEXT: store <2 x float> [[TMP17]], ptr [[VEC_GEP23]], align 8
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/block_scaling_decompr_8bit.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/block_scaling_decompr_8bit.ll
index 7175816963ed1..d13713e09870a 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/block_scaling_decompr_8bit.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/block_scaling_decompr_8bit.ll
@@ -20,7 +20,8 @@ define dso_local noundef i32 @_Z33block_scaling_decompr_8bitjPK27compressed_data
; CHECK: [[FOR_BODY_US]]:
; CHECK-NEXT: [[INDVARS_IV55:%.*]] = phi i64 [ [[INDVARS_IV_NEXT56:%.*]], %[[FOR_BODY_US]] ], [ 0, %[[FOR_BODY_LR_PH]] ]
; CHECK-NEXT: [[DST_ADDR_052_US:%.*]] = phi ptr [ [[DST_ADDR_1_US:%.*]], %[[FOR_BODY_US]] ], [ [[DST]], %[[FOR_BODY_LR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX_US:%.*]] = getelementptr inbounds nuw [[STRUCT_COMPRESSED_DATA_8BIT:%.*]], ptr [[SRC]], i64 [[INDVARS_IV55]]
+; CHECK-NEXT: [[TMP8:%.*]] = mul nuw nsw i64 [[INDVARS_IV55]], 25
+; CHECK-NEXT: [[ARRAYIDX_US:%.*]] = getelementptr inbounds nuw i8, ptr [[SRC]], i64 [[TMP8]]
; CHECK-NEXT: [[MANTISSA_US:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX_US]], i64 1
; CHECK-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[MANTISSA_US]], align 1
; CHECK-NEXT: [[VMOVL_I59_US:%.*]] = sext <8 x i8> [[TMP0]] to <8 x i16>
@@ -53,7 +54,8 @@ define dso_local noundef i32 @_Z33block_scaling_decompr_8bitjPK27compressed_data
; CHECK-NEXT: [[AGG_TMP_COERCE_050:%.*]] = phi i64 [ [[AGG_TMP_COERCE_0_INSERT_INSERT:%.*]], %[[FOR_BODY]] ], [ undef, %[[FOR_BODY_LR_PH]] ]
; CHECK-NEXT: [[AGG_TMP42_COERCE_049:%.*]] = phi i64 [ [[AGG_TMP42_COERCE_0_INSERT_INSERT:%.*]], %[[FOR_BODY]] ], [ undef, %[[FOR_BODY_LR_PH]] ]
; CHECK-NEXT: [[AGG_TMP37_COERCE_048:%.*]] = phi i64 [ [[AGG_TMP37_COERCE_0_INSERT_INSERT:%.*]], %[[FOR_BODY]] ], [ undef, %[[FOR_BODY_LR_PH]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [[STRUCT_COMPRESSED_DATA_8BIT]], ptr [[SRC]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = mul nuw nsw i64 [[INDVARS_IV]], 25
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[SRC]], i64 [[TMP9]]
; CHECK-NEXT: [[MANTISSA:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX]], i64 1
; CHECK-NEXT: [[TMP4:%.*]] = load <8 x i8>, ptr [[MANTISSA]], align 1
; CHECK-NEXT: [[VMOVL_I59:%.*]] = sext <8 x i8> [[TMP4]] to <8 x i16>
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/constraint-elimination-placement.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/constraint-elimination-placement.ll
index f60812f2f39be..2993827db63f6 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/constraint-elimination-placement.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/constraint-elimination-placement.ll
@@ -21,7 +21,8 @@ define i1 @test_order_1(ptr %this, ptr noalias %other, i1 %tobool9.not, i32 %cal
; CHECK-NEXT: [[CMP442:%.*]] = icmp sgt i32 [[CALL431]], 0
; CHECK-NEXT: br i1 [[CMP442]], label [[FOR_BODY45_LR_PH:%.*]], label [[FOR_COND]]
; CHECK: for.body45.lr.ph:
-; CHECK-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr ptr, ptr [[OTHER]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr i8, ptr [[OTHER]], i64 [[TMP2]]
; CHECK-NEXT: br label [[FOR_BODY45:%.*]]
; CHECK: for.cond.cleanup:
; CHECK-NEXT: store i32 0, ptr [[THIS]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/extra-unroll-simplifications.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/extra-unroll-simplifications.ll
index 5e1066ee626fa..9e160086e6ea4 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/extra-unroll-simplifications.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/extra-unroll-simplifications.ll
@@ -21,15 +21,17 @@ define void @partial_unroll_forced(i32 %N, ptr %src, ptr noalias %dst) {
; CHECK: loop.latch:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[LOOP_LATCH_PREHEADER_NEW]] ], [ [[INDVARS_IV_NEXT_1:%.*]], [[LOOP_LATCH]] ]
; CHECK-NEXT: [[NITER:%.*]] = phi i64 [ 0, [[LOOP_LATCH_PREHEADER_NEW]] ], [ [[NITER_NEXT_1:%.*]], [[LOOP_LATCH]] ]
-; CHECK-NEXT: [[SRC_IDX:%.*]] = getelementptr <8 x half>, ptr [[SRC]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 4
+; CHECK-NEXT: [[SRC_IDX:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP1]]
; CHECK-NEXT: [[L:%.*]] = load <8 x half>, ptr [[SRC_IDX]], align 16
-; CHECK-NEXT: [[DST_IDX:%.*]] = getelementptr <8 x half>, ptr [[DST]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[DST_IDX:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP1]]
; CHECK-NEXT: [[ADD:%.*]] = fadd <8 x half> [[L]], [[L]]
; CHECK-NEXT: store <8 x half> [[ADD]], ptr [[DST_IDX]], align 16
-; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = or disjoint i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[SRC_IDX_1:%.*]] = getelementptr <8 x half>, ptr [[SRC]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = shl i64 [[INDVARS_IV]], 4
+; CHECK-NEXT: [[TMP2:%.*]] = or disjoint i64 [[INDVARS_IV_NEXT]], 16
+; CHECK-NEXT: [[SRC_IDX_1:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP2]]
; CHECK-NEXT: [[L_1:%.*]] = load <8 x half>, ptr [[SRC_IDX_1]], align 16
-; CHECK-NEXT: [[DST_IDX_1:%.*]] = getelementptr <8 x half>, ptr [[DST]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[DST_IDX_1:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP2]]
; CHECK-NEXT: [[ADD_1:%.*]] = fadd <8 x half> [[L_1]], [[L_1]]
; CHECK-NEXT: store <8 x half> [[ADD_1]], ptr [[DST_IDX_1]], align 16
; CHECK-NEXT: [[INDVARS_IV_NEXT_1]] = add nuw nsw i64 [[INDVARS_IV]], 2
@@ -43,9 +45,10 @@ define void @partial_unroll_forced(i32 %N, ptr %src, ptr noalias %dst) {
; CHECK-NEXT: [[INDVARS_IV_UNR:%.*]] = phi i64 [ 0, [[LOOP_LATCH_PREHEADER]] ], [ [[INDVARS_IV_NEXT_1]], [[EXIT_LOOPEXIT_UNR_LCSSA]] ]
; CHECK-NEXT: [[LCMP_MOD4:%.*]] = trunc i32 [[N]] to i1
; CHECK-NEXT: tail call void @llvm.assume(i1 [[LCMP_MOD4]])
-; CHECK-NEXT: [[SRC_IDX_EPIL:%.*]] = getelementptr <8 x half>, ptr [[SRC]], i64 [[INDVARS_IV_UNR]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[INDVARS_IV_UNR]], 4
+; CHECK-NEXT: [[SRC_IDX_EPIL:%.*]] = getelementptr i8, ptr [[SRC]], i64 [[TMP3]]
; CHECK-NEXT: [[L_EPIL:%.*]] = load <8 x half>, ptr [[SRC_IDX_EPIL]], align 16
-; CHECK-NEXT: [[DST_IDX_EPIL:%.*]] = getelementptr <8 x half>, ptr [[DST]], i64 [[INDVARS_IV_UNR]]
+; CHECK-NEXT: [[DST_IDX_EPIL:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP3]]
; CHECK-NEXT: [[ADD_EPIL:%.*]] = fadd <8 x half> [[L_EPIL]], [[L_EPIL]]
; CHECK-NEXT: store <8 x half> [[ADD_EPIL]], ptr [[DST_IDX_EPIL]], align 16
; CHECK-NEXT: br label [[EXIT]]
@@ -93,18 +96,20 @@ define void @cse_matching_load_from_previous_unrolled_iteration(i32 %N, ptr %src
; CHECK: loop.latch:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[LOOP_LATCH_PREHEADER_NEW]] ], [ [[INDVARS_IV_NEXT_1:%.*]], [[LOOP_LATCH]] ]
; CHECK-NEXT: [[NITER:%.*]] = phi i64 [ 0, [[LOOP_LATCH_PREHEADER_NEW]] ], [ [[NITER_NEXT_1:%.*]], [[LOOP_LATCH]] ]
-; CHECK-NEXT: [[GEP_SRC_12:%.*]] = getelementptr <2 x i32>, ptr [[SRC_12]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[GEP_SRC_12:%.*]] = getelementptr i8, ptr [[SRC_12]], i64 [[TMP1]]
; CHECK-NEXT: [[L_12:%.*]] = load <2 x i32>, ptr [[GEP_SRC_12]], align 8
-; CHECK-NEXT: [[GEP_SRC_4:%.*]] = getelementptr <2 x i32>, ptr [[SRC_4]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[GEP_SRC_4:%.*]] = getelementptr i8, ptr [[SRC_4]], i64 [[TMP1]]
; CHECK-NEXT: [[L_4:%.*]] = load <2 x i32>, ptr [[GEP_SRC_4]], align 8
; CHECK-NEXT: [[MUL:%.*]] = mul <2 x i32> [[L_4]], [[L_12]]
-; CHECK-NEXT: [[GEP_DST:%.*]] = getelementptr <2 x i32>, ptr [[DST]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[GEP_DST:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP1]]
; CHECK-NEXT: store <2 x i32> [[MUL]], ptr [[GEP_DST]], align 8
-; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = or disjoint i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[GEP_SRC_12_1:%.*]] = getelementptr <2 x i32>, ptr [[SRC_12]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP2:%.*]] = or disjoint i64 [[INDVARS_IV_NEXT]], 8
+; CHECK-NEXT: [[GEP_SRC_12_1:%.*]] = getelementptr i8, ptr [[SRC_12]], i64 [[TMP2]]
; CHECK-NEXT: [[L_12_1:%.*]] = load <2 x i32>, ptr [[GEP_SRC_12_1]], align 8
; CHECK-NEXT: [[MUL_1:%.*]] = mul <2 x i32> [[L_12]], [[L_12_1]]
-; CHECK-NEXT: [[GEP_DST_1:%.*]] = getelementptr <2 x i32>, ptr [[DST]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[GEP_DST_1:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP2]]
; CHECK-NEXT: store <2 x i32> [[MUL_1]], ptr [[GEP_DST_1]], align 8
; CHECK-NEXT: [[INDVARS_IV_NEXT_1]] = add nuw nsw i64 [[INDVARS_IV]], 2
; CHECK-NEXT: [[NITER_NEXT_1]] = add i64 [[NITER]], 2
@@ -117,12 +122,13 @@ define void @cse_matching_load_from_previous_unrolled_iteration(i32 %N, ptr %src
; CHECK-NEXT: [[INDVARS_IV_UNR:%.*]] = phi i64 [ 0, [[LOOP_LATCH_PREHEADER]] ], [ [[INDVARS_IV_NEXT_1]], [[EXIT_LOOPEXIT_UNR_LCSSA]] ]
; CHECK-NEXT: [[LCMP_MOD4:%.*]] = trunc i32 [[N]] to i1
; CHECK-NEXT: tail call void @llvm.assume(i1 [[LCMP_MOD4]])
-; CHECK-NEXT: [[GEP_SRC_12_EPIL:%.*]] = getelementptr <2 x i32>, ptr [[SRC_12]], i64 [[INDVARS_IV_UNR]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[INDVARS_IV_UNR]], 3
+; CHECK-NEXT: [[GEP_SRC_12_EPIL:%.*]] = getelementptr i8, ptr [[SRC_12]], i64 [[TMP3]]
; CHECK-NEXT: [[L_12_EPIL:%.*]] = load <2 x i32>, ptr [[GEP_SRC_12_EPIL]], align 8
-; CHECK-NEXT: [[GEP_SRC_4_EPIL:%.*]] = getelementptr <2 x i32>, ptr [[SRC_4]], i64 [[INDVARS_IV_UNR]]
+; CHECK-NEXT: [[GEP_SRC_4_EPIL:%.*]] = getelementptr i8, ptr [[SRC_4]], i64 [[TMP3]]
; CHECK-NEXT: [[L_4_EPIL:%.*]] = load <2 x i32>, ptr [[GEP_SRC_4_EPIL]], align 8
; CHECK-NEXT: [[MUL_EPIL:%.*]] = mul <2 x i32> [[L_4_EPIL]], [[L_12_EPIL]]
-; CHECK-NEXT: [[GEP_DST_EPIL:%.*]] = getelementptr <2 x i32>, ptr [[DST]], i64 [[INDVARS_IV_UNR]]
+; CHECK-NEXT: [[GEP_DST_EPIL:%.*]] = getelementptr i8, ptr [[DST]], i64 [[TMP3]]
; CHECK-NEXT: store <2 x i32> [[MUL_EPIL]], ptr [[GEP_DST_EPIL]], align 8
; CHECK-NEXT: br label [[EXIT]]
; CHECK: exit:
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-load-from-vector-loop.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-load-from-vector-loop.ll
index c17b15138329c..eb3c90494926e 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-load-from-vector-loop.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-load-from-vector-loop.ll
@@ -34,14 +34,18 @@ define void @hoist_invariant_load(ptr %invariant_ptr, i64 %num_elements, ptr %ar
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[I2:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr nusw %"class.dealii::VectorizedArray", ptr [[ARRAY]], i64 [[I2]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr %"class.dealii::VectorizedArray", ptr [[ARRAY]], i64 [[I2]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[I2]], 5
+; CHECK-NEXT: [[TMP21:%.*]] = shl i64 [[I2]], 5
+; CHECK-NEXT: [[TMP22:%.*]] = shl i64 [[I2]], 5
+; CHECK-NEXT: [[TMP23:%.*]] = shl i64 [[I2]], 5
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr nusw i8, ptr [[ARRAY]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[ARRAY]], i64 [[TMP21]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP6]], i64 32
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr %"class.dealii::VectorizedArray", ptr [[ARRAY]], i64 [[I2]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[ARRAY]], i64 [[TMP22]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 64
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr %"class.dealii::VectorizedArray", ptr [[ARRAY]], i64 [[I2]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[ARRAY]], i64 [[TMP23]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP10]], i64 96
-; CHECK-NEXT: [[TMP12:%.*]] = load <5 x double>, ptr [[GEP]], align 8, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
+; CHECK-NEXT: [[TMP12:%.*]] = load <5 x double>, ptr [[TMP24]], align 8, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
; CHECK-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <5 x double> [[TMP12]], <5 x double> poison, <2 x i32> <i32 0, i32 4>
; CHECK-NEXT: [[TMP13:%.*]] = load <5 x double>, ptr [[TMP9]], align 8, !alias.scope [[META3]], !noalias [[META0]]
; CHECK-NEXT: [[STRIDED_VEC5:%.*]] = shufflevector <5 x double> [[TMP13]], <5 x double> poison, <2 x i32> <i32 0, i32 4>
@@ -51,7 +55,7 @@ define void @hoist_invariant_load(ptr %invariant_ptr, i64 %num_elements, ptr %ar
; CHECK-NEXT: [[TMP17:%.*]] = fadd <2 x double> [[BROADCAST_SPLAT]], [[STRIDED_VEC5]]
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <2 x double> [[TMP17]], i64 0
; CHECK-NEXT: [[TMP19:%.*]] = extractelement <2 x double> [[TMP17]], i64 1
-; CHECK-NEXT: store double [[TMP15]], ptr [[GEP]], align 8, !alias.scope [[META3]], !noalias [[META0]]
+; CHECK-NEXT: store double [[TMP15]], ptr [[TMP24]], align 8, !alias.scope [[META3]], !noalias [[META0]]
; CHECK-NEXT: store double [[TMP16]], ptr [[TMP7]], align 8, !alias.scope [[META3]], !noalias [[META0]]
; CHECK-NEXT: store double [[TMP18]], ptr [[TMP9]], align 8, !alias.scope [[META3]], !noalias [[META0]]
; CHECK-NEXT: store double [[TMP19]], ptr [[TMP11]], align 8, !alias.scope [[META3]], !noalias [[META0]]
@@ -63,7 +67,8 @@ define void @hoist_invariant_load(ptr %invariant_ptr, i64 %num_elements, ptr %ar
; CHECK-NEXT: br label %[[LOOP_LATCH:.*]]
; CHECK: [[LOOP_LATCH]]:
; CHECK-NEXT: [[I3:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[LOOP_LATCH]] ], [ [[I2_PH]], %[[LOOP_LATCH_PREHEADER6]] ]
-; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nusw %"class.dealii::VectorizedArray", ptr [[ARRAY]], i64 [[I3]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nsw i64 [[I3]], 5
+; CHECK-NEXT: [[GEP1:%.*]] = getelementptr nusw i8, ptr [[ARRAY]], i64 [[TMP25]]
; CHECK-NEXT: [[INVARIANT_VAL:%.*]] = load double, ptr [[INVARIANT_PTR]], align 8
; CHECK-NEXT: [[ARRAY_VAL:%.*]] = load double, ptr [[GEP1]], align 8
; CHECK-NEXT: [[SUM:%.*]] = fadd double [[INVARIANT_VAL]], [[ARRAY_VAL]]
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll
index 141503d344fe9..d1187edd9897d 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/hoist-runtime-checks.ll
@@ -25,7 +25,8 @@ define i32 @read_only_loop_with_runtime_check(ptr noundef %array, i32 noundef %c
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI11:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP5:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nuw nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAY]], i64 [[TMP10]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[TMP3]], align 4
@@ -49,7 +50,8 @@ define i32 @read_only_loop_with_runtime_check(ptr noundef %array, i32 noundef %c
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[INDVARS_IV_PH]], [[FOR_BODY_PREHEADER13]] ]
; CHECK-NEXT: [[SUM_07:%.*]] = phi i32 [ [[ADD]], [[FOR_BODY]] ], [ [[SUM_07_PH]], [[FOR_BODY_PREHEADER13]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAY]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAY]], i64 [[TMP9]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[ADD]] = add nsw i32 [[TMP8]], [[SUM_07]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
@@ -140,7 +142,8 @@ define dso_local noundef i32 @sum_prefix_with_sum(ptr %s.coerce0, i64 %s.coerce1
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP3:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI9:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP4:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[S_COERCE0]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[S_COERCE0]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[WIDE_LOAD10:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
@@ -164,7 +167,8 @@ define dso_local noundef i32 @sum_prefix_with_sum(ptr %s.coerce0, i64 %s.coerce1
; CHECK: for.body:
; CHECK-NEXT: [[I_07:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY1]] ], [ [[I_07_PH]], [[FOR_BODY_PREHEADER11]] ]
; CHECK-NEXT: [[RET_06:%.*]] = phi i32 [ [[ADD1]], [[FOR_BODY1]] ], [ [[RET_0_LCSSA]], [[FOR_BODY_PREHEADER11]] ]
-; CHECK-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds i32, ptr [[S_COERCE0]], i64 [[I_07]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nsw i64 [[I_07]], 2
+; CHECK-NEXT: [[ARRAYIDX_I:%.*]] = getelementptr inbounds i8, ptr [[S_COERCE0]], i64 [[TMP8]]
; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX_I]], align 4
; CHECK-NEXT: [[ADD1]] = add nsw i32 [[TMP7]], [[RET_06]]
; CHECK-NEXT: [[INC]] = add nuw i64 [[I_07]], 1
@@ -234,7 +238,8 @@ define hidden noundef nonnull align 4 dereferenceable(4) ptr @span_checked_acces
; CHECK-NEXT: unreachable
; CHECK: cond.end:
; CHECK-NEXT: [[TMP1:%.*]] = load ptr, ptr [[THIS]], align 8
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[__IDX]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[__IDX]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[TMP2]]
; CHECK-NEXT: ret ptr [[ARRAYIDX]]
;
entry:
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 1df9536c3c90f..69023c25ccc1d 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
@@ -48,7 +48,8 @@ define void @loop(ptr %X, ptr %Y) {
; CHECK-NEXT: br i1 [[DIFF_CHECK]], label [[FOR_BODY:%.*]], label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP16:%.*]] = shl nuw nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP16]]
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw 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
@@ -60,7 +61,7 @@ define void @loop(ptr %X, ptr %Y) {
; CHECK-NEXT: [[TMP8:%.*]] = select <2 x i1> [[TMP6]], <2 x double> splat (double 6.000000e+00), <2 x double> [[WIDE_LOAD8]]
; 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 nuw double, ptr [[X]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP16]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw 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
@@ -71,13 +72,14 @@ define void @loop(ptr %X, ptr %Y) {
; CHECK-NEXT: ret void
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP15]]
; CHECK-NEXT: [[TMP14:%.*]] = load double, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[CMP_I:%.*]] = fcmp olt double [[TMP14]], 0.000000e+00
; CHECK-NEXT: [[CMP1_I:%.*]] = fcmp ogt double [[TMP14]], 6.000000e+00
; CHECK-NEXT: [[DOTV_I:%.*]] = select i1 [[CMP1_I]], double 6.000000e+00, double [[TMP14]]
; CHECK-NEXT: [[RETVAL_0_I:%.*]] = select i1 [[CMP_I]], double 0.000000e+00, double [[DOTV_I]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP15]]
; CHECK-NEXT: store double [[RETVAL_0_I]], ptr [[ARRAYIDX2]], align 8
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 20000
@@ -148,19 +150,20 @@ define void @loop2(ptr %A, ptr %B, ptr %C, float %x) {
; 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 nuw i32, ptr [[C]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP12]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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]], splat (i32 20)
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq <4 x i32> [[WIDE_LOAD7]], splat (i32 20)
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP12]]
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw 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> [[BROADCAST_SPLAT]], [[WIDE_LOAD8]]
; CHECK-NEXT: [[TMP7:%.*]] = fmul <4 x float> [[BROADCAST_SPLAT]], [[WIDE_LOAD9]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP12]]
; 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]]
@@ -175,13 +178,14 @@ define void @loop2(ptr %A, ptr %B, ptr %C, float %x) {
; CHECK-NEXT: br i1 [[TMP13]], label [[EXIT:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: loop.body:
; CHECK-NEXT: [[IV1:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT: [[C_GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[C]], i64 [[IV1]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nuw nsw i64 [[IV1]], 2
+; CHECK-NEXT: [[C_GEP:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP14]]
; CHECK-NEXT: [[C_LV:%.*]] = load i32, ptr [[C_GEP]], align 4
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[C_LV]], 20
-; CHECK-NEXT: [[A_GEP_0:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[IV1]]
+; CHECK-NEXT: [[A_GEP_0:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP14]]
; CHECK-NEXT: [[A_LV_0:%.*]] = load float, ptr [[A_GEP_0]], align 4
; CHECK-NEXT: [[MUL2_I81_I:%.*]] = fmul float [[X]], [[A_LV_0]]
-; CHECK-NEXT: [[B_GEP_0:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[IV1]]
+; CHECK-NEXT: [[B_GEP_0:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP14]]
; CHECK-NEXT: br i1 [[CMP]], label [[LOOP_LATCH]], label [[ELSE:%.*]]
; CHECK: else:
; CHECK-NEXT: [[B_LV:%.*]] = load float, ptr [[B_GEP_0]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/indvars-vectorization.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/indvars-vectorization.ll
index 8d20a3ba8ed08..a44cdf3f31447 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/indvars-vectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/indvars-vectorization.ll
@@ -14,72 +14,74 @@ define void @s172(i32 noundef %xa, i32 noundef %xb, ptr noundef %a, ptr noundef
; CHECK-NEXT: [[SUB:%.*]] = add i32 [[XA]], -1
; CHECK-NEXT: [[TMP0:%.*]] = sext i32 [[SUB]] to i64
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[XB]] to i64
-; CHECK-NEXT: [[TMP2:%.*]] = tail call i64 @llvm.smax.i64(i64 [[TMP0]], i64 31999)
-; CHECK-NEXT: [[SMAX10:%.*]] = add nuw nsw i64 [[TMP2]], 1
-; CHECK-NEXT: [[TMP8:%.*]] = sub i64 [[SMAX10]], [[TMP0]]
-; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ugt i64 [[TMP8]], 23
+; CHECK-NEXT: [[TMP22:%.*]] = tail call i64 @llvm.smax.i64(i64 [[TMP0]], i64 31999)
+; CHECK-NEXT: [[SMAX10:%.*]] = add nuw nsw i64 [[TMP22]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = sub i64 [[SMAX10]], [[TMP0]]
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ugt i64 [[TMP3]], 19
; CHECK-NEXT: [[IDENT_CHECK_NOT:%.*]] = icmp eq i32 [[XB]], 1
; CHECK-NEXT: [[OR_COND:%.*]] = and i1 [[MIN_ITERS_CHECK]], [[IDENT_CHECK_NOT]]
-; CHECK-NEXT: br i1 [[OR_COND]], label [[VECTOR_MEMCHECK:%.*]], label [[FOR_BODY_PREHEADER13:%.*]]
+; CHECK-NEXT: br i1 [[OR_COND]], label [[VECTOR_MEMCHECK:%.*]], label [[FOR_BODY_PREHEADER14:%.*]]
; CHECK: vector.memcheck:
-; CHECK-NEXT: [[TMP9:%.*]] = shl nsw i64 [[TMP0]], 2
-; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP9]]
-; CHECK-NEXT: [[TMP10:%.*]] = add nsw i64 [[TMP1]], [[TMP0]]
-; CHECK-NEXT: [[SMAX:%.*]] = tail call i64 @llvm.smax.i64(i64 [[TMP10]], i64 32000)
-; CHECK-NEXT: [[TMP11:%.*]] = icmp slt i64 [[TMP10]], 32000
-; CHECK-NEXT: [[UMIN:%.*]] = zext i1 [[TMP11]] to i64
-; CHECK-NEXT: [[TMP12:%.*]] = add nsw i64 [[TMP10]], [[UMIN]]
-; CHECK-NEXT: [[TMP13:%.*]] = sub i64 [[SMAX]], [[TMP12]]
-; CHECK-NEXT: [[TMP14:%.*]] = add i64 [[TMP13]], [[UMIN]]
-; CHECK-NEXT: [[TMP15:%.*]] = add i64 [[TMP14]], [[TMP0]]
-; CHECK-NEXT: [[TMP16:%.*]] = shl i64 [[TMP15]], 2
-; CHECK-NEXT: [[TMP17:%.*]] = add i64 [[TMP16]], 4
-; CHECK-NEXT: [[SCEVGEP4:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP17]]
-; CHECK-NEXT: [[SCEVGEP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP9]]
-; CHECK-NEXT: [[SCEVGEP6:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP17]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[SCEVGEP:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP5:%.*]] = add nsw i64 [[TMP1]], [[TMP0]]
+; CHECK-NEXT: [[SMAX:%.*]] = tail call i64 @llvm.smax.i64(i64 [[TMP5]], i64 32000)
+; CHECK-NEXT: [[TMP6:%.*]] = icmp slt i64 [[TMP5]], 32000
+; CHECK-NEXT: [[UMIN:%.*]] = zext i1 [[TMP6]] to i64
+; CHECK-NEXT: [[TMP7:%.*]] = add nsw i64 [[TMP5]], [[UMIN]]
+; CHECK-NEXT: [[TMP8:%.*]] = sub i64 [[SMAX]], [[TMP7]]
+; CHECK-NEXT: [[TMP9:%.*]] = add i64 [[TMP8]], [[UMIN]]
+; CHECK-NEXT: [[TMP10:%.*]] = add i64 [[TMP9]], [[TMP0]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl i64 [[TMP10]], 2
+; CHECK-NEXT: [[TMP12:%.*]] = add i64 [[TMP11]], 4
+; CHECK-NEXT: [[SCEVGEP4:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP12]]
+; CHECK-NEXT: [[SCEVGEP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP4]]
+; CHECK-NEXT: [[SCEVGEP6:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP12]]
; CHECK-NEXT: [[BOUND0:%.*]] = icmp ult ptr [[SCEVGEP]], [[SCEVGEP6]]
; CHECK-NEXT: [[BOUND1:%.*]] = icmp ult ptr [[SCEVGEP5]], [[SCEVGEP4]]
; CHECK-NEXT: [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
-; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[FOR_BODY_PREHEADER13]], label [[VECTOR_PH:%.*]]
+; CHECK-NEXT: br i1 [[FOUND_CONFLICT]], label [[FOR_BODY_PREHEADER14]], label [[VECTOR_PH:%.*]]
; CHECK: vector.ph:
-; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP8]], -8
-; CHECK-NEXT: [[IND_END:%.*]] = add i64 [[N_VEC]], [[TMP0]]
-; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
+; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP3]], -8
+; CHECK-NEXT: [[TMP13:%.*]] = add i64 [[N_VEC]], [[TMP0]]
+; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[FOR_BODY]] ]
; CHECK-NEXT: [[OFFSET_IDX:%.*]] = add i64 [[INDEX]], [[TMP0]]
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP20]], i64 16
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP20]], align 4, !alias.scope [[META0:![0-9]+]]
-; CHECK-NEXT: [[WIDE_LOAD10:%.*]] = load <4 x i32>, ptr [[TMP21]], align 4, !alias.scope [[META0]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP22]], i64 16
-; CHECK-NEXT: [[WIDE_LOAD11:%.*]] = load <4 x i32>, ptr [[TMP22]], align 4, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
-; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[TMP23]], align 4, !alias.scope [[META3]], !noalias [[META0]]
-; CHECK-NEXT: [[TMP24:%.*]] = add nsw <4 x i32> [[WIDE_LOAD11]], [[WIDE_LOAD]]
-; CHECK-NEXT: [[TMP25:%.*]] = add nsw <4 x i32> [[WIDE_LOAD12]], [[WIDE_LOAD10]]
-; CHECK-NEXT: store <4 x i32> [[TMP24]], ptr [[TMP22]], align 4, !alias.scope [[META3]], !noalias [[META0]]
-; CHECK-NEXT: store <4 x i32> [[TMP25]], ptr [[TMP23]], align 4, !alias.scope [[META3]], !noalias [[META0]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw i64 [[OFFSET_IDX]], 2
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP15]], i64 16
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP15]], align 4, !alias.scope [[META0:![0-9]+]]
+; CHECK-NEXT: [[WIDE_LOAD11:%.*]] = load <4 x i32>, ptr [[TMP16]], align 4, !alias.scope [[META0]]
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP14]]
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP17]], i64 16
+; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[TMP17]], align 4, !alias.scope [[META3:![0-9]+]], !noalias [[META0]]
+; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <4 x i32>, ptr [[TMP18]], align 4, !alias.scope [[META3]], !noalias [[META0]]
+; CHECK-NEXT: [[TMP19:%.*]] = add nsw <4 x i32> [[WIDE_LOAD12]], [[WIDE_LOAD]]
+; CHECK-NEXT: [[TMP20:%.*]] = add nsw <4 x i32> [[WIDE_LOAD13]], [[WIDE_LOAD11]]
+; CHECK-NEXT: store <4 x i32> [[TMP19]], ptr [[TMP17]], align 4, !alias.scope [[META3]], !noalias [[META0]]
+; CHECK-NEXT: store <4 x i32> [[TMP20]], ptr [[TMP18]], align 4, !alias.scope [[META3]], !noalias [[META0]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; CHECK-NEXT: [[TMP26:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP26]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK-NEXT: [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK: middle.block:
-; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP8]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END]], label [[FOR_BODY_PREHEADER13]]
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP3]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END]], label [[FOR_BODY_PREHEADER14]]
; CHECK: for.body.preheader14:
-; CHECK-NEXT: [[INDVARS_IV_PH:%.*]] = phi i64 [ [[TMP0]], [[VECTOR_MEMCHECK]] ], [ [[TMP0]], [[FOR_BODY_PREHEADER]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT: br label [[FOR_BODY:%.*]]
+; CHECK-NEXT: [[INDVARS_IV_PH:%.*]] = phi i64 [ [[TMP0]], [[VECTOR_MEMCHECK]] ], [ [[TMP0]], [[FOR_BODY_PREHEADER]] ], [ [[TMP13]], [[MIDDLE_BLOCK]] ]
+; CHECK-NEXT: br label [[FOR_BODY1:%.*]]
; CHECK: for.body:
-; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[INDVARS_IV_PH]], [[FOR_BODY_PREHEADER13]] ]
-; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY1]] ], [ [[INDVARS_IV_PH]], [[FOR_BODY_PREHEADER14]] ]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP2]]
; CHECK-NEXT: [[L_B:%.*]] = load i32, ptr [[GEP_B]], align 4
-; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: [[L_A:%.*]] = load i32, ptr [[GEP_A]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[L_A]], [[L_B]]
; CHECK-NEXT: store i32 [[ADD]], ptr [[GEP_A]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nsw i64 [[INDVARS_IV]], [[TMP1]]
; CHECK-NEXT: [[CMP:%.*]] = icmp slt i64 [[INDVARS_IV_NEXT]], 32000
-; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY]], label [[FOR_END]], !llvm.loop [[LOOP9:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY1]], label [[FOR_END]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK: for.end:
; CHECK-NEXT: ret void
;
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/interleave_vec.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/interleave_vec.ll
index a9280b7baaf2f..842d633b04f3c 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/interleave_vec.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/interleave_vec.ll
@@ -11,18 +11,19 @@ define void @same_op2(ptr noalias noundef %a, ptr noundef %b, ptr noundef %c) {
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP0:%.*]] = or disjoint i64 [[OFFSET_IDX]], 8
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[TMP0]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = or disjoint i64 [[OFFSET_IDX]], 32
+; CHECK-NEXT: [[TMP10:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP0]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x float>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[WIDE_VEC15:%.*]] = load <8 x float>, ptr [[TMP2]], align 4
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP0]]
; CHECK-NEXT: [[WIDE_VEC18:%.*]] = load <8 x float>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[WIDE_VEC21:%.*]] = load <8 x float>, ptr [[TMP4]], align 4
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP10]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP0]]
; CHECK-NEXT: [[WIDE_VEC24:%.*]] = load <8 x float>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[WIDE_VEC27:%.*]] = load <8 x float>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP7:%.*]] = fmul fast <8 x float> [[WIDE_VEC18]], [[WIDE_VEC]]
@@ -123,14 +124,15 @@ define void @same_op2_splat(ptr noalias noundef %a, ptr noundef %b, ptr noundef
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 8
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP3]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 32
+; CHECK-NEXT: [[TMP11:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP3]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <8 x float>, ptr [[TMP4]], align 4
; CHECK-NEXT: [[WIDE_VEC13:%.*]] = load <8 x float>, ptr [[TMP5]], align 4
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP11]]
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP3]]
; CHECK-NEXT: [[WIDE_VEC16:%.*]] = load <8 x float>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[WIDE_VEC19:%.*]] = load <8 x float>, ptr [[TMP7]], align 4
; CHECK-NEXT: [[TMP8:%.*]] = fmul fast <8 x float> [[WIDE_VEC]], [[TMP1]]
@@ -223,12 +225,12 @@ define void @same_op3(ptr noalias noundef %a, ptr noundef %b, ptr noundef %c) {
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP5:%.*]] = mul i64 [[INDEX]], 12
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP5]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <12 x float>, ptr [[TMP0]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP5]]
; CHECK-NEXT: [[WIDE_VEC16:%.*]] = load <12 x float>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT: [[WIDE_VEC20:%.*]] = load <12 x float>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[TMP3:%.*]] = fmul fast <12 x float> [[WIDE_VEC16]], [[WIDE_VEC]]
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = fadd fast <12 x float> [[WIDE_VEC20]], [[TMP3]]
@@ -324,10 +326,10 @@ define void @same_op3_splat(ptr noalias noundef %a, ptr noundef %b, ptr noundef
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP5:%.*]] = mul i64 [[INDEX]], 12
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP5]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <12 x float>, ptr [[TMP1]], align 4
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT: [[WIDE_VEC14:%.*]] = load <12 x float>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = fmul fast <12 x float> [[WIDE_VEC]], [[TMP2]]
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = fadd fast <12 x float> [[WIDE_VEC14]], [[TMP4]]
@@ -416,22 +418,25 @@ define void @same_op4(ptr noalias noundef %a, ptr noundef %b, ptr noundef %c) {
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[OFFSET_IDX]], 4
-; CHECK-NEXT: [[TMP7:%.*]] = or disjoint i64 [[OFFSET_IDX]], 8
-; CHECK-NEXT: [[TMP15:%.*]] = or disjoint i64 [[OFFSET_IDX]], 12
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[TMP7]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[TMP15]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP7:%.*]] = or disjoint i64 [[OFFSET_IDX]], 48
+; CHECK-NEXT: [[TMP15:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP24:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP3:%.*]] = or disjoint i64 [[TMP24]], 16
+; CHECK-NEXT: [[TMP25:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP26:%.*]] = or disjoint i64 [[TMP25]], 32
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP26]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP0]], align 4
; CHECK-NEXT: [[WIDE_LOAD14:%.*]] = load <4 x float>, ptr [[TMP23]], align 4
; CHECK-NEXT: [[WIDE_LOAD15:%.*]] = load <4 x float>, ptr [[TMP5]], align 4
; CHECK-NEXT: [[WIDE_LOAD16:%.*]] = load <4 x float>, ptr [[TMP6]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP7]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP26]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD17:%.*]] = load <4 x float>, ptr [[TMP1]], align 4
; CHECK-NEXT: [[WIDE_LOAD18:%.*]] = load <4 x float>, ptr [[TMP8]], align 4
; CHECK-NEXT: [[WIDE_LOAD19:%.*]] = load <4 x float>, ptr [[TMP9]], align 4
@@ -440,10 +445,10 @@ define void @same_op4(ptr noalias noundef %a, ptr noundef %b, ptr noundef %c) {
; CHECK-NEXT: [[TMP12:%.*]] = fmul fast <4 x float> [[WIDE_LOAD18]], [[WIDE_LOAD14]]
; CHECK-NEXT: [[TMP13:%.*]] = fmul fast <4 x float> [[WIDE_LOAD19]], [[WIDE_LOAD15]]
; CHECK-NEXT: [[TMP14:%.*]] = fmul fast <4 x float> [[WIDE_LOAD20]], [[WIDE_LOAD16]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP7]]
-; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP15]]
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP26]]
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP7]]
; CHECK-NEXT: [[WIDE_LOAD21:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[WIDE_LOAD22:%.*]] = load <4 x float>, ptr [[TMP16]], align 4
; CHECK-NEXT: [[WIDE_LOAD23:%.*]] = load <4 x float>, ptr [[TMP17]], align 4
@@ -547,14 +552,17 @@ define void @same_op4_splat(ptr noalias noundef %a, ptr noundef %b, ptr noundef
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP1:%.*]] = or disjoint i64 [[OFFSET_IDX]], 4
-; CHECK-NEXT: [[TMP4:%.*]] = or disjoint i64 [[OFFSET_IDX]], 8
-; CHECK-NEXT: [[TMP12:%.*]] = or disjoint i64 [[OFFSET_IDX]], 12
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP12]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP1:%.*]] = or disjoint i64 [[OFFSET_IDX]], 48
+; CHECK-NEXT: [[TMP12:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP21:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP4:%.*]] = or disjoint i64 [[TMP21]], 16
+; CHECK-NEXT: [[TMP22:%.*]] = shl i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP23:%.*]] = or disjoint i64 [[TMP22]], 32
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP23]]
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP1]]
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP2]], align 4
; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x float>, ptr [[TMP20]], align 4
; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <4 x float>, ptr [[TMP6]], align 4
@@ -563,10 +571,10 @@ define void @same_op4_splat(ptr noalias noundef %a, ptr noundef %b, ptr noundef
; CHECK-NEXT: [[TMP9:%.*]] = fmul fast <4 x float> [[WIDE_LOAD12]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP10:%.*]] = fmul fast <4 x float> [[WIDE_LOAD13]], [[BROADCAST_SPLAT]]
; CHECK-NEXT: [[TMP11:%.*]] = fmul fast <4 x float> [[WIDE_LOAD14]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP12]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP23]]
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP1]]
; CHECK-NEXT: [[WIDE_LOAD15:%.*]] = load <4 x float>, ptr [[TMP3]], align 4
; CHECK-NEXT: [[WIDE_LOAD16:%.*]] = load <4 x float>, ptr [[TMP13]], align 4
; CHECK-NEXT: [[WIDE_LOAD17:%.*]] = load <4 x float>, ptr [[TMP14]], align 4
@@ -662,29 +670,20 @@ define void @same_op6(ptr noalias noundef %a, ptr noundef %b, ptr noundef %c) {
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: br label %[[FOR_COND1_PREHEADER:.*]]
; CHECK: [[FOR_COND1_PREHEADER]]:
-; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[ARRAYIDX6]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = fmul fast <4 x float> [[TMP1]], [[TMP0]]
-; CHECK-NEXT: [[TMP3:%.*]] = load <4 x float>, ptr [[ARRAYIDX9]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = fadd fast <4 x float> [[TMP3]], [[TMP2]]
-; CHECK-NEXT: store <4 x float> [[TMP4]], ptr [[ARRAYIDX9]], align 4
-; CHECK-NEXT: [[TMP5:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 4
-; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[TMP5]]
-; CHECK-NEXT: [[ARRAYIDX6_4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP5]]
-; CHECK-NEXT: [[ARRAYIDX9_4:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP6:%.*]] = load <2 x float>, ptr [[ARRAYIDX_4]], align 4
-; CHECK-NEXT: [[TMP7:%.*]] = load <2 x float>, ptr [[ARRAYIDX6_4]], align 4
-; CHECK-NEXT: [[TMP8:%.*]] = fmul fast <2 x float> [[TMP7]], [[TMP6]]
-; CHECK-NEXT: [[TMP9:%.*]] = load <2 x float>, ptr [[ARRAYIDX9_4]], align 4
-; CHECK-NEXT: [[TMP10:%.*]] = fadd fast <2 x float> [[TMP9]], [[TMP8]]
-; CHECK-NEXT: store <2 x float> [[TMP10]], ptr [[ARRAYIDX9_4]], align 4
-; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 6
-; CHECK-NEXT: [[CMP:%.*]] = icmp samesign ult i64 [[INDVARS_IV]], 1146
-; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_COND1_PREHEADER]], label %[[FOR_END13:.*]]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = mul i64 [[INDEX]], 24
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <12 x float>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_VEC19:%.*]] = load <12 x float>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP0]]
+; CHECK-NEXT: [[WIDE_VEC26:%.*]] = load <12 x float>, ptr [[TMP3]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = fmul fast <12 x float> [[WIDE_VEC19]], [[WIDE_VEC]]
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = fadd fast <12 x float> [[WIDE_VEC26]], [[TMP4]]
+; CHECK-NEXT: store <12 x float> [[INTERLEAVED_VEC]], ptr [[TMP3]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 192
+; CHECK-NEXT: br i1 [[TMP5]], label %[[FOR_END13:.*]], label %[[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK: [[FOR_END13]]:
; CHECK-NEXT: ret void
;
@@ -768,31 +767,22 @@ define void @same_op6_splat(ptr noalias noundef %a, ptr noundef %b, ptr noundef
; CHECK-SAME: ptr noalias noundef captures(none) [[A:%.*]], ptr noundef readonly captures(none) [[B:%.*]], ptr noundef readonly captures(none) [[C:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[C]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x float> poison, float [[TMP0]], i64 0
-; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <4 x float> [[TMP1]], <4 x float> poison, <4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x float> poison, float [[TMP0]], i64 0
-; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <2 x float> [[TMP3]], <2 x float> poison, <2 x i32> zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = shufflevector <2 x float> [[TMP3]], <2 x float> poison, <12 x i32> zeroinitializer
; CHECK-NEXT: br label %[[FOR_COND1_PREHEADER:.*]]
; CHECK: [[FOR_COND1_PREHEADER]]:
-; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ]
-; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[TMP5:%.*]] = load <4 x float>, ptr [[ARRAYIDX4]], align 4
-; CHECK-NEXT: [[TMP6:%.*]] = fmul fast <4 x float> [[TMP5]], [[TMP2]]
-; CHECK-NEXT: [[TMP7:%.*]] = load <4 x float>, ptr [[ARRAYIDX7]], align 4
-; CHECK-NEXT: [[TMP8:%.*]] = fadd fast <4 x float> [[TMP7]], [[TMP6]]
-; CHECK-NEXT: store <4 x float> [[TMP8]], ptr [[ARRAYIDX7]], align 4
-; CHECK-NEXT: [[TMP9:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 4
-; CHECK-NEXT: [[ARRAYIDX4_4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP9]]
-; CHECK-NEXT: [[ARRAYIDX7_4:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP9]]
-; CHECK-NEXT: [[TMP10:%.*]] = load <2 x float>, ptr [[ARRAYIDX4_4]], align 4
-; CHECK-NEXT: [[TMP11:%.*]] = fmul fast <2 x float> [[TMP10]], [[TMP4]]
-; CHECK-NEXT: [[TMP12:%.*]] = load <2 x float>, ptr [[ARRAYIDX7_4]], align 4
-; CHECK-NEXT: [[TMP13:%.*]] = fadd fast <2 x float> [[TMP12]], [[TMP11]]
-; CHECK-NEXT: store <2 x float> [[TMP13]], ptr [[ARRAYIDX7_4]], align 4
-; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 6
-; CHECK-NEXT: [[CMP:%.*]] = icmp samesign ult i64 [[INDVARS_IV]], 1146
-; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_COND1_PREHEADER]], label %[[FOR_END11:.*]]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ]
+; CHECK-NEXT: [[TMP2:%.*]] = mul i64 [[INDEX]], 24
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP2]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <12 x float>, ptr [[TMP7]], align 4
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP2]]
+; CHECK-NEXT: [[WIDE_VEC17:%.*]] = load <12 x float>, ptr [[TMP4]], align 4
+; CHECK-NEXT: [[TMP5:%.*]] = fmul fast <12 x float> [[WIDE_VEC]], [[TMP1]]
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = fadd fast <12 x float> [[WIDE_VEC17]], [[TMP5]]
+; CHECK-NEXT: store <12 x float> [[INTERLEAVED_VEC]], ptr [[TMP4]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], 192
+; CHECK-NEXT: br i1 [[TMP6]], label %[[FOR_END11:.*]], label %[[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP9:![0-9]+]]
; CHECK: [[FOR_END11]]:
; CHECK-NEXT: ret void
;
@@ -873,29 +863,20 @@ define void @same_op8(ptr noalias noundef %a, ptr noundef %b, ptr noundef %c) {
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: br label %[[FOR_COND1_PREHEADER:.*]]
; CHECK: [[FOR_COND1_PREHEADER]]:
-; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT: [[TMP0:%.*]] = load <4 x float>, ptr [[ARRAYIDX]], align 4
-; CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, ptr [[ARRAYIDX6]], align 4
-; CHECK-NEXT: [[TMP2:%.*]] = fmul fast <4 x float> [[TMP1]], [[TMP0]]
-; CHECK-NEXT: [[TMP3:%.*]] = load <4 x float>, ptr [[ARRAYIDX9]], align 4
-; CHECK-NEXT: [[TMP4:%.*]] = fadd fast <4 x float> [[TMP3]], [[TMP2]]
-; CHECK-NEXT: store <4 x float> [[TMP4]], ptr [[ARRAYIDX9]], align 4
-; CHECK-NEXT: [[TMP5:%.*]] = or disjoint i64 [[INDVARS_IV]], 4
-; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[TMP5]]
-; CHECK-NEXT: [[ARRAYIDX6_4:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[TMP5]]
-; CHECK-NEXT: [[ARRAYIDX9_4:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP6:%.*]] = load <4 x float>, ptr [[ARRAYIDX_4]], align 4
-; CHECK-NEXT: [[TMP7:%.*]] = load <4 x float>, ptr [[ARRAYIDX6_4]], align 4
-; CHECK-NEXT: [[TMP8:%.*]] = fmul fast <4 x float> [[TMP7]], [[TMP6]]
-; CHECK-NEXT: [[TMP9:%.*]] = load <4 x float>, ptr [[ARRAYIDX9_4]], align 4
-; CHECK-NEXT: [[TMP10:%.*]] = fadd fast <4 x float> [[TMP9]], [[TMP8]]
-; CHECK-NEXT: store <4 x float> [[TMP10]], ptr [[ARRAYIDX9_4]], align 4
-; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 8
-; CHECK-NEXT: [[CMP:%.*]] = icmp samesign ult i64 [[INDVARS_IV]], 1144
-; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_COND1_PREHEADER]], label %[[FOR_END13:.*]]
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[FOR_COND1_PREHEADER]] ]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 5
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <16 x float>, ptr [[TMP0]], align 4
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC21:%.*]] = load <16 x float>, ptr [[TMP1]], align 4
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[WIDE_VEC30:%.*]] = load <16 x float>, ptr [[TMP2]], align 4
+; CHECK-NEXT: [[TMP3:%.*]] = fmul fast <16 x float> [[WIDE_VEC21]], [[WIDE_VEC]]
+; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = fadd fast <16 x float> [[WIDE_VEC30]], [[TMP3]]
+; CHECK-NEXT: store <16 x float> [[INTERLEAVED_VEC]], ptr [[TMP2]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], 144
+; CHECK-NEXT: br i1 [[TMP4]], label %[[FOR_END13:.*]], label %[[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK: [[FOR_END13]]:
; CHECK-NEXT: ret void
;
@@ -984,17 +965,17 @@ define void @same_op8_splat(ptr noalias noundef %a, ptr noundef %b, ptr noundef
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 5
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <16 x float>, ptr [[TMP5]], align 4
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC19:%.*]] = load <16 x float>, ptr [[TMP6]], align 4
; CHECK-NEXT: [[TMP4:%.*]] = fmul fast <16 x float> [[WIDE_VEC]], [[TMP1]]
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = fadd fast <16 x float> [[WIDE_VEC19]], [[TMP4]]
; CHECK-NEXT: store <16 x float> [[INTERLEAVED_VEC]], ptr [[TMP6]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP25:%.*]] = icmp eq i64 [[INDEX_NEXT]], 144
-; CHECK-NEXT: br i1 [[TMP25]], label %[[FOR_END11:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP25]], label %[[FOR_END11:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK: [[FOR_END11]]:
; CHECK-NEXT: ret void
;
@@ -1100,17 +1081,17 @@ define void @saxpy_5(i64 %n, float %a, ptr readonly %x, ptr noalias %y) {
; 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:%.*]] = mul i64 [[INDEX]], 5
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw float, ptr [[X]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP16:%.*]] = mul i64 [[INDEX]], 20
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP16]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <10 x float>, ptr [[TMP6]], align 4
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw float, ptr [[Y]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP16]]
; CHECK-NEXT: [[WIDE_VEC5:%.*]] = load <10 x float>, ptr [[TMP7]], align 4
; CHECK-NEXT: [[TMP8:%.*]] = fmul fast <10 x float> [[WIDE_VEC]], [[TMP5]]
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = fadd fast <10 x float> [[WIDE_VEC5]], [[TMP8]]
; CHECK-NEXT: store <10 x float> [[INTERLEAVED_VEC]], ptr [[TMP7]], align 4
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: [[MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP3]], [[N_VEC]]
; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT]], label %[[LOOP_PREHEADER11]]
@@ -1121,24 +1102,25 @@ define void @saxpy_5(i64 %n, float %a, ptr readonly %x, ptr noalias %y) {
; CHECK-NEXT: br label %[[LOOP:.*]]
; CHECK: [[LOOP]]:
; CHECK-NEXT: [[I1:%.*]] = phi i64 [ [[I_NEXT:%.*]], %[[LOOP]] ], [ [[I1_PH]], %[[LOOP_PREHEADER11]] ]
-; CHECK-NEXT: [[XGEP1:%.*]] = getelementptr inbounds nuw float, ptr [[X]], i64 [[I1]]
-; CHECK-NEXT: [[YGEP1:%.*]] = getelementptr inbounds nuw float, ptr [[Y]], i64 [[I1]]
+; CHECK-NEXT: [[TMP17:%.*]] = shl i64 [[I1]], 2
+; CHECK-NEXT: [[XGEP1:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP17]]
+; CHECK-NEXT: [[YGEP1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP17]]
; CHECK-NEXT: [[TMP12:%.*]] = load <4 x float>, ptr [[XGEP1]], align 4
; CHECK-NEXT: [[TMP13:%.*]] = fmul fast <4 x float> [[TMP12]], [[TMP11]]
; CHECK-NEXT: [[TMP14:%.*]] = load <4 x float>, ptr [[YGEP1]], align 4
; CHECK-NEXT: [[TMP15:%.*]] = fadd fast <4 x float> [[TMP14]], [[TMP13]]
; CHECK-NEXT: store <4 x float> [[TMP15]], ptr [[YGEP1]], align 4
-; CHECK-NEXT: [[I5:%.*]] = add nuw nsw i64 [[I1]], 4
-; CHECK-NEXT: [[XGEP5:%.*]] = getelementptr inbounds nuw float, ptr [[X]], i64 [[I5]]
+; CHECK-NEXT: [[TMP18:%.*]] = add nuw i64 [[TMP17]], 16
+; CHECK-NEXT: [[XGEP5:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP18]]
; CHECK-NEXT: [[X5:%.*]] = load float, ptr [[XGEP5]], align 4
; CHECK-NEXT: [[AX5:%.*]] = fmul fast float [[X5]], [[A]]
-; CHECK-NEXT: [[YGEP5:%.*]] = getelementptr inbounds nuw float, ptr [[Y]], i64 [[I5]]
+; CHECK-NEXT: [[YGEP5:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP18]]
; CHECK-NEXT: [[Y5:%.*]] = load float, ptr [[YGEP5]], align 4
; CHECK-NEXT: [[AXPY5:%.*]] = fadd fast float [[Y5]], [[AX5]]
; CHECK-NEXT: store float [[AXPY5]], ptr [[YGEP5]], align 4
; CHECK-NEXT: [[I_NEXT]] = add nuw nsw i64 [[I1]], 5
; CHECK-NEXT: [[CMP:%.*]] = icmp sgt i64 [[N]], [[I_NEXT]]
-; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP10:![0-9]+]]
+; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP13:![0-9]+]]
; CHECK: [[EXIT]]:
; CHECK-NEXT: ret void
;
@@ -1205,5 +1187,8 @@ exit:
; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META1]], [[META2]]}
; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META1]], [[META2]]}
; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META1]], [[META2]]}
-; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META2]], [[META1]]}
+; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META1]], [[META2]]}
+; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META1]], [[META2]]}
+; CHECK: [[LOOP12]] = distinct !{[[LOOP12]], [[META1]], [[META2]]}
+; CHECK: [[LOOP13]] = distinct !{[[LOOP13]], [[META2]], [[META1]]}
;.
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll
index be57809132d3f..de581146f55f9 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/interleavevectorization.ll
@@ -16,10 +16,10 @@ define void @add4(ptr noalias noundef %x, ptr noalias noundef %y, i32 noundef %n
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC24:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
; CHECK-NEXT: store <32 x i16> [[INTERLEAVED_VEC]], ptr [[TMP1]], align 2
@@ -136,10 +136,10 @@ define void @addsubs(ptr noalias noundef %x, ptr noundef %y, i32 noundef %n) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC24:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
; CHECK-NEXT: [[TMP3:%.*]] = sub <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
@@ -262,10 +262,10 @@ define void @add2sub2(ptr noalias noundef %x, ptr noundef %y, i32 noundef %n) {
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC24:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
; CHECK-NEXT: [[TMP3:%.*]] = add <32 x i16> [[WIDE_VEC24]], [[WIDE_VEC]]
@@ -388,12 +388,12 @@ define void @addmul(ptr noalias noundef %x, ptr noundef %y, ptr noundef %z, i32
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC31:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC36:%.*]] = load <32 x i16>, ptr [[TMP2]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
; CHECK-NEXT: [[INTERLEAVED_VEC:%.*]] = add <32 x i16> [[TMP4]], [[WIDE_VEC36]]
@@ -544,13 +544,13 @@ define void @addsubsmul(ptr noalias noundef %x, ptr noundef %y, ptr noundef %z,
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC31:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC36:%.*]] = load <32 x i16>, ptr [[TMP3]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = add <32 x i16> [[TMP2]], [[WIDE_VEC36]]
; CHECK-NEXT: [[TMP5:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
@@ -709,13 +709,13 @@ define void @add2sub2mul(ptr noalias noundef %x, ptr noundef %y, ptr noundef %z,
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[OFFSET_IDX:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr [[Y:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC:%.*]] = load <32 x i16>, ptr [[TMP0]], align 2
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i16, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[Z:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC31:%.*]] = load <32 x i16>, ptr [[TMP1]], align 2
; CHECK-NEXT: [[TMP2:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[X:%.*]], i64 [[OFFSET_IDX]]
; CHECK-NEXT: [[WIDE_VEC36:%.*]] = load <32 x i16>, ptr [[TMP3]], align 2
; CHECK-NEXT: [[TMP4:%.*]] = add <32 x i16> [[TMP2]], [[WIDE_VEC36]]
; CHECK-NEXT: [[TMP5:%.*]] = mul <32 x i16> [[WIDE_VEC31]], [[WIDE_VEC]]
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/loopflatten.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/loopflatten.ll
index 2703d2390ce52..db4dcd2075a31 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/loopflatten.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/loopflatten.ll
@@ -18,7 +18,8 @@ define dso_local void @_Z3fooPiii(ptr %A, i32 %N, i32 %M) #0 {
; CHECK-NEXT: br label [[FOR_COND1_PREHEADER_US:%.*]]
; CHECK: for.cond1.preheader.us:
; CHECK-NEXT: [[INDVAR6:%.*]] = phi i64 [ [[INDVAR_NEXT7:%.*]], [[FOR_COND1_PREHEADER_US]] ], [ 0, [[FOR_COND1_PREHEADER_LR_PH_SPLIT_US]] ]
-; CHECK-NEXT: [[ARRAYIDX_US:%.*]] = getelementptr inbounds nuw i32, ptr [[A:%.*]], i64 [[INDVAR6]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[INDVAR6]], 2
+; CHECK-NEXT: [[ARRAYIDX_US:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARRAYIDX_US]], align 4
; CHECK-NEXT: tail call void @_Z1fi(i32 [[TMP2]])
; CHECK-NEXT: [[INDVAR_NEXT7]] = add nuw i64 [[INDVAR6]], 1
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll
index 75276c0412647..f19f5fe0928f0 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/matrix-extract-insert.ll
@@ -12,16 +12,18 @@ define void @matrix_extract_insert_scalar(i32 %i, i32 %k, i32 %j, ptr nonnull al
; CHECK-NEXT: [[TMP1:%.*]] = add nuw nsw i64 [[TMP0]], [[CONV]]
; CHECK-NEXT: [[TMP2:%.*]] = icmp samesign ult i64 [[TMP1]], 225
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP2]])
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw double, ptr [[A:%.*]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[TMP1]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP8]]
; CHECK-NEXT: [[MATRIXEXT:%.*]] = load double, ptr [[TMP3]], align 8
; CHECK-NEXT: [[CONV2:%.*]] = zext i32 [[I:%.*]] to i64
; CHECK-NEXT: [[TMP4:%.*]] = add nuw nsw i64 [[TMP0]], [[CONV2]]
; CHECK-NEXT: [[TMP5:%.*]] = icmp samesign ult i64 [[TMP4]], 225
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP5]])
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw double, ptr [[B:%.*]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP9:%.*]] = shl nuw nsw i64 [[TMP4]], 3
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[B:%.*]], i64 [[TMP9]]
; CHECK-NEXT: [[MATRIXEXT4:%.*]] = load double, ptr [[TMP6]], align 8
; CHECK-NEXT: [[MUL:%.*]] = fmul double [[MATRIXEXT]], [[MATRIXEXT4]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP8]]
; CHECK-NEXT: [[MATRIXEXT7:%.*]] = load double, ptr [[TMP7]], align 8
; CHECK-NEXT: [[SUB:%.*]] = fsub double [[MATRIXEXT7]], [[MUL]]
; CHECK-NEXT: store double [[SUB]], ptr [[TMP7]], align 8
@@ -93,7 +95,8 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[SCEVGEP20:%.*]] = getelementptr i8, ptr [[A:%.*]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = icmp ult i32 [[I]], 225
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP2]])
-; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[CONV6]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[CONV6]], 3
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP4]]
; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp samesign ult i32 [[I]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY4_US_PREHEADER:%.*]], label [[FOR_COND1_PREHEADER_US_PREHEADER:%.*]]
; CHECK: vector.memcheck:
@@ -106,7 +109,8 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP5:%.*]] = shl nuw nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP17]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x double>, ptr [[TMP17]], align 8, !alias.scope [[META0:![0-9]+]]
; CHECK-NEXT: [[WIDE_LOAD21:%.*]] = load <2 x double>, ptr [[TMP18]], align 8, !alias.scope [[META0]]
@@ -115,7 +119,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP20:%.*]] = fmul <2 x double> [[WIDE_LOAD]], [[BROADCAST_SPLAT23]]
; CHECK-NEXT: [[TMP21:%.*]] = fmul <2 x double> [[WIDE_LOAD21]], [[BROADCAST_SPLAT23]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP5]]
; CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP22]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24:%.*]] = load <2 x double>, ptr [[TMP22]], align 8, !alias.scope [[META5:![0-9]+]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25:%.*]] = load <2 x double>, ptr [[TMP23]], align 8, !alias.scope [[META5]], !noalias [[META0]]
@@ -136,11 +140,12 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY4_US]] ], [ [[INDVARS_IV_PH]], [[FOR_BODY4_US_PREHEADER]] ]
; CHECK-NEXT: [[TMP27:%.*]] = icmp samesign ult i64 [[INDVARS_IV]], 225
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP27]])
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP30]]
; CHECK-NEXT: [[MATRIXEXT_US:%.*]] = load double, ptr [[TMP28]], align 8
; CHECK-NEXT: [[MATRIXEXT8_US:%.*]] = load double, ptr [[TMP3]], align 8
; CHECK-NEXT: [[MUL_US:%.*]] = fmul double [[MATRIXEXT_US]], [[MATRIXEXT8_US]]
-; CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP30]]
; CHECK-NEXT: [[MATRIXEXT11_US:%.*]] = load double, ptr [[TMP29]], align 8
; CHECK-NEXT: [[SUB_US:%.*]] = fsub double [[MATRIXEXT11_US]], [[MUL_US]]
; CHECK-NEXT: store double [[SUB_US]], ptr [[TMP29]], align 8
@@ -150,7 +155,8 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK: for.cond1.for.cond.cleanup3_crit_edge.us:
; CHECK-NEXT: [[TMP31:%.*]] = icmp samesign ult i32 [[I]], 210
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP31]])
-; CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[CONV6]]
+; CHECK-NEXT: [[TMP33:%.*]] = shl nuw nsw i64 [[CONV6]], 3
+; CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP33]]
; CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP61]], i64 120
; CHECK-NEXT: [[MIN_ITERS_CHECK_1:%.*]] = icmp samesign ult i32 [[I]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK_1]], label [[FOR_BODY4_US_PREHEADER_1:%.*]], label [[FOR_COND1_FOR_COND_CLEANUP3_CRIT_EDGE_US:%.*]]
@@ -164,8 +170,9 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: br label [[VECTOR_BODY_1:%.*]]
; CHECK: vector.body.1:
; CHECK-NEXT: [[INDEX_1:%.*]] = phi i64 [ 0, [[VECTOR_PH_1]] ], [ [[INDEX_NEXT_1:%.*]], [[VECTOR_BODY_1]] ]
-; CHECK-NEXT: [[TMP33:%.*]] = add nuw nsw i64 [[INDEX_1]], 15
-; CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[TMP33]]
+; CHECK-NEXT: [[TMP34:%.*]] = shl i64 [[INDEX_1]], 3
+; CHECK-NEXT: [[TMP35:%.*]] = add i64 [[TMP34]], 120
+; CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP47]], i64 16
; CHECK-NEXT: [[WIDE_LOAD_1:%.*]] = load <2 x double>, ptr [[TMP47]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[WIDE_LOAD21_1:%.*]] = load <2 x double>, ptr [[TMP48]], align 8, !alias.scope [[META0]]
@@ -174,7 +181,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23_1:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22_1]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP50:%.*]] = fmul <2 x double> [[WIDE_LOAD_1]], [[BROADCAST_SPLAT23_1]]
; CHECK-NEXT: [[TMP51:%.*]] = fmul <2 x double> [[WIDE_LOAD21_1]], [[BROADCAST_SPLAT23_1]]
-; CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[TMP33]]
+; CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP35]]
; CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP52]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24_1:%.*]] = load <2 x double>, ptr [[TMP52]], align 8, !alias.scope [[META5]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25_1:%.*]] = load <2 x double>, ptr [[TMP53]], align 8, !alias.scope [[META5]], !noalias [[META0]]
@@ -193,14 +200,15 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: br label [[FOR_BODY4_US_1:%.*]]
; CHECK: for.body4.us.1:
; CHECK-NEXT: [[INDVARS_IV_1:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_1:%.*]], [[FOR_BODY4_US_1]] ], [ [[INDVARS_IV_PH_1]], [[FOR_BODY4_US_PREHEADER_1]] ]
-; CHECK-NEXT: [[TMP57:%.*]] = add nuw nsw i64 [[INDVARS_IV_1]], 15
; CHECK-NEXT: [[TMP58:%.*]] = icmp samesign ult i64 [[INDVARS_IV_1]], 210
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP58]])
-; CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[TMP57]]
+; CHECK-NEXT: [[TMP37:%.*]] = shl nuw nsw i64 [[INDVARS_IV_1]], 3
+; CHECK-NEXT: [[TMP38:%.*]] = add nuw nsw i64 [[TMP37]], 120
+; CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP38]]
; CHECK-NEXT: [[MATRIXEXT_US_1:%.*]] = load double, ptr [[TMP59]], align 8
; CHECK-NEXT: [[MATRIXEXT8_US_1:%.*]] = load double, ptr [[TMP32]], align 8
; CHECK-NEXT: [[MUL_US_1:%.*]] = fmul double [[MATRIXEXT_US_1]], [[MATRIXEXT8_US_1]]
-; CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[TMP57]]
+; CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP38]]
; CHECK-NEXT: [[MATRIXEXT11_US_1:%.*]] = load double, ptr [[TMP60]], align 8
; CHECK-NEXT: [[SUB_US_1:%.*]] = fsub double [[MATRIXEXT11_US_1]], [[MUL_US_1]]
; CHECK-NEXT: store double [[SUB_US_1]], ptr [[TMP60]], align 8
@@ -210,7 +218,8 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK: for.cond1.for.cond.cleanup3_crit_edge.us.1:
; CHECK-NEXT: [[TMP62:%.*]] = icmp samesign ult i32 [[I]], 195
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP62]])
-; CHECK-NEXT: [[TMP92:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[CONV6]]
+; CHECK-NEXT: [[TMP42:%.*]] = shl nuw nsw i64 [[CONV6]], 3
+; CHECK-NEXT: [[TMP92:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP42]]
; CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP92]], i64 240
; CHECK-NEXT: [[MIN_ITERS_CHECK_2:%.*]] = icmp samesign ult i32 [[I]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK_2]], label [[FOR_BODY4_US_PREHEADER_2:%.*]], label [[FOR_COND1_FOR_COND_CLEANUP3_CRIT_EDGE_US_1:%.*]]
@@ -224,8 +233,9 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: br label [[VECTOR_BODY_2:%.*]]
; CHECK: vector.body.2:
; CHECK-NEXT: [[INDEX_2:%.*]] = phi i64 [ 0, [[VECTOR_PH_2]] ], [ [[INDEX_NEXT_2:%.*]], [[VECTOR_BODY_2]] ]
-; CHECK-NEXT: [[TMP64:%.*]] = add nuw nsw i64 [[INDEX_2]], 30
-; CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[TMP64]]
+; CHECK-NEXT: [[TMP45:%.*]] = shl i64 [[INDEX_2]], 3
+; CHECK-NEXT: [[TMP46:%.*]] = add i64 [[TMP45]], 240
+; CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP46]]
; CHECK-NEXT: [[TMP79:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP78]], i64 16
; CHECK-NEXT: [[WIDE_LOAD_2:%.*]] = load <2 x double>, ptr [[TMP78]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[WIDE_LOAD21_2:%.*]] = load <2 x double>, ptr [[TMP79]], align 8, !alias.scope [[META0]]
@@ -234,7 +244,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23_2:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22_2]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP81:%.*]] = fmul <2 x double> [[WIDE_LOAD_2]], [[BROADCAST_SPLAT23_2]]
; CHECK-NEXT: [[TMP82:%.*]] = fmul <2 x double> [[WIDE_LOAD21_2]], [[BROADCAST_SPLAT23_2]]
-; CHECK-NEXT: [[TMP83:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[TMP64]]
+; CHECK-NEXT: [[TMP83:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP46]]
; CHECK-NEXT: [[TMP84:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP83]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24_2:%.*]] = load <2 x double>, ptr [[TMP83]], align 8, !alias.scope [[META5]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25_2:%.*]] = load <2 x double>, ptr [[TMP84]], align 8, !alias.scope [[META5]], !noalias [[META0]]
@@ -253,14 +263,15 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: br label [[FOR_BODY4_US_2:%.*]]
; CHECK: for.body4.us.2:
; CHECK-NEXT: [[INDVARS_IV_2:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_2:%.*]], [[FOR_BODY4_US_2]] ], [ [[INDVARS_IV_PH_2]], [[FOR_BODY4_US_PREHEADER_2]] ]
-; CHECK-NEXT: [[TMP88:%.*]] = add nuw nsw i64 [[INDVARS_IV_2]], 30
; CHECK-NEXT: [[TMP89:%.*]] = icmp samesign ult i64 [[INDVARS_IV_2]], 195
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP89]])
-; CHECK-NEXT: [[TMP90:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[TMP88]]
+; CHECK-NEXT: [[TMP64:%.*]] = shl nuw nsw i64 [[INDVARS_IV_2]], 3
+; CHECK-NEXT: [[TMP65:%.*]] = add nuw nsw i64 [[TMP64]], 240
+; CHECK-NEXT: [[TMP90:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP65]]
; CHECK-NEXT: [[MATRIXEXT_US_2:%.*]] = load double, ptr [[TMP90]], align 8
; CHECK-NEXT: [[MATRIXEXT8_US_2:%.*]] = load double, ptr [[TMP63]], align 8
; CHECK-NEXT: [[MUL_US_2:%.*]] = fmul double [[MATRIXEXT_US_2]], [[MATRIXEXT8_US_2]]
-; CHECK-NEXT: [[TMP91:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[TMP88]]
+; CHECK-NEXT: [[TMP91:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP65]]
; CHECK-NEXT: [[MATRIXEXT11_US_2:%.*]] = load double, ptr [[TMP91]], align 8
; CHECK-NEXT: [[SUB_US_2:%.*]] = fsub double [[MATRIXEXT11_US_2]], [[MUL_US_2]]
; CHECK-NEXT: store double [[SUB_US_2]], ptr [[TMP91]], align 8
@@ -270,7 +281,8 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK: for.cond1.for.cond.cleanup3_crit_edge.us.2:
; CHECK-NEXT: [[TMP93:%.*]] = icmp samesign ult i32 [[I]], 180
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP93]])
-; CHECK-NEXT: [[TMP123:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[CONV6]]
+; CHECK-NEXT: [[TMP68:%.*]] = shl nuw nsw i64 [[CONV6]], 3
+; CHECK-NEXT: [[TMP123:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP68]]
; CHECK-NEXT: [[TMP94:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP123]], i64 360
; CHECK-NEXT: [[MIN_ITERS_CHECK_3:%.*]] = icmp samesign ult i32 [[I]], 4
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK_3]], label [[FOR_BODY4_US_PREHEADER_3:%.*]], label [[FOR_COND1_FOR_COND_CLEANUP3_CRIT_EDGE_US_2:%.*]]
@@ -284,8 +296,9 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: br label [[VECTOR_BODY_3:%.*]]
; CHECK: vector.body.3:
; CHECK-NEXT: [[INDEX_3:%.*]] = phi i64 [ 0, [[VECTOR_PH_3]] ], [ [[INDEX_NEXT_3:%.*]], [[VECTOR_BODY_3]] ]
-; CHECK-NEXT: [[TMP95:%.*]] = add nuw nsw i64 [[INDEX_3]], 45
-; CHECK-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[TMP95]]
+; CHECK-NEXT: [[TMP66:%.*]] = shl i64 [[INDEX_3]], 3
+; CHECK-NEXT: [[TMP67:%.*]] = add i64 [[TMP66]], 360
+; CHECK-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP67]]
; CHECK-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP109]], i64 16
; CHECK-NEXT: [[WIDE_LOAD_3:%.*]] = load <2 x double>, ptr [[TMP109]], align 8, !alias.scope [[META0]]
; CHECK-NEXT: [[WIDE_LOAD21_3:%.*]] = load <2 x double>, ptr [[TMP110]], align 8, !alias.scope [[META0]]
@@ -294,7 +307,7 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: [[BROADCAST_SPLAT23_3:%.*]] = shufflevector <2 x double> [[BROADCAST_SPLATINSERT22_3]], <2 x double> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP112:%.*]] = fmul <2 x double> [[WIDE_LOAD_3]], [[BROADCAST_SPLAT23_3]]
; CHECK-NEXT: [[TMP113:%.*]] = fmul <2 x double> [[WIDE_LOAD21_3]], [[BROADCAST_SPLAT23_3]]
-; CHECK-NEXT: [[TMP114:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[TMP95]]
+; CHECK-NEXT: [[TMP114:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP67]]
; CHECK-NEXT: [[TMP115:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP114]], i64 16
; CHECK-NEXT: [[WIDE_LOAD24_3:%.*]] = load <2 x double>, ptr [[TMP114]], align 8, !alias.scope [[META5]], !noalias [[META0]]
; CHECK-NEXT: [[WIDE_LOAD25_3:%.*]] = load <2 x double>, ptr [[TMP115]], align 8, !alias.scope [[META5]], !noalias [[META0]]
@@ -313,14 +326,15 @@ define void @matrix_extract_insert_loop(i32 %i, ptr nonnull align 8 dereferencea
; CHECK-NEXT: br label [[FOR_BODY4_US_3:%.*]]
; CHECK: for.body4.us.3:
; CHECK-NEXT: [[INDVARS_IV_3:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_3:%.*]], [[FOR_BODY4_US_3]] ], [ [[INDVARS_IV_PH_3]], [[FOR_BODY4_US_PREHEADER_3]] ]
-; CHECK-NEXT: [[TMP119:%.*]] = add nuw nsw i64 [[INDVARS_IV_3]], 45
; CHECK-NEXT: [[TMP120:%.*]] = icmp samesign ult i64 [[INDVARS_IV_3]], 180
; CHECK-NEXT: tail call void @llvm.assume(i1 [[TMP120]])
-; CHECK-NEXT: [[TMP121:%.*]] = getelementptr inbounds nuw double, ptr [[A]], i64 [[TMP119]]
+; CHECK-NEXT: [[TMP88:%.*]] = shl nuw nsw i64 [[INDVARS_IV_3]], 3
+; CHECK-NEXT: [[TMP95:%.*]] = add nuw nsw i64 [[TMP88]], 360
+; CHECK-NEXT: [[TMP121:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP95]]
; CHECK-NEXT: [[MATRIXEXT_US_3:%.*]] = load double, ptr [[TMP121]], align 8
; CHECK-NEXT: [[MATRIXEXT8_US_3:%.*]] = load double, ptr [[TMP94]], align 8
; CHECK-NEXT: [[MUL_US_3:%.*]] = fmul double [[MATRIXEXT_US_3]], [[MATRIXEXT8_US_3]]
-; CHECK-NEXT: [[TMP122:%.*]] = getelementptr inbounds nuw double, ptr [[B]], i64 [[TMP119]]
+; CHECK-NEXT: [[TMP122:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP95]]
; CHECK-NEXT: [[MATRIXEXT11_US_3:%.*]] = load double, ptr [[TMP122]], align 8
; CHECK-NEXT: [[SUB_US_3:%.*]] = fsub double [[MATRIXEXT11_US_3]], [[MUL_US_3]]
; CHECK-NEXT: store double [[SUB_US_3]], ptr [[TMP122]], align 8
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 2fe49a31b7722..053210e8eeba8 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
@@ -46,11 +46,12 @@ define i64 @sum_2_at_with_int_conversion(ptr %A, ptr %B, i64 %N) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI13:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i64, ptr [[START_I]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i8, ptr [[START_I]], i64 [[TMP18]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP8]], align 8
; CHECK-NEXT: [[WIDE_LOAD14:%.*]] = load <2 x i64>, ptr [[TMP9]], align 8
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i64, ptr [[START_I1]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i8, ptr [[START_I1]], i64 [[TMP18]]
; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[TMP10]], i64 16
; CHECK-NEXT: [[WIDE_LOAD15:%.*]] = load <2 x i64>, ptr [[TMP10]], align 8
; CHECK-NEXT: [[WIDE_LOAD16:%.*]] = load <2 x i64>, ptr [[TMP11]], align 8
@@ -73,9 +74,10 @@ define i64 @sum_2_at_with_int_conversion(ptr %A, ptr %B, i64 %N) {
; CHECK: loop:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[LOOP]] ], [ [[IV_PH]], [[LOOP_PREHEADER17]] ]
; CHECK-NEXT: [[SUM:%.*]] = phi i64 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ [[SUM_PH]], [[LOOP_PREHEADER17]] ]
-; CHECK-NEXT: [[GEP_IDX_I:%.*]] = getelementptr i64, ptr [[START_I]], i64 [[IV]]
+; CHECK-NEXT: [[TMP19:%.*]] = shl i64 [[IV]], 3
+; CHECK-NEXT: [[GEP_IDX_I:%.*]] = getelementptr i8, ptr [[START_I]], i64 [[TMP19]]
; CHECK-NEXT: [[LV_I:%.*]] = load i64, ptr [[GEP_IDX_I]], align 8
-; CHECK-NEXT: [[GEP_IDX_I8:%.*]] = getelementptr i64, ptr [[START_I1]], i64 [[IV]]
+; CHECK-NEXT: [[GEP_IDX_I8:%.*]] = getelementptr i8, ptr [[START_I1]], i64 [[TMP19]]
; CHECK-NEXT: [[LV_I9:%.*]] = load i64, ptr [[GEP_IDX_I8]], align 8
; CHECK-NEXT: [[ADD:%.*]] = add i64 [[LV_I]], [[SUM]]
; CHECK-NEXT: [[SUM_NEXT]] = add i64 [[ADD]], [[LV_I9]]
@@ -161,15 +163,16 @@ define i64 @sum_3_at_with_int_conversion(ptr %A, ptr %B, ptr %C, i64 %N) {
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP21:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI25:%.*]] = phi <2 x i64> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP22:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i64, ptr [[START_I]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i8, ptr [[START_I]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i8, ptr [[TMP11]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP11]], align 8
; CHECK-NEXT: [[WIDE_LOAD26:%.*]] = load <2 x i64>, ptr [[TMP12]], align 8
-; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i64, ptr [[START_I1]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i8, ptr [[START_I1]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i8, ptr [[TMP13]], i64 16
; CHECK-NEXT: [[WIDE_LOAD27:%.*]] = load <2 x i64>, ptr [[TMP13]], align 8
; CHECK-NEXT: [[WIDE_LOAD28:%.*]] = load <2 x i64>, ptr [[TMP14]], align 8
-; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i64, ptr [[START_I12]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i8, ptr [[START_I12]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i8, ptr [[TMP15]], i64 16
; CHECK-NEXT: [[WIDE_LOAD29:%.*]] = load <2 x i64>, ptr [[TMP15]], align 8
; CHECK-NEXT: [[WIDE_LOAD30:%.*]] = load <2 x i64>, ptr [[TMP16]], align 8
@@ -194,11 +197,12 @@ define i64 @sum_3_at_with_int_conversion(ptr %A, ptr %B, ptr %C, i64 %N) {
; CHECK: loop:
; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[LOOP]] ], [ [[IV_PH]], [[LOOP_PREHEADER31]] ]
; CHECK-NEXT: [[SUM:%.*]] = phi i64 [ [[SUM_NEXT:%.*]], [[LOOP]] ], [ [[SUM_PH]], [[LOOP_PREHEADER31]] ]
-; CHECK-NEXT: [[GEP_IDX_I:%.*]] = getelementptr i64, ptr [[START_I]], i64 [[IV]]
+; CHECK-NEXT: [[TMP26:%.*]] = shl i64 [[IV]], 3
+; CHECK-NEXT: [[GEP_IDX_I:%.*]] = getelementptr i8, ptr [[START_I]], i64 [[TMP26]]
; CHECK-NEXT: [[LV_I:%.*]] = load i64, ptr [[GEP_IDX_I]], align 8
-; CHECK-NEXT: [[GEP_IDX_I8:%.*]] = getelementptr i64, ptr [[START_I1]], i64 [[IV]]
+; CHECK-NEXT: [[GEP_IDX_I8:%.*]] = getelementptr i8, ptr [[START_I1]], i64 [[TMP26]]
; CHECK-NEXT: [[LV_I9:%.*]] = load i64, ptr [[GEP_IDX_I8]], align 8
-; CHECK-NEXT: [[GEP_IDX_I19:%.*]] = getelementptr i64, ptr [[START_I12]], i64 [[IV]]
+; CHECK-NEXT: [[GEP_IDX_I19:%.*]] = getelementptr i8, ptr [[START_I12]], i64 [[TMP26]]
; CHECK-NEXT: [[LV_I20:%.*]] = load i64, ptr [[GEP_IDX_I19]], align 8
; CHECK-NEXT: [[ADD_1:%.*]] = add i64 [[LV_I]], [[SUM]]
; CHECK-NEXT: [[ADD_2:%.*]] = add i64 [[ADD_1]], [[LV_I9]]
@@ -251,7 +255,8 @@ define i64 @at_with_int_conversion(ptr %ptr, i64 %idx) {
; CHECK-NEXT: [[INRANGE:%.*]] = icmp ugt i64 [[IDX:%.*]], [[SUB]]
; CHECK-NEXT: br i1 [[INRANGE]], label [[ERROR:%.*]], label [[EXIT:%.*]]
; CHECK: exit:
-; CHECK-NEXT: [[GEP_IDX:%.*]] = getelementptr i64, ptr [[START]], i64 [[IDX]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl i64 [[IDX]], 3
+; CHECK-NEXT: [[GEP_IDX:%.*]] = getelementptr i8, ptr [[START]], i64 [[TMP1]]
; CHECK-NEXT: [[LV:%.*]] = load i64, ptr [[GEP_IDX]], align 8
; CHECK-NEXT: ret i64 [[LV]]
; CHECK: error:
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
index 48e8564322153..dc32f8d2ed46e 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/predicated-reduction.ll
@@ -27,7 +27,8 @@ define nofpclass(nan inf) double @monte_simple(i32 noundef %nblocks, i32 noundef
; CHECK-NEXT: [[VEC_PHI16:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP19:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI17:%.*]] = phi <4 x double> [ <double 0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %[[VECTOR_PH]] ], [ [[TMP14:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI18:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP15:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[SAMPLES]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP23:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[SAMPLES]], i64 [[TMP23]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[WIDE_LOAD19:%.*]] = load <4 x float>, ptr [[TMP1]], align 4
@@ -68,7 +69,8 @@ define nofpclass(nan inf) double @monte_simple(i32 noundef %nblocks, i32 noundef
; CHECK-NEXT: [[INDVARS_IV1:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ], [ [[INDVARS_IV_PH]], %[[FOR_BODY_PREHEADER22]] ]
; CHECK-NEXT: [[V1_012:%.*]] = phi double [ [[V1_2:%.*]], %[[FOR_BODY]] ], [ [[V1_011_PH]], %[[FOR_BODY_PREHEADER22]] ]
; CHECK-NEXT: [[V0_011:%.*]] = phi double [ [[V0_2:%.*]], %[[FOR_BODY]] ], [ [[V0_010_PH]], %[[FOR_BODY_PREHEADER22]] ]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw float, ptr [[SAMPLES]], i64 [[INDVARS_IV1]]
+; CHECK-NEXT: [[TMP25:%.*]] = shl nuw nsw i64 [[INDVARS_IV1]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[SAMPLES]], i64 [[TMP25]]
; CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[CONV:%.*]] = fpext nnan ninf float [[TMP0]] to double
; CHECK-NEXT: [[MUL:%.*]] = fmul fast double [[Y]], [[CONV]]
@@ -213,7 +215,8 @@ define nofpclass(nan inf) double @monte_exp(i32 noundef %nblocks, i32 noundef %R
; CHECK-NEXT: [[VEC_PHI31:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP23:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI32:%.*]] = phi <4 x double> [ [[TMP27]], %[[VECTOR_PH]] ], [ [[TMP18:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[VEC_PHI33:%.*]] = phi <4 x double> [ splat (double -0.000000e+00), %[[VECTOR_PH]] ], [ [[TMP19:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX_US:%.*]] = getelementptr inbounds nuw float, ptr [[SAMPLES]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP30:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX_US:%.*]] = getelementptr inbounds nuw i8, ptr [[SAMPLES]], i64 [[TMP30]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX_US]], i64 16
; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[ARRAYIDX_US]], align 4
; CHECK-NEXT: [[WIDE_LOAD34:%.*]] = load <4 x float>, ptr [[TMP3]], align 4
@@ -255,7 +258,8 @@ define nofpclass(nan inf) double @monte_exp(i32 noundef %nblocks, i32 noundef %R
; CHECK-NEXT: [[INDVARS_IV1:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY3_US]] ], [ [[INDVARS_IV_PH]], %[[FOR_BODY3_US_PREHEADER]] ]
; CHECK-NEXT: [[V1_116_US:%.*]] = phi double [ [[V1_2_US:%.*]], %[[FOR_BODY3_US]] ], [ [[V1_114_US_PH]], %[[FOR_BODY3_US_PREHEADER]] ]
; CHECK-NEXT: [[V0_115_US:%.*]] = phi double [ [[V0_2_US:%.*]], %[[FOR_BODY3_US]] ], [ [[V0_113_US_PH]], %[[FOR_BODY3_US_PREHEADER]] ]
-; CHECK-NEXT: [[ARRAYIDX_US1:%.*]] = getelementptr inbounds nuw float, ptr [[SAMPLES]], i64 [[INDVARS_IV1]]
+; CHECK-NEXT: [[TMP28:%.*]] = shl nuw nsw i64 [[INDVARS_IV1]], 2
+; CHECK-NEXT: [[ARRAYIDX_US1:%.*]] = getelementptr inbounds nuw i8, ptr [[SAMPLES]], i64 [[TMP28]]
; CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX_US1]], align 4
; CHECK-NEXT: [[CONV_US:%.*]] = fpext nnan float [[TMP0]] to double
; CHECK-NEXT: [[TMP1:%.*]] = tail call fast double @llvm.exp2.f64(double [[CONV_US]])
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll
index 258ef63503bec..131f06f5b864c 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/quant_4x4.ll
@@ -63,458 +63,49 @@ define i32 @quant_4x4(ptr noundef %dct, ptr noundef %mf, ptr noundef %bias) {
; CHECK-NEXT: store <8 x i16> [[PREDPHI34]], ptr [[TMP0]], align 2, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT: [[BIN_RDX35:%.*]] = or <8 x i16> [[PREDPHI34]], [[PREDPHI]]
; CHECK-NEXT: [[TMP29:%.*]] = tail call i16 @llvm.vector.reduce.or.v8i16(<8 x i16> [[BIN_RDX35]])
+; CHECK-NEXT: [[TMP30:%.*]] = sext i16 [[TMP29]] to i32
; CHECK-NEXT: br label [[FOR_COND_CLEANUP:%.*]]
; CHECK: for.cond.cleanup:
-; CHECK-NEXT: [[OR_LCSSA_IN:%.*]] = phi i16 [ [[TMP29]], [[VECTOR_BODY]] ], [ [[OR_1551:%.*]], [[IF_END_15:%.*]] ]
-; CHECK-NEXT: [[TOBOOL:%.*]] = icmp ne i16 [[OR_LCSSA_IN]], 0
+; CHECK-NEXT: [[OR_LCSSA:%.*]] = phi i32 [ [[TMP30]], [[VECTOR_BODY]] ], [ [[OR:%.*]], [[IF_END:%.*]] ]
+; CHECK-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[OR_LCSSA]], 0
; CHECK-NEXT: [[LNOT_EXT:%.*]] = zext i1 [[TOBOOL]] to i32
; CHECK-NEXT: ret i32 [[LNOT_EXT]]
; CHECK: for.body:
-; CHECK-NEXT: [[TMP30:%.*]] = load i16, ptr [[DCT]], align 2
-; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP30]] to i32
-; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i16 [[TMP30]], 0
-; CHECK-NEXT: [[TMP31:%.*]] = load i16, ptr [[BIAS]], align 2
-; CHECK-NEXT: [[CONV5:%.*]] = zext i16 [[TMP31]] to i32
-; CHECK-NEXT: [[TMP32:%.*]] = load i16, ptr [[MF]], align 2
-; CHECK-NEXT: [[CONV11:%.*]] = zext i16 [[TMP32]] to i32
+; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[IF_END]] ], [ 0, [[ENTRY:%.*]] ]
+; CHECK-NEXT: [[NZ_020:%.*]] = phi i32 [ [[OR]], [[IF_END]] ], [ 0, [[ENTRY]] ]
+; CHECK-NEXT: [[TMP31:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP32:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
+; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP32]] to i32
+; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i16 [[TMP32]], 0
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP33:%.*]] = load i16, ptr [[ARRAYIDX4]], align 2
+; CHECK-NEXT: [[CONV5:%.*]] = zext i16 [[TMP33]] to i32
+; CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 [[TMP31]]
+; CHECK-NEXT: [[TMP34:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
+; CHECK-NEXT: [[CONV11:%.*]] = zext i16 [[TMP34]] to i32
; CHECK-NEXT: br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then:
; CHECK-NEXT: [[ADD:%.*]] = add nuw nsw i32 [[CONV5]], [[CONV]]
; CHECK-NEXT: [[MUL:%.*]] = mul i32 [[ADD]], [[CONV11]]
; CHECK-NEXT: [[SHR:%.*]] = lshr i32 [[MUL]], 16
; CHECK-NEXT: [[CONV12:%.*]] = trunc nuw i32 [[SHR]] to i16
-; CHECK-NEXT: br label [[IF_END:%.*]]
+; CHECK-NEXT: br label [[IF_END]]
; CHECK: if.else:
; CHECK-NEXT: [[ADD21:%.*]] = sub nsw i32 [[CONV5]], [[CONV]]
; CHECK-NEXT: [[MUL25:%.*]] = mul i32 [[ADD21]], [[CONV11]]
; CHECK-NEXT: [[SHR26:%.*]] = lshr i32 [[MUL25]], 16
-; CHECK-NEXT: [[TMP33:%.*]] = trunc nuw i32 [[SHR26]] to i16
-; CHECK-NEXT: [[CONV28:%.*]] = sub i16 0, [[TMP33]]
+; CHECK-NEXT: [[TMP35:%.*]] = trunc nuw i32 [[SHR26]] to i16
+; CHECK-NEXT: [[CONV28:%.*]] = sub i16 0, [[TMP35]]
; CHECK-NEXT: br label [[IF_END]]
; 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 nuw i8, ptr [[DCT]], i64 2
-; CHECK-NEXT: [[TMP34:%.*]] = load i16, ptr [[ARRAYIDX_1]], align 2
-; CHECK-NEXT: [[CONV_1:%.*]] = sext i16 [[TMP34]] to i32
-; CHECK-NEXT: [[CMP1_1:%.*]] = icmp sgt i16 [[TMP34]], 0
-; CHECK-NEXT: [[ARRAYIDX4_1:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 2
-; CHECK-NEXT: [[TMP35:%.*]] = load i16, ptr [[ARRAYIDX4_1]], align 2
-; CHECK-NEXT: [[CONV5_1:%.*]] = zext i16 [[TMP35]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_1:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 2
-; CHECK-NEXT: [[TMP36:%.*]] = load i16, ptr [[ARRAYIDX10_1]], align 2
-; CHECK-NEXT: [[CONV11_1:%.*]] = zext i16 [[TMP36]] to i32
-; CHECK-NEXT: br i1 [[CMP1_1]], label [[IF_THEN_1:%.*]], label [[IF_ELSE_1:%.*]]
-; CHECK: if.else.1:
-; CHECK-NEXT: [[ADD21_1:%.*]] = sub nsw i32 [[CONV5_1]], [[CONV_1]]
-; CHECK-NEXT: [[MUL25_1:%.*]] = mul i32 [[ADD21_1]], [[CONV11_1]]
-; CHECK-NEXT: [[SHR26_1:%.*]] = lshr i32 [[MUL25_1]], 16
-; CHECK-NEXT: [[TMP37:%.*]] = trunc nuw i32 [[SHR26_1]] to i16
-; CHECK-NEXT: [[CONV28_1:%.*]] = sub i16 0, [[TMP37]]
-; CHECK-NEXT: br label [[IF_END_1:%.*]]
-; CHECK: if.then.1:
-; CHECK-NEXT: [[ADD_1:%.*]] = add nuw nsw i32 [[CONV5_1]], [[CONV_1]]
-; CHECK-NEXT: [[MUL_1:%.*]] = mul i32 [[ADD_1]], [[CONV11_1]]
-; CHECK-NEXT: [[SHR_1:%.*]] = lshr i32 [[MUL_1]], 16
-; CHECK-NEXT: [[CONV12_1:%.*]] = trunc nuw i32 [[SHR_1]] to i16
-; CHECK-NEXT: br label [[IF_END_1]]
-; CHECK: if.end.1:
-; 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_137:%.*]] = or i16 [[STOREMERGE]], [[STOREMERGE_1]]
-; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 4
-; CHECK-NEXT: [[TMP38:%.*]] = load i16, ptr [[ARRAYIDX_2]], align 2
-; CHECK-NEXT: [[CONV_2:%.*]] = sext i16 [[TMP38]] to i32
-; CHECK-NEXT: [[CMP1_2:%.*]] = icmp sgt i16 [[TMP38]], 0
-; CHECK-NEXT: [[ARRAYIDX4_2:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 4
-; CHECK-NEXT: [[TMP39:%.*]] = load i16, ptr [[ARRAYIDX4_2]], align 2
-; CHECK-NEXT: [[CONV5_2:%.*]] = zext i16 [[TMP39]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_2:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 4
-; CHECK-NEXT: [[TMP40:%.*]] = load i16, ptr [[ARRAYIDX10_2]], align 2
-; CHECK-NEXT: [[CONV11_2:%.*]] = zext i16 [[TMP40]] to i32
-; CHECK-NEXT: br i1 [[CMP1_2]], label [[IF_THEN_2:%.*]], label [[IF_ELSE_2:%.*]]
-; CHECK: if.else.2:
-; CHECK-NEXT: [[ADD21_2:%.*]] = sub nsw i32 [[CONV5_2]], [[CONV_2]]
-; CHECK-NEXT: [[MUL25_2:%.*]] = mul i32 [[ADD21_2]], [[CONV11_2]]
-; CHECK-NEXT: [[SHR26_2:%.*]] = lshr i32 [[MUL25_2]], 16
-; CHECK-NEXT: [[TMP41:%.*]] = trunc nuw i32 [[SHR26_2]] to i16
-; CHECK-NEXT: [[CONV28_2:%.*]] = sub i16 0, [[TMP41]]
-; CHECK-NEXT: br label [[IF_END_2:%.*]]
-; CHECK: if.then.2:
-; CHECK-NEXT: [[ADD_2:%.*]] = add nuw nsw i32 [[CONV5_2]], [[CONV_2]]
-; CHECK-NEXT: [[MUL_2:%.*]] = mul i32 [[ADD_2]], [[CONV11_2]]
-; CHECK-NEXT: [[SHR_2:%.*]] = lshr i32 [[MUL_2]], 16
-; CHECK-NEXT: [[CONV12_2:%.*]] = trunc nuw i32 [[SHR_2]] to i16
-; CHECK-NEXT: br label [[IF_END_2]]
-; CHECK: if.end.2:
-; 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_238:%.*]] = or i16 [[OR_137]], [[STOREMERGE_2]]
-; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 6
-; CHECK-NEXT: [[TMP42:%.*]] = load i16, ptr [[ARRAYIDX_3]], align 2
-; CHECK-NEXT: [[CONV_3:%.*]] = sext i16 [[TMP42]] to i32
-; CHECK-NEXT: [[CMP1_3:%.*]] = icmp sgt i16 [[TMP42]], 0
-; CHECK-NEXT: [[ARRAYIDX4_3:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 6
-; CHECK-NEXT: [[TMP43:%.*]] = load i16, ptr [[ARRAYIDX4_3]], align 2
-; CHECK-NEXT: [[CONV5_3:%.*]] = zext i16 [[TMP43]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_3:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 6
-; CHECK-NEXT: [[TMP44:%.*]] = load i16, ptr [[ARRAYIDX10_3]], align 2
-; CHECK-NEXT: [[CONV11_3:%.*]] = zext i16 [[TMP44]] to i32
-; CHECK-NEXT: br i1 [[CMP1_3]], label [[IF_THEN_3:%.*]], label [[IF_ELSE_3:%.*]]
-; CHECK: if.else.3:
-; CHECK-NEXT: [[ADD21_3:%.*]] = sub nsw i32 [[CONV5_3]], [[CONV_3]]
-; CHECK-NEXT: [[MUL25_3:%.*]] = mul i32 [[ADD21_3]], [[CONV11_3]]
-; CHECK-NEXT: [[SHR26_3:%.*]] = lshr i32 [[MUL25_3]], 16
-; CHECK-NEXT: [[TMP45:%.*]] = trunc nuw i32 [[SHR26_3]] to i16
-; CHECK-NEXT: [[CONV28_3:%.*]] = sub i16 0, [[TMP45]]
-; CHECK-NEXT: br label [[IF_END_3:%.*]]
-; CHECK: if.then.3:
-; CHECK-NEXT: [[ADD_3:%.*]] = add nuw nsw i32 [[CONV5_3]], [[CONV_3]]
-; CHECK-NEXT: [[MUL_3:%.*]] = mul i32 [[ADD_3]], [[CONV11_3]]
-; CHECK-NEXT: [[SHR_3:%.*]] = lshr i32 [[MUL_3]], 16
-; CHECK-NEXT: [[CONV12_3:%.*]] = trunc nuw i32 [[SHR_3]] to i16
-; CHECK-NEXT: br label [[IF_END_3]]
-; CHECK: if.end.3:
-; 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_339:%.*]] = or i16 [[OR_238]], [[STOREMERGE_3]]
-; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 8
-; CHECK-NEXT: [[TMP46:%.*]] = load i16, ptr [[ARRAYIDX_4]], align 2
-; CHECK-NEXT: [[CONV_4:%.*]] = sext i16 [[TMP46]] to i32
-; CHECK-NEXT: [[CMP1_4:%.*]] = icmp sgt i16 [[TMP46]], 0
-; CHECK-NEXT: [[ARRAYIDX4_4:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 8
-; CHECK-NEXT: [[TMP47:%.*]] = load i16, ptr [[ARRAYIDX4_4]], align 2
-; CHECK-NEXT: [[CONV5_4:%.*]] = zext i16 [[TMP47]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_4:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 8
-; CHECK-NEXT: [[TMP48:%.*]] = load i16, ptr [[ARRAYIDX10_4]], align 2
-; CHECK-NEXT: [[CONV11_4:%.*]] = zext i16 [[TMP48]] to i32
-; CHECK-NEXT: br i1 [[CMP1_4]], label [[IF_THEN_4:%.*]], label [[IF_ELSE_4:%.*]]
-; CHECK: if.else.4:
-; CHECK-NEXT: [[ADD21_4:%.*]] = sub nsw i32 [[CONV5_4]], [[CONV_4]]
-; CHECK-NEXT: [[MUL25_4:%.*]] = mul i32 [[ADD21_4]], [[CONV11_4]]
-; CHECK-NEXT: [[SHR26_4:%.*]] = lshr i32 [[MUL25_4]], 16
-; CHECK-NEXT: [[TMP49:%.*]] = trunc nuw i32 [[SHR26_4]] to i16
-; CHECK-NEXT: [[CONV28_4:%.*]] = sub i16 0, [[TMP49]]
-; CHECK-NEXT: br label [[IF_END_4:%.*]]
-; CHECK: if.then.4:
-; CHECK-NEXT: [[ADD_4:%.*]] = add nuw nsw i32 [[CONV5_4]], [[CONV_4]]
-; CHECK-NEXT: [[MUL_4:%.*]] = mul i32 [[ADD_4]], [[CONV11_4]]
-; CHECK-NEXT: [[SHR_4:%.*]] = lshr i32 [[MUL_4]], 16
-; CHECK-NEXT: [[CONV12_4:%.*]] = trunc nuw i32 [[SHR_4]] to i16
-; CHECK-NEXT: br label [[IF_END_4]]
-; CHECK: if.end.4:
-; 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_440:%.*]] = or i16 [[OR_339]], [[STOREMERGE_4]]
-; CHECK-NEXT: [[ARRAYIDX_5:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 10
-; CHECK-NEXT: [[TMP50:%.*]] = load i16, ptr [[ARRAYIDX_5]], align 2
-; CHECK-NEXT: [[CONV_5:%.*]] = sext i16 [[TMP50]] to i32
-; CHECK-NEXT: [[CMP1_5:%.*]] = icmp sgt i16 [[TMP50]], 0
-; CHECK-NEXT: [[ARRAYIDX4_5:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 10
-; CHECK-NEXT: [[TMP51:%.*]] = load i16, ptr [[ARRAYIDX4_5]], align 2
-; CHECK-NEXT: [[CONV5_5:%.*]] = zext i16 [[TMP51]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_5:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 10
-; CHECK-NEXT: [[TMP52:%.*]] = load i16, ptr [[ARRAYIDX10_5]], align 2
-; CHECK-NEXT: [[CONV11_5:%.*]] = zext i16 [[TMP52]] to i32
-; CHECK-NEXT: br i1 [[CMP1_5]], label [[IF_THEN_5:%.*]], label [[IF_ELSE_5:%.*]]
-; CHECK: if.else.5:
-; CHECK-NEXT: [[ADD21_5:%.*]] = sub nsw i32 [[CONV5_5]], [[CONV_5]]
-; CHECK-NEXT: [[MUL25_5:%.*]] = mul i32 [[ADD21_5]], [[CONV11_5]]
-; CHECK-NEXT: [[SHR26_5:%.*]] = lshr i32 [[MUL25_5]], 16
-; CHECK-NEXT: [[TMP53:%.*]] = trunc nuw i32 [[SHR26_5]] to i16
-; CHECK-NEXT: [[CONV28_5:%.*]] = sub i16 0, [[TMP53]]
-; CHECK-NEXT: br label [[IF_END_5:%.*]]
-; CHECK: if.then.5:
-; CHECK-NEXT: [[ADD_5:%.*]] = add nuw nsw i32 [[CONV5_5]], [[CONV_5]]
-; CHECK-NEXT: [[MUL_5:%.*]] = mul i32 [[ADD_5]], [[CONV11_5]]
-; CHECK-NEXT: [[SHR_5:%.*]] = lshr i32 [[MUL_5]], 16
-; CHECK-NEXT: [[CONV12_5:%.*]] = trunc nuw i32 [[SHR_5]] to i16
-; CHECK-NEXT: br label [[IF_END_5]]
-; CHECK: if.end.5:
-; 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_541:%.*]] = or i16 [[OR_440]], [[STOREMERGE_5]]
-; CHECK-NEXT: [[ARRAYIDX_6:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 12
-; CHECK-NEXT: [[TMP54:%.*]] = load i16, ptr [[ARRAYIDX_6]], align 2
-; CHECK-NEXT: [[CONV_6:%.*]] = sext i16 [[TMP54]] to i32
-; CHECK-NEXT: [[CMP1_6:%.*]] = icmp sgt i16 [[TMP54]], 0
-; CHECK-NEXT: [[ARRAYIDX4_6:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 12
-; CHECK-NEXT: [[TMP55:%.*]] = load i16, ptr [[ARRAYIDX4_6]], align 2
-; CHECK-NEXT: [[CONV5_6:%.*]] = zext i16 [[TMP55]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_6:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 12
-; CHECK-NEXT: [[TMP56:%.*]] = load i16, ptr [[ARRAYIDX10_6]], align 2
-; CHECK-NEXT: [[CONV11_6:%.*]] = zext i16 [[TMP56]] to i32
-; CHECK-NEXT: br i1 [[CMP1_6]], label [[IF_THEN_6:%.*]], label [[IF_ELSE_6:%.*]]
-; CHECK: if.else.6:
-; CHECK-NEXT: [[ADD21_6:%.*]] = sub nsw i32 [[CONV5_6]], [[CONV_6]]
-; CHECK-NEXT: [[MUL25_6:%.*]] = mul i32 [[ADD21_6]], [[CONV11_6]]
-; CHECK-NEXT: [[SHR26_6:%.*]] = lshr i32 [[MUL25_6]], 16
-; CHECK-NEXT: [[TMP57:%.*]] = trunc nuw i32 [[SHR26_6]] to i16
-; CHECK-NEXT: [[CONV28_6:%.*]] = sub i16 0, [[TMP57]]
-; CHECK-NEXT: br label [[IF_END_6:%.*]]
-; CHECK: if.then.6:
-; CHECK-NEXT: [[ADD_6:%.*]] = add nuw nsw i32 [[CONV5_6]], [[CONV_6]]
-; CHECK-NEXT: [[MUL_6:%.*]] = mul i32 [[ADD_6]], [[CONV11_6]]
-; CHECK-NEXT: [[SHR_6:%.*]] = lshr i32 [[MUL_6]], 16
-; CHECK-NEXT: [[CONV12_6:%.*]] = trunc nuw i32 [[SHR_6]] to i16
-; CHECK-NEXT: br label [[IF_END_6]]
-; CHECK: if.end.6:
-; 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_642:%.*]] = or i16 [[OR_541]], [[STOREMERGE_6]]
-; CHECK-NEXT: [[ARRAYIDX_7:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 14
-; CHECK-NEXT: [[TMP58:%.*]] = load i16, ptr [[ARRAYIDX_7]], align 2
-; CHECK-NEXT: [[CONV_7:%.*]] = sext i16 [[TMP58]] to i32
-; CHECK-NEXT: [[CMP1_7:%.*]] = icmp sgt i16 [[TMP58]], 0
-; CHECK-NEXT: [[ARRAYIDX4_7:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 14
-; CHECK-NEXT: [[TMP59:%.*]] = load i16, ptr [[ARRAYIDX4_7]], align 2
-; CHECK-NEXT: [[CONV5_7:%.*]] = zext i16 [[TMP59]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_7:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 14
-; CHECK-NEXT: [[TMP60:%.*]] = load i16, ptr [[ARRAYIDX10_7]], align 2
-; CHECK-NEXT: [[CONV11_7:%.*]] = zext i16 [[TMP60]] to i32
-; CHECK-NEXT: br i1 [[CMP1_7]], label [[IF_THEN_7:%.*]], label [[IF_ELSE_7:%.*]]
-; CHECK: if.else.7:
-; CHECK-NEXT: [[ADD21_7:%.*]] = sub nsw i32 [[CONV5_7]], [[CONV_7]]
-; CHECK-NEXT: [[MUL25_7:%.*]] = mul i32 [[ADD21_7]], [[CONV11_7]]
-; CHECK-NEXT: [[SHR26_7:%.*]] = lshr i32 [[MUL25_7]], 16
-; CHECK-NEXT: [[TMP61:%.*]] = trunc nuw i32 [[SHR26_7]] to i16
-; CHECK-NEXT: [[CONV28_7:%.*]] = sub i16 0, [[TMP61]]
-; CHECK-NEXT: br label [[IF_END_7:%.*]]
-; CHECK: if.then.7:
-; CHECK-NEXT: [[ADD_7:%.*]] = add nuw nsw i32 [[CONV5_7]], [[CONV_7]]
-; CHECK-NEXT: [[MUL_7:%.*]] = mul i32 [[ADD_7]], [[CONV11_7]]
-; CHECK-NEXT: [[SHR_7:%.*]] = lshr i32 [[MUL_7]], 16
-; CHECK-NEXT: [[CONV12_7:%.*]] = trunc nuw i32 [[SHR_7]] to i16
-; CHECK-NEXT: br label [[IF_END_7]]
-; CHECK: if.end.7:
-; 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_743:%.*]] = or i16 [[OR_642]], [[STOREMERGE_7]]
-; CHECK-NEXT: [[ARRAYIDX_8:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 16
-; CHECK-NEXT: [[TMP62:%.*]] = load i16, ptr [[ARRAYIDX_8]], align 2
-; CHECK-NEXT: [[CONV_8:%.*]] = sext i16 [[TMP62]] to i32
-; CHECK-NEXT: [[CMP1_8:%.*]] = icmp sgt i16 [[TMP62]], 0
-; CHECK-NEXT: [[ARRAYIDX4_8:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 16
-; CHECK-NEXT: [[TMP63:%.*]] = load i16, ptr [[ARRAYIDX4_8]], align 2
-; CHECK-NEXT: [[CONV5_8:%.*]] = zext i16 [[TMP63]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_8:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 16
-; CHECK-NEXT: [[TMP64:%.*]] = load i16, ptr [[ARRAYIDX10_8]], align 2
-; CHECK-NEXT: [[CONV11_8:%.*]] = zext i16 [[TMP64]] to i32
-; CHECK-NEXT: br i1 [[CMP1_8]], label [[IF_THEN_8:%.*]], label [[IF_ELSE_8:%.*]]
-; CHECK: if.else.8:
-; CHECK-NEXT: [[ADD21_8:%.*]] = sub nsw i32 [[CONV5_8]], [[CONV_8]]
-; CHECK-NEXT: [[MUL25_8:%.*]] = mul i32 [[ADD21_8]], [[CONV11_8]]
-; CHECK-NEXT: [[SHR26_8:%.*]] = lshr i32 [[MUL25_8]], 16
-; CHECK-NEXT: [[TMP65:%.*]] = trunc nuw i32 [[SHR26_8]] to i16
-; CHECK-NEXT: [[CONV28_8:%.*]] = sub i16 0, [[TMP65]]
-; CHECK-NEXT: br label [[IF_END_8:%.*]]
-; CHECK: if.then.8:
-; CHECK-NEXT: [[ADD_8:%.*]] = add nuw nsw i32 [[CONV5_8]], [[CONV_8]]
-; CHECK-NEXT: [[MUL_8:%.*]] = mul i32 [[ADD_8]], [[CONV11_8]]
-; CHECK-NEXT: [[SHR_8:%.*]] = lshr i32 [[MUL_8]], 16
-; CHECK-NEXT: [[CONV12_8:%.*]] = trunc nuw i32 [[SHR_8]] to i16
-; CHECK-NEXT: br label [[IF_END_8]]
-; CHECK: if.end.8:
-; 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_844:%.*]] = or i16 [[OR_743]], [[STOREMERGE_8]]
-; CHECK-NEXT: [[ARRAYIDX_9:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 18
-; CHECK-NEXT: [[TMP66:%.*]] = load i16, ptr [[ARRAYIDX_9]], align 2
-; CHECK-NEXT: [[CONV_9:%.*]] = sext i16 [[TMP66]] to i32
-; CHECK-NEXT: [[CMP1_9:%.*]] = icmp sgt i16 [[TMP66]], 0
-; CHECK-NEXT: [[ARRAYIDX4_9:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 18
-; CHECK-NEXT: [[TMP67:%.*]] = load i16, ptr [[ARRAYIDX4_9]], align 2
-; CHECK-NEXT: [[CONV5_9:%.*]] = zext i16 [[TMP67]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_9:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 18
-; CHECK-NEXT: [[TMP68:%.*]] = load i16, ptr [[ARRAYIDX10_9]], align 2
-; CHECK-NEXT: [[CONV11_9:%.*]] = zext i16 [[TMP68]] to i32
-; CHECK-NEXT: br i1 [[CMP1_9]], label [[IF_THEN_9:%.*]], label [[IF_ELSE_9:%.*]]
-; CHECK: if.else.9:
-; CHECK-NEXT: [[ADD21_9:%.*]] = sub nsw i32 [[CONV5_9]], [[CONV_9]]
-; CHECK-NEXT: [[MUL25_9:%.*]] = mul i32 [[ADD21_9]], [[CONV11_9]]
-; CHECK-NEXT: [[SHR26_9:%.*]] = lshr i32 [[MUL25_9]], 16
-; CHECK-NEXT: [[TMP69:%.*]] = trunc nuw i32 [[SHR26_9]] to i16
-; CHECK-NEXT: [[CONV28_9:%.*]] = sub i16 0, [[TMP69]]
-; CHECK-NEXT: br label [[IF_END_9:%.*]]
-; CHECK: if.then.9:
-; CHECK-NEXT: [[ADD_9:%.*]] = add nuw nsw i32 [[CONV5_9]], [[CONV_9]]
-; CHECK-NEXT: [[MUL_9:%.*]] = mul i32 [[ADD_9]], [[CONV11_9]]
-; CHECK-NEXT: [[SHR_9:%.*]] = lshr i32 [[MUL_9]], 16
-; CHECK-NEXT: [[CONV12_9:%.*]] = trunc nuw i32 [[SHR_9]] to i16
-; CHECK-NEXT: br label [[IF_END_9]]
-; CHECK: if.end.9:
-; 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_945:%.*]] = or i16 [[OR_844]], [[STOREMERGE_9]]
-; CHECK-NEXT: [[ARRAYIDX_10:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 20
-; CHECK-NEXT: [[TMP70:%.*]] = load i16, ptr [[ARRAYIDX_10]], align 2
-; CHECK-NEXT: [[CONV_10:%.*]] = sext i16 [[TMP70]] to i32
-; CHECK-NEXT: [[CMP1_10:%.*]] = icmp sgt i16 [[TMP70]], 0
-; CHECK-NEXT: [[ARRAYIDX4_10:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 20
-; CHECK-NEXT: [[TMP71:%.*]] = load i16, ptr [[ARRAYIDX4_10]], align 2
-; CHECK-NEXT: [[CONV5_10:%.*]] = zext i16 [[TMP71]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_10:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 20
-; CHECK-NEXT: [[TMP72:%.*]] = load i16, ptr [[ARRAYIDX10_10]], align 2
-; CHECK-NEXT: [[CONV11_10:%.*]] = zext i16 [[TMP72]] to i32
-; CHECK-NEXT: br i1 [[CMP1_10]], label [[IF_THEN_10:%.*]], label [[IF_ELSE_10:%.*]]
-; CHECK: if.else.10:
-; CHECK-NEXT: [[ADD21_10:%.*]] = sub nsw i32 [[CONV5_10]], [[CONV_10]]
-; CHECK-NEXT: [[MUL25_10:%.*]] = mul i32 [[ADD21_10]], [[CONV11_10]]
-; CHECK-NEXT: [[SHR26_10:%.*]] = lshr i32 [[MUL25_10]], 16
-; CHECK-NEXT: [[TMP73:%.*]] = trunc nuw i32 [[SHR26_10]] to i16
-; CHECK-NEXT: [[CONV28_10:%.*]] = sub i16 0, [[TMP73]]
-; CHECK-NEXT: br label [[IF_END_10:%.*]]
-; CHECK: if.then.10:
-; CHECK-NEXT: [[ADD_10:%.*]] = add nuw nsw i32 [[CONV5_10]], [[CONV_10]]
-; CHECK-NEXT: [[MUL_10:%.*]] = mul i32 [[ADD_10]], [[CONV11_10]]
-; CHECK-NEXT: [[SHR_10:%.*]] = lshr i32 [[MUL_10]], 16
-; CHECK-NEXT: [[CONV12_10:%.*]] = trunc nuw i32 [[SHR_10]] to i16
-; CHECK-NEXT: br label [[IF_END_10]]
-; CHECK: if.end.10:
-; 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_1046:%.*]] = or i16 [[OR_945]], [[STOREMERGE_10]]
-; CHECK-NEXT: [[ARRAYIDX_11:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 22
-; CHECK-NEXT: [[TMP74:%.*]] = load i16, ptr [[ARRAYIDX_11]], align 2
-; CHECK-NEXT: [[CONV_11:%.*]] = sext i16 [[TMP74]] to i32
-; CHECK-NEXT: [[CMP1_11:%.*]] = icmp sgt i16 [[TMP74]], 0
-; CHECK-NEXT: [[ARRAYIDX4_11:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 22
-; CHECK-NEXT: [[TMP75:%.*]] = load i16, ptr [[ARRAYIDX4_11]], align 2
-; CHECK-NEXT: [[CONV5_11:%.*]] = zext i16 [[TMP75]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_11:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 22
-; CHECK-NEXT: [[TMP76:%.*]] = load i16, ptr [[ARRAYIDX10_11]], align 2
-; CHECK-NEXT: [[CONV11_11:%.*]] = zext i16 [[TMP76]] to i32
-; CHECK-NEXT: br i1 [[CMP1_11]], label [[IF_THEN_11:%.*]], label [[IF_ELSE_11:%.*]]
-; CHECK: if.else.11:
-; CHECK-NEXT: [[ADD21_11:%.*]] = sub nsw i32 [[CONV5_11]], [[CONV_11]]
-; CHECK-NEXT: [[MUL25_11:%.*]] = mul i32 [[ADD21_11]], [[CONV11_11]]
-; CHECK-NEXT: [[SHR26_11:%.*]] = lshr i32 [[MUL25_11]], 16
-; CHECK-NEXT: [[TMP77:%.*]] = trunc nuw i32 [[SHR26_11]] to i16
-; CHECK-NEXT: [[CONV28_11:%.*]] = sub i16 0, [[TMP77]]
-; CHECK-NEXT: br label [[IF_END_11:%.*]]
-; CHECK: if.then.11:
-; CHECK-NEXT: [[ADD_11:%.*]] = add nuw nsw i32 [[CONV5_11]], [[CONV_11]]
-; CHECK-NEXT: [[MUL_11:%.*]] = mul i32 [[ADD_11]], [[CONV11_11]]
-; CHECK-NEXT: [[SHR_11:%.*]] = lshr i32 [[MUL_11]], 16
-; CHECK-NEXT: [[CONV12_11:%.*]] = trunc nuw i32 [[SHR_11]] to i16
-; CHECK-NEXT: br label [[IF_END_11]]
-; CHECK: if.end.11:
-; 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_1147:%.*]] = or i16 [[OR_1046]], [[STOREMERGE_11]]
-; CHECK-NEXT: [[ARRAYIDX_12:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 24
-; CHECK-NEXT: [[TMP78:%.*]] = load i16, ptr [[ARRAYIDX_12]], align 2
-; CHECK-NEXT: [[CONV_12:%.*]] = sext i16 [[TMP78]] to i32
-; CHECK-NEXT: [[CMP1_12:%.*]] = icmp sgt i16 [[TMP78]], 0
-; CHECK-NEXT: [[ARRAYIDX4_12:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 24
-; CHECK-NEXT: [[TMP79:%.*]] = load i16, ptr [[ARRAYIDX4_12]], align 2
-; CHECK-NEXT: [[CONV5_12:%.*]] = zext i16 [[TMP79]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_12:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 24
-; CHECK-NEXT: [[TMP80:%.*]] = load i16, ptr [[ARRAYIDX10_12]], align 2
-; CHECK-NEXT: [[CONV11_12:%.*]] = zext i16 [[TMP80]] to i32
-; CHECK-NEXT: br i1 [[CMP1_12]], label [[IF_THEN_12:%.*]], label [[IF_ELSE_12:%.*]]
-; CHECK: if.else.12:
-; CHECK-NEXT: [[ADD21_12:%.*]] = sub nsw i32 [[CONV5_12]], [[CONV_12]]
-; CHECK-NEXT: [[MUL25_12:%.*]] = mul i32 [[ADD21_12]], [[CONV11_12]]
-; CHECK-NEXT: [[SHR26_12:%.*]] = lshr i32 [[MUL25_12]], 16
-; CHECK-NEXT: [[TMP81:%.*]] = trunc nuw i32 [[SHR26_12]] to i16
-; CHECK-NEXT: [[CONV28_12:%.*]] = sub i16 0, [[TMP81]]
-; CHECK-NEXT: br label [[IF_END_12:%.*]]
-; CHECK: if.then.12:
-; CHECK-NEXT: [[ADD_12:%.*]] = add nuw nsw i32 [[CONV5_12]], [[CONV_12]]
-; CHECK-NEXT: [[MUL_12:%.*]] = mul i32 [[ADD_12]], [[CONV11_12]]
-; CHECK-NEXT: [[SHR_12:%.*]] = lshr i32 [[MUL_12]], 16
-; CHECK-NEXT: [[CONV12_12:%.*]] = trunc nuw i32 [[SHR_12]] to i16
-; CHECK-NEXT: br label [[IF_END_12]]
-; CHECK: if.end.12:
-; 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_1248:%.*]] = or i16 [[OR_1147]], [[STOREMERGE_12]]
-; CHECK-NEXT: [[ARRAYIDX_13:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 26
-; CHECK-NEXT: [[TMP82:%.*]] = load i16, ptr [[ARRAYIDX_13]], align 2
-; CHECK-NEXT: [[CONV_13:%.*]] = sext i16 [[TMP82]] to i32
-; CHECK-NEXT: [[CMP1_13:%.*]] = icmp sgt i16 [[TMP82]], 0
-; CHECK-NEXT: [[ARRAYIDX4_13:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 26
-; CHECK-NEXT: [[TMP83:%.*]] = load i16, ptr [[ARRAYIDX4_13]], align 2
-; CHECK-NEXT: [[CONV5_13:%.*]] = zext i16 [[TMP83]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_13:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 26
-; CHECK-NEXT: [[TMP84:%.*]] = load i16, ptr [[ARRAYIDX10_13]], align 2
-; CHECK-NEXT: [[CONV11_13:%.*]] = zext i16 [[TMP84]] to i32
-; CHECK-NEXT: br i1 [[CMP1_13]], label [[IF_THEN_13:%.*]], label [[IF_ELSE_13:%.*]]
-; CHECK: if.else.13:
-; CHECK-NEXT: [[ADD21_13:%.*]] = sub nsw i32 [[CONV5_13]], [[CONV_13]]
-; CHECK-NEXT: [[MUL25_13:%.*]] = mul i32 [[ADD21_13]], [[CONV11_13]]
-; CHECK-NEXT: [[SHR26_13:%.*]] = lshr i32 [[MUL25_13]], 16
-; CHECK-NEXT: [[TMP85:%.*]] = trunc nuw i32 [[SHR26_13]] to i16
-; CHECK-NEXT: [[CONV28_13:%.*]] = sub i16 0, [[TMP85]]
-; CHECK-NEXT: br label [[IF_END_13:%.*]]
-; CHECK: if.then.13:
-; CHECK-NEXT: [[ADD_13:%.*]] = add nuw nsw i32 [[CONV5_13]], [[CONV_13]]
-; CHECK-NEXT: [[MUL_13:%.*]] = mul i32 [[ADD_13]], [[CONV11_13]]
-; CHECK-NEXT: [[SHR_13:%.*]] = lshr i32 [[MUL_13]], 16
-; CHECK-NEXT: [[CONV12_13:%.*]] = trunc nuw i32 [[SHR_13]] to i16
-; CHECK-NEXT: br label [[IF_END_13]]
-; CHECK: if.end.13:
-; 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_1349:%.*]] = or i16 [[OR_1248]], [[STOREMERGE_13]]
-; CHECK-NEXT: [[ARRAYIDX_14:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 28
-; CHECK-NEXT: [[TMP86:%.*]] = load i16, ptr [[ARRAYIDX_14]], align 2
-; CHECK-NEXT: [[CONV_14:%.*]] = sext i16 [[TMP86]] to i32
-; CHECK-NEXT: [[CMP1_14:%.*]] = icmp sgt i16 [[TMP86]], 0
-; CHECK-NEXT: [[ARRAYIDX4_14:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 28
-; CHECK-NEXT: [[TMP87:%.*]] = load i16, ptr [[ARRAYIDX4_14]], align 2
-; CHECK-NEXT: [[CONV5_14:%.*]] = zext i16 [[TMP87]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_14:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 28
-; CHECK-NEXT: [[TMP88:%.*]] = load i16, ptr [[ARRAYIDX10_14]], align 2
-; CHECK-NEXT: [[CONV11_14:%.*]] = zext i16 [[TMP88]] to i32
-; CHECK-NEXT: br i1 [[CMP1_14]], label [[IF_THEN_14:%.*]], label [[IF_ELSE_14:%.*]]
-; CHECK: if.else.14:
-; CHECK-NEXT: [[ADD21_14:%.*]] = sub nsw i32 [[CONV5_14]], [[CONV_14]]
-; CHECK-NEXT: [[MUL25_14:%.*]] = mul i32 [[ADD21_14]], [[CONV11_14]]
-; CHECK-NEXT: [[SHR26_14:%.*]] = lshr i32 [[MUL25_14]], 16
-; CHECK-NEXT: [[TMP89:%.*]] = trunc nuw i32 [[SHR26_14]] to i16
-; CHECK-NEXT: [[CONV28_14:%.*]] = sub i16 0, [[TMP89]]
-; CHECK-NEXT: br label [[IF_END_14:%.*]]
-; CHECK: if.then.14:
-; CHECK-NEXT: [[ADD_14:%.*]] = add nuw nsw i32 [[CONV5_14]], [[CONV_14]]
-; CHECK-NEXT: [[MUL_14:%.*]] = mul i32 [[ADD_14]], [[CONV11_14]]
-; CHECK-NEXT: [[SHR_14:%.*]] = lshr i32 [[MUL_14]], 16
-; CHECK-NEXT: [[CONV12_14:%.*]] = trunc nuw i32 [[SHR_14]] to i16
-; CHECK-NEXT: br label [[IF_END_14]]
-; CHECK: if.end.14:
-; 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_1450:%.*]] = or i16 [[OR_1349]], [[STOREMERGE_14]]
-; CHECK-NEXT: [[ARRAYIDX_15:%.*]] = getelementptr inbounds nuw i8, ptr [[DCT]], i64 30
-; CHECK-NEXT: [[TMP90:%.*]] = load i16, ptr [[ARRAYIDX_15]], align 2
-; CHECK-NEXT: [[CONV_15:%.*]] = sext i16 [[TMP90]] to i32
-; CHECK-NEXT: [[CMP1_15:%.*]] = icmp sgt i16 [[TMP90]], 0
-; CHECK-NEXT: [[ARRAYIDX4_15:%.*]] = getelementptr inbounds nuw i8, ptr [[BIAS]], i64 30
-; CHECK-NEXT: [[TMP91:%.*]] = load i16, ptr [[ARRAYIDX4_15]], align 2
-; CHECK-NEXT: [[CONV5_15:%.*]] = zext i16 [[TMP91]] to i32
-; CHECK-NEXT: [[ARRAYIDX10_15:%.*]] = getelementptr inbounds nuw i8, ptr [[MF]], i64 30
-; CHECK-NEXT: [[TMP92:%.*]] = load i16, ptr [[ARRAYIDX10_15]], align 2
-; CHECK-NEXT: [[CONV11_15:%.*]] = zext i16 [[TMP92]] to i32
-; CHECK-NEXT: br i1 [[CMP1_15]], label [[IF_THEN_15:%.*]], label [[IF_ELSE_15:%.*]]
-; CHECK: if.else.15:
-; CHECK-NEXT: [[ADD21_15:%.*]] = sub nsw i32 [[CONV5_15]], [[CONV_15]]
-; CHECK-NEXT: [[MUL25_15:%.*]] = mul i32 [[ADD21_15]], [[CONV11_15]]
-; CHECK-NEXT: [[SHR26_15:%.*]] = lshr i32 [[MUL25_15]], 16
-; CHECK-NEXT: [[TMP93:%.*]] = trunc nuw i32 [[SHR26_15]] to i16
-; CHECK-NEXT: [[CONV28_15:%.*]] = sub i16 0, [[TMP93]]
-; CHECK-NEXT: br label [[IF_END_15]]
-; CHECK: if.then.15:
-; CHECK-NEXT: [[ADD_15:%.*]] = add nuw nsw i32 [[CONV5_15]], [[CONV_15]]
-; CHECK-NEXT: [[MUL_15:%.*]] = mul i32 [[ADD_15]], [[CONV11_15]]
-; CHECK-NEXT: [[SHR_15:%.*]] = lshr i32 [[MUL_15]], 16
-; CHECK-NEXT: [[CONV12_15:%.*]] = trunc nuw i32 [[SHR_15]] to i16
-; CHECK-NEXT: br label [[IF_END_15]]
-; CHECK: if.end.15:
-; CHECK-NEXT: [[STOREMERGE_15:%.*]] = phi i16 [ [[CONV28_15]], [[IF_ELSE_15]] ], [ [[CONV12_15]], [[IF_THEN_15]] ]
-; CHECK-NEXT: store i16 [[STOREMERGE_15]], ptr [[ARRAYIDX_15]], align 2
-; CHECK-NEXT: [[OR_1551]] = or i16 [[OR_1450]], [[STOREMERGE_15]]
-; CHECK-NEXT: br label [[FOR_COND_CLEANUP]]
+; CHECK-NEXT: store i16 [[STOREMERGE]], ptr [[ARRAYIDX]], align 2
+; CHECK-NEXT: [[CONV33:%.*]] = sext i16 [[STOREMERGE]] to i32
+; CHECK-NEXT: [[OR]] = or i32 [[NZ_020]], [[CONV33]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
+; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 16
+; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
;
entry:
%dct.addr = alloca ptr, align 8
diff --git a/llvm/test/Transforms/PhaseOrdering/AArch64/reduce_submuladd.ll b/llvm/test/Transforms/PhaseOrdering/AArch64/reduce_submuladd.ll
index eefde9ddfdc1d..ab6de948c2c02 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/reduce_submuladd.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/reduce_submuladd.ll
@@ -11,101 +11,103 @@ define dso_local noundef nofpclass(nan inf) float @_Z4testPKfS0_ii(ptr noundef %
; CHECK-SAME: (ptr noundef readonly captures(none) [[TMP0:%.*]], ptr noundef readonly captures(none) [[TMP1:%.*]], i32 noundef [[TMP2:%.*]], i32 noundef [[TMP3:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: .preheader.i:
; CHECK-NEXT: [[TMP4:%.*]] = sext i32 [[TMP3]] to i64
-; CHECK-NEXT: [[TMP5:%.*]] = sext i32 [[TMP2]] to i64
-; CHECK-NEXT: [[TMP6:%.*]] = load <20 x float>, ptr [[TMP0]], align 4, !tbaa [[TBAA4:![0-9]+]]
-; CHECK-NEXT: [[TMP7:%.*]] = load <20 x float>, ptr [[TMP1]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP8:%.*]] = fsub fast <20 x float> [[TMP6]], [[TMP7]]
-; CHECK-NEXT: [[TMP9:%.*]] = fmul fast <20 x float> [[TMP8]], [[TMP8]]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 80
-; CHECK-NEXT: [[TMP11:%.*]] = load float, ptr [[TMP10]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 80
+; CHECK-NEXT: [[TMP5:%.*]] = shl nsw i64 [[TMP4]], 2
+; CHECK-NEXT: [[TMP6:%.*]] = sext i32 [[TMP2]] to i64
+; CHECK-NEXT: [[TMP7:%.*]] = shl nsw i64 [[TMP6]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = load <20 x float>, ptr [[TMP0]], align 4, !tbaa [[TBAA4:![0-9]+]]
+; CHECK-NEXT: [[TMP9:%.*]] = load <20 x float>, ptr [[TMP1]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP10:%.*]] = fsub fast <20 x float> [[TMP8]], [[TMP9]]
+; CHECK-NEXT: [[TMP11:%.*]] = fmul fast <20 x float> [[TMP10]], [[TMP10]]
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 80
; CHECK-NEXT: [[TMP13:%.*]] = load float, ptr [[TMP12]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP14:%.*]] = fsub fast float [[TMP11]], [[TMP13]]
-; CHECK-NEXT: [[TMP15:%.*]] = fmul fast float [[TMP14]], [[TMP14]]
-; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds float, ptr [[TMP0]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds float, ptr [[TMP1]], i64 [[TMP4]]
-; CHECK-NEXT: [[OP_RDX:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP15]], <20 x float> [[TMP9]])
-; CHECK-NEXT: [[TMP18:%.*]] = load <20 x float>, ptr [[TMP16]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP19:%.*]] = load <20 x float>, ptr [[TMP17]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP20:%.*]] = fsub fast <20 x float> [[TMP18]], [[TMP19]]
-; CHECK-NEXT: [[TMP21:%.*]] = fmul fast <20 x float> [[TMP20]], [[TMP20]]
-; CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP16]], i64 80
-; CHECK-NEXT: [[TMP23:%.*]] = load float, ptr [[TMP22]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP17]], i64 80
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 80
+; CHECK-NEXT: [[TMP15:%.*]] = load float, ptr [[TMP14]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP16:%.*]] = fsub fast float [[TMP13]], [[TMP15]]
+; CHECK-NEXT: [[TMP17:%.*]] = fmul fast float [[TMP16]], [[TMP16]]
+; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[TMP5]]
+; CHECK-NEXT: [[OP_RDX:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP17]], <20 x float> [[TMP11]])
+; CHECK-NEXT: [[TMP20:%.*]] = load <20 x float>, ptr [[TMP18]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP21:%.*]] = load <20 x float>, ptr [[TMP19]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP22:%.*]] = fsub fast <20 x float> [[TMP20]], [[TMP21]]
+; CHECK-NEXT: [[TMP23:%.*]] = fmul fast <20 x float> [[TMP22]], [[TMP22]]
+; CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP18]], i64 80
; CHECK-NEXT: [[TMP25:%.*]] = load float, ptr [[TMP24]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP26:%.*]] = fsub fast float [[TMP23]], [[TMP25]]
-; CHECK-NEXT: [[TMP27:%.*]] = fmul fast float [[TMP26]], [[TMP26]]
-; CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds float, ptr [[TMP16]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds float, ptr [[TMP17]], i64 [[TMP4]]
-; CHECK-NEXT: [[OP_RDX_1:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP27]], <20 x float> [[TMP21]])
+; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP19]], i64 80
+; CHECK-NEXT: [[TMP27:%.*]] = load float, ptr [[TMP26]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP28:%.*]] = fsub fast float [[TMP25]], [[TMP27]]
+; CHECK-NEXT: [[TMP29:%.*]] = fmul fast float [[TMP28]], [[TMP28]]
+; CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds i8, ptr [[TMP18]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[TMP19]], i64 [[TMP5]]
+; CHECK-NEXT: [[OP_RDX_1:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP29]], <20 x float> [[TMP23]])
; CHECK-NEXT: [[OP_RDX3_1:%.*]] = fadd fast float [[OP_RDX_1]], [[OP_RDX]]
-; CHECK-NEXT: [[TMP30:%.*]] = load <20 x float>, ptr [[TMP28]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP31:%.*]] = load <20 x float>, ptr [[TMP29]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP32:%.*]] = fsub fast <20 x float> [[TMP30]], [[TMP31]]
-; CHECK-NEXT: [[TMP33:%.*]] = fmul fast <20 x float> [[TMP32]], [[TMP32]]
-; CHECK-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP28]], i64 80
-; CHECK-NEXT: [[TMP35:%.*]] = load float, ptr [[TMP34]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP29]], i64 80
+; CHECK-NEXT: [[TMP32:%.*]] = load <20 x float>, ptr [[TMP30]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP33:%.*]] = load <20 x float>, ptr [[TMP31]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP34:%.*]] = fsub fast <20 x float> [[TMP32]], [[TMP33]]
+; CHECK-NEXT: [[TMP35:%.*]] = fmul fast <20 x float> [[TMP34]], [[TMP34]]
+; CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP30]], i64 80
; CHECK-NEXT: [[TMP37:%.*]] = load float, ptr [[TMP36]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP38:%.*]] = fsub fast float [[TMP35]], [[TMP37]]
-; CHECK-NEXT: [[TMP39:%.*]] = fmul fast float [[TMP38]], [[TMP38]]
-; CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds float, ptr [[TMP28]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds float, ptr [[TMP29]], i64 [[TMP4]]
-; CHECK-NEXT: [[OP_RDX_2:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP39]], <20 x float> [[TMP33]])
+; CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP31]], i64 80
+; CHECK-NEXT: [[TMP39:%.*]] = load float, ptr [[TMP38]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP40:%.*]] = fsub fast float [[TMP37]], [[TMP39]]
+; CHECK-NEXT: [[TMP41:%.*]] = fmul fast float [[TMP40]], [[TMP40]]
+; CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[TMP30]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds i8, ptr [[TMP31]], i64 [[TMP5]]
+; CHECK-NEXT: [[OP_RDX_2:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP41]], <20 x float> [[TMP35]])
; CHECK-NEXT: [[OP_RDX3_2:%.*]] = fadd fast float [[OP_RDX_2]], [[OP_RDX3_1]]
-; CHECK-NEXT: [[TMP42:%.*]] = load <20 x float>, ptr [[TMP40]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP43:%.*]] = load <20 x float>, ptr [[TMP41]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP44:%.*]] = fsub fast <20 x float> [[TMP42]], [[TMP43]]
-; CHECK-NEXT: [[TMP45:%.*]] = fmul fast <20 x float> [[TMP44]], [[TMP44]]
-; CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP40]], i64 80
-; CHECK-NEXT: [[TMP47:%.*]] = load float, ptr [[TMP46]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP41]], i64 80
+; CHECK-NEXT: [[TMP44:%.*]] = load <20 x float>, ptr [[TMP42]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP45:%.*]] = load <20 x float>, ptr [[TMP43]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP46:%.*]] = fsub fast <20 x float> [[TMP44]], [[TMP45]]
+; CHECK-NEXT: [[TMP47:%.*]] = fmul fast <20 x float> [[TMP46]], [[TMP46]]
+; CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP42]], i64 80
; CHECK-NEXT: [[TMP49:%.*]] = load float, ptr [[TMP48]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP50:%.*]] = fsub fast float [[TMP47]], [[TMP49]]
-; CHECK-NEXT: [[TMP51:%.*]] = fmul fast float [[TMP50]], [[TMP50]]
-; CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds float, ptr [[TMP40]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds float, ptr [[TMP41]], i64 [[TMP4]]
-; CHECK-NEXT: [[OP_RDX_3:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP51]], <20 x float> [[TMP45]])
+; CHECK-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP43]], i64 80
+; CHECK-NEXT: [[TMP51:%.*]] = load float, ptr [[TMP50]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP52:%.*]] = fsub fast float [[TMP49]], [[TMP51]]
+; CHECK-NEXT: [[TMP53:%.*]] = fmul fast float [[TMP52]], [[TMP52]]
+; CHECK-NEXT: [[TMP54:%.*]] = getelementptr inbounds i8, ptr [[TMP42]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds i8, ptr [[TMP43]], i64 [[TMP5]]
+; CHECK-NEXT: [[OP_RDX_3:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP53]], <20 x float> [[TMP47]])
; CHECK-NEXT: [[OP_RDX3_3:%.*]] = fadd fast float [[OP_RDX_3]], [[OP_RDX3_2]]
-; CHECK-NEXT: [[TMP54:%.*]] = load <20 x float>, ptr [[TMP52]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP55:%.*]] = load <20 x float>, ptr [[TMP53]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP56:%.*]] = fsub fast <20 x float> [[TMP54]], [[TMP55]]
-; CHECK-NEXT: [[TMP57:%.*]] = fmul fast <20 x float> [[TMP56]], [[TMP56]]
-; CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP52]], i64 80
-; CHECK-NEXT: [[TMP59:%.*]] = load float, ptr [[TMP58]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP53]], i64 80
+; CHECK-NEXT: [[TMP56:%.*]] = load <20 x float>, ptr [[TMP54]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP57:%.*]] = load <20 x float>, ptr [[TMP55]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP58:%.*]] = fsub fast <20 x float> [[TMP56]], [[TMP57]]
+; CHECK-NEXT: [[TMP59:%.*]] = fmul fast <20 x float> [[TMP58]], [[TMP58]]
+; CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP54]], i64 80
; CHECK-NEXT: [[TMP61:%.*]] = load float, ptr [[TMP60]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP62:%.*]] = fsub fast float [[TMP59]], [[TMP61]]
-; CHECK-NEXT: [[TMP63:%.*]] = fmul fast float [[TMP62]], [[TMP62]]
-; CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds float, ptr [[TMP52]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP65:%.*]] = getelementptr inbounds float, ptr [[TMP53]], i64 [[TMP4]]
-; CHECK-NEXT: [[OP_RDX_4:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP63]], <20 x float> [[TMP57]])
+; CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP55]], i64 80
+; CHECK-NEXT: [[TMP63:%.*]] = load float, ptr [[TMP62]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP64:%.*]] = fsub fast float [[TMP61]], [[TMP63]]
+; CHECK-NEXT: [[TMP65:%.*]] = fmul fast float [[TMP64]], [[TMP64]]
+; CHECK-NEXT: [[TMP66:%.*]] = getelementptr inbounds i8, ptr [[TMP54]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP67:%.*]] = getelementptr inbounds i8, ptr [[TMP55]], i64 [[TMP5]]
+; CHECK-NEXT: [[OP_RDX_4:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP65]], <20 x float> [[TMP59]])
; CHECK-NEXT: [[OP_RDX3_4:%.*]] = fadd fast float [[OP_RDX_4]], [[OP_RDX3_3]]
-; CHECK-NEXT: [[TMP66:%.*]] = load <20 x float>, ptr [[TMP64]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP67:%.*]] = load <20 x float>, ptr [[TMP65]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP68:%.*]] = fsub fast <20 x float> [[TMP66]], [[TMP67]]
-; CHECK-NEXT: [[TMP69:%.*]] = fmul fast <20 x float> [[TMP68]], [[TMP68]]
-; CHECK-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP64]], i64 80
-; CHECK-NEXT: [[TMP71:%.*]] = load float, ptr [[TMP70]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP65]], i64 80
+; CHECK-NEXT: [[TMP68:%.*]] = load <20 x float>, ptr [[TMP66]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP69:%.*]] = load <20 x float>, ptr [[TMP67]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP70:%.*]] = fsub fast <20 x float> [[TMP68]], [[TMP69]]
+; CHECK-NEXT: [[TMP71:%.*]] = fmul fast <20 x float> [[TMP70]], [[TMP70]]
+; CHECK-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP66]], i64 80
; CHECK-NEXT: [[TMP73:%.*]] = load float, ptr [[TMP72]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP74:%.*]] = fsub fast float [[TMP71]], [[TMP73]]
-; CHECK-NEXT: [[TMP75:%.*]] = fmul fast float [[TMP74]], [[TMP74]]
-; CHECK-NEXT: [[TMP76:%.*]] = getelementptr inbounds float, ptr [[TMP64]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP77:%.*]] = getelementptr inbounds float, ptr [[TMP65]], i64 [[TMP4]]
-; CHECK-NEXT: [[OP_RDX_5:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP75]], <20 x float> [[TMP69]])
+; CHECK-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP67]], i64 80
+; CHECK-NEXT: [[TMP75:%.*]] = load float, ptr [[TMP74]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP76:%.*]] = fsub fast float [[TMP73]], [[TMP75]]
+; CHECK-NEXT: [[TMP77:%.*]] = fmul fast float [[TMP76]], [[TMP76]]
+; CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds i8, ptr [[TMP66]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP79:%.*]] = getelementptr inbounds i8, ptr [[TMP67]], i64 [[TMP5]]
+; CHECK-NEXT: [[OP_RDX_5:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP77]], <20 x float> [[TMP71]])
; CHECK-NEXT: [[OP_RDX3_5:%.*]] = fadd fast float [[OP_RDX_5]], [[OP_RDX3_4]]
-; CHECK-NEXT: [[TMP78:%.*]] = load <20 x float>, ptr [[TMP76]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP79:%.*]] = load <20 x float>, ptr [[TMP77]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP80:%.*]] = fsub fast <20 x float> [[TMP78]], [[TMP79]]
-; CHECK-NEXT: [[TMP81:%.*]] = fmul fast <20 x float> [[TMP80]], [[TMP80]]
-; CHECK-NEXT: [[TMP82:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP76]], i64 80
-; CHECK-NEXT: [[TMP83:%.*]] = load float, ptr [[TMP82]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP84:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP77]], i64 80
+; CHECK-NEXT: [[TMP80:%.*]] = load <20 x float>, ptr [[TMP78]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP81:%.*]] = load <20 x float>, ptr [[TMP79]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP82:%.*]] = fsub fast <20 x float> [[TMP80]], [[TMP81]]
+; CHECK-NEXT: [[TMP83:%.*]] = fmul fast <20 x float> [[TMP82]], [[TMP82]]
+; CHECK-NEXT: [[TMP84:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP78]], i64 80
; CHECK-NEXT: [[TMP85:%.*]] = load float, ptr [[TMP84]], align 4, !tbaa [[TBAA4]]
-; CHECK-NEXT: [[TMP86:%.*]] = fsub fast float [[TMP83]], [[TMP85]]
-; CHECK-NEXT: [[TMP87:%.*]] = fmul fast float [[TMP86]], [[TMP86]]
-; CHECK-NEXT: [[OP_RDX_6:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP87]], <20 x float> [[TMP81]])
+; CHECK-NEXT: [[TMP86:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP79]], i64 80
+; CHECK-NEXT: [[TMP87:%.*]] = load float, ptr [[TMP86]], align 4, !tbaa [[TBAA4]]
+; CHECK-NEXT: [[TMP88:%.*]] = fsub fast float [[TMP85]], [[TMP87]]
+; CHECK-NEXT: [[TMP89:%.*]] = fmul fast float [[TMP88]], [[TMP88]]
+; CHECK-NEXT: [[OP_RDX_6:%.*]] = tail call fast float @llvm.vector.reduce.fadd.v20f32(float [[TMP89]], <20 x float> [[TMP83]])
; CHECK-NEXT: [[OP_RDX3_6:%.*]] = fadd fast float [[OP_RDX_6]], [[OP_RDX3_5]]
; CHECK-NEXT: ret float [[OP_RDX3_6]]
;
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 eda54d999a79f..eccf2ecf19c77 100644
--- a/llvm/test/Transforms/PhaseOrdering/AArch64/sinking-vs-if-conversion.ll
+++ b/llvm/test/Transforms/PhaseOrdering/AArch64/sinking-vs-if-conversion.ll
@@ -25,7 +25,8 @@ define void @test_find_min(ptr noundef nonnull align 8 dereferenceable(24) %this
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[FOR_BODY_LR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[COND_END7:%.*]] ]
; CHECK-NEXT: [[MIN_010:%.*]] = phi ptr [ [[TMP1]], [[FOR_BODY_LR_PH]] ], [ [[COND8:%.*]], [[COND_END7]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw ptr, ptr [[TMP2]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP2]], i64 [[TMP6]]
; CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[ARRAYIDX]], align 8
; CHECK-NEXT: [[CMP3:%.*]] = icmp eq ptr [[MIN_010]], null
; CHECK-NEXT: br i1 [[CMP3]], label [[COND_END7]], label [[COND_FALSE:%.*]]
@@ -143,17 +144,18 @@ define void @cond_select_loop(ptr noalias nocapture noundef readonly %a, ptr noa
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[I_07:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[COND_END:%.*]] ]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[I_07]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[I_07]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP0:%.*]] = load float, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[CMP2:%.*]] = fcmp ogt float [[TMP0]], 0.000000e+00
; CHECK-NEXT: br i1 [[CMP2]], label [[COND_END]], label [[COND_FALSE:%.*]]
; CHECK: cond.false:
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[B]], i64 [[I_07]]
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP2]]
; CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: br label [[COND_END]]
; CHECK: cond.end:
; CHECK-NEXT: [[COND:%.*]] = phi float [ [[TMP1]], [[COND_FALSE]] ], [ [[TMP0]], [[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw float, ptr [[C]], i64 [[I_07]]
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr [[C]], i64 [[TMP2]]
; CHECK-NEXT: store float [[COND]], ptr [[ARRAYIDX4]], align 4
; CHECK-NEXT: [[INC]] = add nuw nsw i64 [[I_07]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], 1000
diff --git a/llvm/test/Transforms/PhaseOrdering/SystemZ/sub-xor.ll b/llvm/test/Transforms/PhaseOrdering/SystemZ/sub-xor.ll
index 13eed2e918aa0..d2aba99355d73 100644
--- a/llvm/test/Transforms/PhaseOrdering/SystemZ/sub-xor.ll
+++ b/llvm/test/Transforms/PhaseOrdering/SystemZ/sub-xor.ll
@@ -16,124 +16,163 @@ define dso_local zeroext i32 @foo(ptr noundef %a) #0 {
; CHECK-SAME: ptr noundef readnone captures(none) [[A:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[ENTRY:.*]]:
; CHECK-NEXT: tail call void @populate(ptr noundef nonnull @ARR) #[[ATTR2:[0-9]+]]
-; CHECK-NEXT: br label %[[FOR_BODY4:.*]]
-; CHECK: [[FOR_BODY4]]:
-; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDVARS_IV_NEXT_7:%.*]], %[[FOR_BODY4]] ]
-; CHECK-NEXT: [[SUM_11:%.*]] = phi i32 [ 0, %[[ENTRY]] ], [ [[ADD_7:%.*]], %[[FOR_BODY4]] ]
-; CHECK-NEXT: [[IDX_NEG:%.*]] = sub nsw i64 0, [[INDVARS_IV]]
-; CHECK-NEXT: [[ADD_PTR:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[IDX_NEG]]
-; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ADD_PTR]], align 4, !tbaa [[INT_TBAA3:![0-9]+]]
-; CHECK-NEXT: [[ADD:%.*]] = add i32 [[TMP0]], [[SUM_11]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_NEG:%.*]] = xor i64 [[INDVARS_IV]], -1
-; CHECK-NEXT: [[ADD_PTR_110:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_NEG]]
-; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[ADD_PTR_110]], align 4, !tbaa [[INT_TBAA3]]
-; CHECK-NEXT: [[ADD_111:%.*]] = add i32 [[TMP1]], [[ADD]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_112_NEG:%.*]] = sub nuw nsw i64 -2, [[INDVARS_IV]]
-; CHECK-NEXT: [[ADD_PTR_217:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_112_NEG]]
-; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[ADD_PTR_217]], align 4, !tbaa [[INT_TBAA3]]
-; CHECK-NEXT: [[ADD_218:%.*]] = add i32 [[TMP2]], [[ADD_111]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_219_NEG:%.*]] = sub nuw nsw i64 -3, [[INDVARS_IV]]
-; CHECK-NEXT: [[ADD_PTR_3:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_219_NEG]]
-; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[ADD_PTR_3]], align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), align 4, !tbaa [[INT_TBAA3:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 392), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_112:%.*]] = add i32 [[TMP1]], [[TMP0]]
+; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 388), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_218:%.*]] = add i32 [[TMP2]], [[ADD_112]]
+; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 384), align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[ADD_3:%.*]] = add i32 [[TMP3]], [[ADD_218]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_3_NEG:%.*]] = sub nuw nsw i64 -4, [[INDVARS_IV]]
-; CHECK-NEXT: [[ADD_PTR_4:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_3_NEG]]
-; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr [[ADD_PTR_4]], align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[TMP4:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 380), align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[ADD_4:%.*]] = add i32 [[TMP4]], [[ADD_3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_4_NEG:%.*]] = sub nuw nsw i64 -5, [[INDVARS_IV]]
-; CHECK-NEXT: [[ADD_PTR_5:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_4_NEG]]
-; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ADD_PTR_5]], align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 376), align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[ADD_5:%.*]] = add i32 [[TMP5]], [[ADD_4]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_5_NEG:%.*]] = sub nuw nsw i64 -6, [[INDVARS_IV]]
-; CHECK-NEXT: [[ADD_PTR_6:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_5_NEG]]
-; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ADD_PTR_6]], align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 372), align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[ADD_6:%.*]] = add i32 [[TMP6]], [[ADD_5]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_6_NEG:%.*]] = sub nuw nsw i64 -7, [[INDVARS_IV]]
-; CHECK-NEXT: [[ADD_PTR_7:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_6_NEG]]
-; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ADD_PTR_7]], align 4, !tbaa [[INT_TBAA3]]
-; CHECK-NEXT: [[ADD_7]] = add i32 [[TMP7]], [[ADD_6]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_7]] = add nuw nsw i64 [[INDVARS_IV]], 8
-; CHECK-NEXT: [[EXITCOND_NOT_7:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT_7]], 32
-; CHECK-NEXT: br i1 [[EXITCOND_NOT_7]], label %[[FOR_BODY4_1:.*]], label %[[FOR_BODY4]], !llvm.loop [[LOOP7:![0-9]+]]
-; CHECK: [[FOR_BODY4_1]]:
-; CHECK-NEXT: [[INDVARS_IV_1:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_1_7:%.*]], %[[FOR_BODY4_1]] ], [ 0, %[[FOR_BODY4]] ]
-; CHECK-NEXT: [[SUM_11_1:%.*]] = phi i32 [ [[ADD_1_7:%.*]], %[[FOR_BODY4_1]] ], [ [[ADD_7]], %[[FOR_BODY4]] ]
-; CHECK-NEXT: [[IDX_NEG_1:%.*]] = sub nsw i64 0, [[INDVARS_IV_1]]
-; CHECK-NEXT: [[ADD_PTR_1:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[IDX_NEG_1]]
+; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 368), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_7:%.*]] = add i32 [[TMP7]], [[ADD_6]]
+; CHECK-NEXT: [[TMP32:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 364), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_8:%.*]] = add i32 [[TMP32]], [[ADD_7]]
+; CHECK-NEXT: [[TMP33:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 360), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_9:%.*]] = add i32 [[TMP33]], [[ADD_8]]
+; CHECK-NEXT: [[TMP34:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 356), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_10:%.*]] = add i32 [[TMP34]], [[ADD_9]]
+; CHECK-NEXT: [[TMP36:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 352), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_11:%.*]] = add i32 [[TMP36]], [[ADD_10]]
+; CHECK-NEXT: [[TMP37:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 348), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_12:%.*]] = add i32 [[TMP37]], [[ADD_11]]
+; CHECK-NEXT: [[TMP39:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 344), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_13:%.*]] = add i32 [[TMP39]], [[ADD_12]]
+; CHECK-NEXT: [[TMP40:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 340), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_14:%.*]] = add i32 [[TMP40]], [[ADD_13]]
+; CHECK-NEXT: [[TMP42:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 336), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_15:%.*]] = add i32 [[TMP42]], [[ADD_14]]
+; CHECK-NEXT: [[TMP43:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 332), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_16:%.*]] = add i32 [[TMP43]], [[ADD_15]]
+; CHECK-NEXT: [[TMP45:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 328), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_17:%.*]] = add i32 [[TMP45]], [[ADD_16]]
+; CHECK-NEXT: [[TMP46:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 324), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_18:%.*]] = add i32 [[TMP46]], [[ADD_17]]
+; CHECK-NEXT: [[TMP48:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 320), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_19:%.*]] = add i32 [[TMP48]], [[ADD_18]]
+; CHECK-NEXT: [[TMP49:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 316), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_20:%.*]] = add i32 [[TMP49]], [[ADD_19]]
+; CHECK-NEXT: [[TMP51:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 312), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_21:%.*]] = add i32 [[TMP51]], [[ADD_20]]
+; CHECK-NEXT: [[TMP52:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 308), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_22:%.*]] = add i32 [[TMP52]], [[ADD_21]]
+; CHECK-NEXT: [[TMP53:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 304), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_23:%.*]] = add i32 [[TMP53]], [[ADD_22]]
+; CHECK-NEXT: [[TMP54:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 300), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_24:%.*]] = add i32 [[TMP54]], [[ADD_23]]
+; CHECK-NEXT: [[TMP55:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 296), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_25:%.*]] = add i32 [[TMP55]], [[ADD_24]]
+; CHECK-NEXT: [[TMP57:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 292), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_26:%.*]] = add i32 [[TMP57]], [[ADD_25]]
+; CHECK-NEXT: [[TMP59:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 288), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_27:%.*]] = add i32 [[TMP59]], [[ADD_26]]
+; CHECK-NEXT: [[TMP61:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 284), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_28:%.*]] = add i32 [[TMP61]], [[ADD_27]]
+; CHECK-NEXT: [[TMP63:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 280), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_29:%.*]] = add i32 [[TMP63]], [[ADD_28]]
+; CHECK-NEXT: [[TMP65:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 276), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_30:%.*]] = add i32 [[TMP65]], [[ADD_29]]
+; CHECK-NEXT: [[TMP67:%.*]] = load i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 272), align 4, !tbaa [[INT_TBAA3]]
+; CHECK-NEXT: [[ADD_31:%.*]] = add i32 [[TMP67]], [[ADD_30]]
+; CHECK-NEXT: br label %[[FOR_BODY4:.*]]
+; CHECK: [[FOR_BODY4]]:
+; CHECK-NEXT: [[INDVARS_IV_1:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDVARS_IV_NEXT_1_7:%.*]], %[[FOR_BODY4]] ]
+; CHECK-NEXT: [[SUM_11_1:%.*]] = phi i32 [ [[ADD_31]], %[[ENTRY]] ], [ [[ADD_1_7:%.*]], %[[FOR_BODY4]] ]
+; CHECK-NEXT: [[DOTNEG7:%.*]] = mul nsw i64 [[INDVARS_IV_1]], -4
+; CHECK-NEXT: [[ADD_PTR_1:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7]]
; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ADD_PTR_1]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_1_NEG:%.*]] = xor i64 [[INDVARS_IV_1]], -1
-; CHECK-NEXT: [[ADD_PTR_1_1:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_1_NEG]]
+; CHECK-NEXT: [[DOTNEG7_1:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_1_NEG]], 2
+; CHECK-NEXT: [[ADD_PTR_1_1:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7_1]]
; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[ADD_PTR_1_1]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[TMP10:%.*]] = add i32 [[TMP8]], [[TMP9]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_1_1_NEG:%.*]] = sub nuw nsw i64 -2, [[INDVARS_IV_1]]
-; CHECK-NEXT: [[ADD_PTR_1_2:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_1_1_NEG]]
+; CHECK-NEXT: [[TMP35:%.*]] = shl i64 [[INDVARS_IV_1]], 2
+; CHECK-NEXT: [[DOTNEG7_2:%.*]] = sub nuw nsw i64 -8, [[TMP35]]
+; CHECK-NEXT: [[ADD_PTR_1_2:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7_2]]
; CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[ADD_PTR_1_2]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[TMP12:%.*]] = add i32 [[TMP10]], [[TMP11]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_1_2_NEG:%.*]] = sub nuw nsw i64 -3, [[INDVARS_IV_1]]
-; CHECK-NEXT: [[ADD_PTR_1_3:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_1_2_NEG]]
+; CHECK-NEXT: [[TMP38:%.*]] = shl i64 [[INDVARS_IV_1]], 2
+; CHECK-NEXT: [[DOTNEG7_3:%.*]] = sub nuw nsw i64 -12, [[TMP38]]
+; CHECK-NEXT: [[ADD_PTR_1_3:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7_3]]
; CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[ADD_PTR_1_3]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[TMP14:%.*]] = add i32 [[TMP12]], [[TMP13]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_1_3_NEG:%.*]] = sub nuw nsw i64 -4, [[INDVARS_IV_1]]
-; CHECK-NEXT: [[ADD_PTR_1_4:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_1_3_NEG]]
+; CHECK-NEXT: [[TMP41:%.*]] = shl i64 [[INDVARS_IV_1]], 2
+; CHECK-NEXT: [[DOTNEG7_4:%.*]] = sub nuw nsw i64 -16, [[TMP41]]
+; CHECK-NEXT: [[ADD_PTR_1_4:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7_4]]
; CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[ADD_PTR_1_4]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[TMP16:%.*]] = add i32 [[TMP14]], [[TMP15]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_1_4_NEG:%.*]] = sub nuw nsw i64 -5, [[INDVARS_IV_1]]
-; CHECK-NEXT: [[ADD_PTR_1_5:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_1_4_NEG]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl i64 [[INDVARS_IV_1]], 2
+; CHECK-NEXT: [[DOTNEG7_5:%.*]] = sub nuw nsw i64 -20, [[TMP44]]
+; CHECK-NEXT: [[ADD_PTR_1_5:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7_5]]
; CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[ADD_PTR_1_5]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[TMP18:%.*]] = add i32 [[TMP16]], [[TMP17]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_1_5_NEG:%.*]] = sub nuw nsw i64 -6, [[INDVARS_IV_1]]
-; CHECK-NEXT: [[ADD_PTR_1_6:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_1_5_NEG]]
+; CHECK-NEXT: [[TMP47:%.*]] = shl i64 [[INDVARS_IV_1]], 2
+; CHECK-NEXT: [[DOTNEG7_6:%.*]] = sub nuw nsw i64 -24, [[TMP47]]
+; CHECK-NEXT: [[ADD_PTR_1_6:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7_6]]
; CHECK-NEXT: [[TMP19:%.*]] = load i32, ptr [[ADD_PTR_1_6]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[TMP20:%.*]] = add i32 [[TMP18]], [[TMP19]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_1_6_NEG:%.*]] = sub nuw nsw i64 -7, [[INDVARS_IV_1]]
-; CHECK-NEXT: [[ADD_PTR_1_7:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_1_6_NEG]]
+; CHECK-NEXT: [[TMP50:%.*]] = shl i64 [[INDVARS_IV_1]], 2
+; CHECK-NEXT: [[DOTNEG7_7:%.*]] = sub nuw nsw i64 -28, [[TMP50]]
+; CHECK-NEXT: [[ADD_PTR_1_7:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG7_7]]
; CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[ADD_PTR_1_7]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[TMP22:%.*]] = add i32 [[TMP20]], [[TMP21]]
; CHECK-NEXT: [[TMP23:%.*]] = shl i32 [[TMP22]], 1
; CHECK-NEXT: [[ADD_1_7]] = add i32 [[TMP23]], [[SUM_11_1]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_1_7]] = add nuw nsw i64 [[INDVARS_IV_1]], 8
; CHECK-NEXT: [[EXITCOND_1_NOT_7:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT_1_7]], 32
-; CHECK-NEXT: br i1 [[EXITCOND_1_NOT_7]], label %[[FOR_BODY4_2:.*]], label %[[FOR_BODY4_1]], !llvm.loop [[LOOP7]]
+; CHECK-NEXT: br i1 [[EXITCOND_1_NOT_7]], label %[[FOR_BODY4_2:.*]], label %[[FOR_BODY4]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK: [[FOR_BODY4_2]]:
-; CHECK-NEXT: [[INDVARS_IV_2:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_2_7:%.*]], %[[FOR_BODY4_2]] ], [ 0, %[[FOR_BODY4_1]] ]
-; CHECK-NEXT: [[SUM_11_2:%.*]] = phi i32 [ [[ADD_2_7:%.*]], %[[FOR_BODY4_2]] ], [ [[ADD_1_7]], %[[FOR_BODY4_1]] ]
-; CHECK-NEXT: [[IDX_NEG_2:%.*]] = sub nsw i64 0, [[INDVARS_IV_2]]
-; CHECK-NEXT: [[ADD_PTR_2:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[IDX_NEG_2]]
+; CHECK-NEXT: [[INDVARS_IV_2:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_2_7:%.*]], %[[FOR_BODY4_2]] ], [ 0, %[[FOR_BODY4]] ]
+; CHECK-NEXT: [[SUM_11_2:%.*]] = phi i32 [ [[ADD_2_7:%.*]], %[[FOR_BODY4_2]] ], [ [[ADD_1_7]], %[[FOR_BODY4]] ]
+; CHECK-NEXT: [[DOTNEG8:%.*]] = mul nsw i64 [[INDVARS_IV_2]], -4
+; CHECK-NEXT: [[ADD_PTR_2:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8]]
; CHECK-NEXT: [[TMP24:%.*]] = load i32, ptr [[ADD_PTR_2]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2:%.*]] = mul i32 [[TMP24]], 3
; CHECK-NEXT: [[ADD_2:%.*]] = add i32 [[MUL_2]], [[SUM_11_2]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_2_NEG:%.*]] = xor i64 [[INDVARS_IV_2]], -1
-; CHECK-NEXT: [[ADD_PTR_2_1:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_2_NEG]]
+; CHECK-NEXT: [[DOTNEG8_1:%.*]] = shl nsw i64 [[INDVARS_IV_NEXT_2_NEG]], 2
+; CHECK-NEXT: [[ADD_PTR_2_1:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8_1]]
; CHECK-NEXT: [[TMP25:%.*]] = load i32, ptr [[ADD_PTR_2_1]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2_1:%.*]] = mul i32 [[TMP25]], 3
; CHECK-NEXT: [[ADD_2_1:%.*]] = add i32 [[MUL_2_1]], [[ADD_2]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_2_1_NEG:%.*]] = sub nuw nsw i64 -2, [[INDVARS_IV_2]]
-; CHECK-NEXT: [[ADD_PTR_2_2:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_2_1_NEG]]
+; CHECK-NEXT: [[TMP56:%.*]] = shl i64 [[INDVARS_IV_2]], 2
+; CHECK-NEXT: [[DOTNEG8_2:%.*]] = sub nuw nsw i64 -8, [[TMP56]]
+; CHECK-NEXT: [[ADD_PTR_2_2:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8_2]]
; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[ADD_PTR_2_2]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2_2:%.*]] = mul i32 [[TMP26]], 3
; CHECK-NEXT: [[ADD_2_2:%.*]] = add i32 [[MUL_2_2]], [[ADD_2_1]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_2_2_NEG:%.*]] = sub nuw nsw i64 -3, [[INDVARS_IV_2]]
-; CHECK-NEXT: [[ADD_PTR_2_3:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_2_2_NEG]]
+; CHECK-NEXT: [[TMP58:%.*]] = shl i64 [[INDVARS_IV_2]], 2
+; CHECK-NEXT: [[DOTNEG8_3:%.*]] = sub nuw nsw i64 -12, [[TMP58]]
+; CHECK-NEXT: [[ADD_PTR_2_3:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8_3]]
; CHECK-NEXT: [[TMP27:%.*]] = load i32, ptr [[ADD_PTR_2_3]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2_3:%.*]] = mul i32 [[TMP27]], 3
; CHECK-NEXT: [[ADD_2_3:%.*]] = add i32 [[MUL_2_3]], [[ADD_2_2]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_2_3_NEG:%.*]] = sub nuw nsw i64 -4, [[INDVARS_IV_2]]
-; CHECK-NEXT: [[ADD_PTR_2_4:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_2_3_NEG]]
+; CHECK-NEXT: [[TMP60:%.*]] = shl i64 [[INDVARS_IV_2]], 2
+; CHECK-NEXT: [[DOTNEG8_4:%.*]] = sub nuw nsw i64 -16, [[TMP60]]
+; CHECK-NEXT: [[ADD_PTR_2_4:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8_4]]
; CHECK-NEXT: [[TMP28:%.*]] = load i32, ptr [[ADD_PTR_2_4]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2_4:%.*]] = mul i32 [[TMP28]], 3
; CHECK-NEXT: [[ADD_2_4:%.*]] = add i32 [[MUL_2_4]], [[ADD_2_3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_2_4_NEG:%.*]] = sub nuw nsw i64 -5, [[INDVARS_IV_2]]
-; CHECK-NEXT: [[ADD_PTR_2_5:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_2_4_NEG]]
+; CHECK-NEXT: [[TMP62:%.*]] = shl i64 [[INDVARS_IV_2]], 2
+; CHECK-NEXT: [[DOTNEG8_5:%.*]] = sub nuw nsw i64 -20, [[TMP62]]
+; CHECK-NEXT: [[ADD_PTR_2_5:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8_5]]
; CHECK-NEXT: [[TMP29:%.*]] = load i32, ptr [[ADD_PTR_2_5]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2_5:%.*]] = mul i32 [[TMP29]], 3
; CHECK-NEXT: [[ADD_2_5:%.*]] = add i32 [[MUL_2_5]], [[ADD_2_4]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_2_5_NEG:%.*]] = sub nuw nsw i64 -6, [[INDVARS_IV_2]]
-; CHECK-NEXT: [[ADD_PTR_2_6:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_2_5_NEG]]
+; CHECK-NEXT: [[TMP64:%.*]] = shl i64 [[INDVARS_IV_2]], 2
+; CHECK-NEXT: [[DOTNEG8_6:%.*]] = sub nuw nsw i64 -24, [[TMP64]]
+; CHECK-NEXT: [[ADD_PTR_2_6:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8_6]]
; CHECK-NEXT: [[TMP30:%.*]] = load i32, ptr [[ADD_PTR_2_6]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2_6:%.*]] = mul i32 [[TMP30]], 3
; CHECK-NEXT: [[ADD_2_6:%.*]] = add i32 [[MUL_2_6]], [[ADD_2_5]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_2_6_NEG:%.*]] = sub nuw nsw i64 -7, [[INDVARS_IV_2]]
-; CHECK-NEXT: [[ADD_PTR_2_7:%.*]] = getelementptr inbounds i32, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[INDVARS_IV_NEXT_2_6_NEG]]
+; CHECK-NEXT: [[TMP66:%.*]] = shl i64 [[INDVARS_IV_2]], 2
+; CHECK-NEXT: [[DOTNEG8_7:%.*]] = sub nuw nsw i64 -28, [[TMP66]]
+; CHECK-NEXT: [[ADD_PTR_2_7:%.*]] = getelementptr inbounds i8, ptr getelementptr inbounds nuw (i8, ptr @ARR, i64 396), i64 [[DOTNEG8_7]]
; CHECK-NEXT: [[TMP31:%.*]] = load i32, ptr [[ADD_PTR_2_7]], align 4, !tbaa [[INT_TBAA3]]
; CHECK-NEXT: [[MUL_2_7:%.*]] = mul i32 [[TMP31]], 3
; CHECK-NEXT: [[ADD_2_7]] = add i32 [[MUL_2_7]], [[ADD_2_6]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/SROA-after-final-loop-unrolling-2.ll b/llvm/test/Transforms/PhaseOrdering/X86/SROA-after-final-loop-unrolling-2.ll
index f42101ffe89aa..32ead07ca3cef 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/SROA-after-final-loop-unrolling-2.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/SROA-after-final-loop-unrolling-2.ll
@@ -23,35 +23,39 @@ define dso_local void @foo(i32 noundef %arg, ptr noundef nonnull align 4 derefer
; CHECK-LABEL: define dso_local void @foo(
; CHECK-SAME: i32 noundef [[ARG:%.*]], ptr noundef nonnull writeonly align 4 captures(none) dereferenceable(8) [[ARG1:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[BB:.*]]:
+; CHECK-NEXT: [[I3:%.*]] = alloca [[T0:%.*]], align 8
+; CHECK-NEXT: call void @llvm.lifetime.start.p0(ptr nonnull [[I3]]) #[[ATTR2:[0-9]+]]
+; CHECK-NEXT: store i64 180388626456, ptr [[I3]], align 8
; CHECK-NEXT: [[I9:%.*]] = sdiv i32 [[ARG]], 128
; CHECK-NEXT: [[I10:%.*]] = shl nsw i32 [[I9]], 7
; CHECK-NEXT: [[ARG_OFF:%.*]] = add i32 [[ARG]], 127
; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[ARG_OFF]], 255
; CHECK-NEXT: br i1 [[TMP0]], label %[[BB12:.*]], label %[[BB13:.*]]
+; CHECK: [[BB13]]:
+; CHECK-NEXT: [[I5_I_I_1:%.*]] = getelementptr inbounds nuw i8, ptr [[I3]], i64 4
+; CHECK-NEXT: [[I3_PROMOTED:%.*]] = load i32, ptr [[I3]], align 8, !tbaa [[INT_TBAA5:![0-9]+]]
+; CHECK-NEXT: [[I5_I_I_1_PROMOTED:%.*]] = load i32, ptr [[I5_I_I_1]], align 4, !tbaa [[INT_TBAA5]]
+; CHECK-NEXT: br label %[[BB14:.*]]
; CHECK: [[BB12_LOOPEXIT:.*]]:
-; CHECK-NEXT: [[I3_SROA_8_0_INSERT_EXT:%.*]] = zext i32 [[I21_3:%.*]] to i64
-; CHECK-NEXT: [[I3_SROA_8_0_INSERT_SHIFT:%.*]] = shl nuw i64 [[I3_SROA_8_0_INSERT_EXT]], 32
-; CHECK-NEXT: [[I3_SROA_0_0_INSERT_EXT:%.*]] = zext i32 [[I21_2:%.*]] to i64
-; CHECK-NEXT: [[I3_SROA_0_0_INSERT_INSERT:%.*]] = or disjoint i64 [[I3_SROA_8_0_INSERT_SHIFT]], [[I3_SROA_0_0_INSERT_EXT]]
+; CHECK-NEXT: store i32 [[I21:%.*]], ptr [[I3]], align 8, !tbaa [[INT_TBAA5]]
+; CHECK-NEXT: store i32 [[I21_1:%.*]], ptr [[I5_I_I_1]], align 4, !tbaa [[INT_TBAA5]]
+; CHECK-NEXT: [[DOTPRE:%.*]] = load i64, ptr [[I3]], align 8, !tbaa [[CHAR_TBAA9:![0-9]+]]
; CHECK-NEXT: br label %[[BB12]]
; CHECK: [[BB12]]:
-; CHECK-NEXT: [[TMP1:%.*]] = phi i64 [ [[I3_SROA_0_0_INSERT_INSERT]], %[[BB12_LOOPEXIT]] ], [ 180388626456, %[[BB]] ]
-; CHECK-NEXT: store i64 [[TMP1]], ptr [[ARG1]], align 4, !tbaa [[CHAR_TBAA5:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = phi i64 [ [[DOTPRE]], %[[BB12_LOOPEXIT]] ], [ 180388626456, %[[BB]] ]
+; CHECK-NEXT: store i64 [[TMP1]], ptr [[ARG1]], align 4, !tbaa [[CHAR_TBAA9]]
+; CHECK-NEXT: call void @llvm.lifetime.end.p0(ptr nonnull [[I3]]) #[[ATTR2]]
; CHECK-NEXT: ret void
-; CHECK: [[BB13]]:
-; CHECK-NEXT: [[I3_SROA_8_0:%.*]] = phi i32 [ [[I21_3]], %[[BB13]] ], [ 42, %[[BB]] ]
-; CHECK-NEXT: [[I3_SROA_0_0:%.*]] = phi i32 [ [[I21_2]], %[[BB13]] ], [ 24, %[[BB]] ]
-; CHECK-NEXT: [[I4_05:%.*]] = phi i32 [ [[I24_3:%.*]], %[[BB13]] ], [ 0, %[[BB]] ]
-; CHECK-NEXT: [[I21:%.*]] = mul nsw i32 [[I3_SROA_0_0]], [[I4_05]]
+; CHECK: [[BB14]]:
+; CHECK-NEXT: [[I3_SROA_8_0:%.*]] = phi i32 [ [[I5_I_I_1_PROMOTED]], %[[BB13]] ], [ [[I21_1]], %[[BB14]] ]
+; CHECK-NEXT: [[I3_SROA_0_0:%.*]] = phi i32 [ [[I3_PROMOTED]], %[[BB13]] ], [ [[I21]], %[[BB14]] ]
+; CHECK-NEXT: [[I4_05:%.*]] = phi i32 [ 0, %[[BB13]] ], [ [[I24_3:%.*]], %[[BB14]] ]
+; CHECK-NEXT: [[I21]] = mul nsw i32 [[I3_SROA_0_0]], [[I4_05]]
; CHECK-NEXT: [[I24:%.*]] = or disjoint i32 [[I4_05]], 1
-; CHECK-NEXT: [[I21_1:%.*]] = mul nsw i32 [[I3_SROA_8_0]], [[I24]]
-; CHECK-NEXT: [[I24_1:%.*]] = or disjoint i32 [[I4_05]], 2
-; CHECK-NEXT: [[I21_2]] = mul nsw i32 [[I21]], [[I24_1]]
-; CHECK-NEXT: [[I24_2:%.*]] = or disjoint i32 [[I4_05]], 3
-; CHECK-NEXT: [[I21_3]] = mul nsw i32 [[I21_1]], [[I24_2]]
-; CHECK-NEXT: [[I24_3]] = add nuw nsw i32 [[I4_05]], 4
+; CHECK-NEXT: [[I21_1]] = mul nsw i32 [[I3_SROA_8_0]], [[I24]]
+; CHECK-NEXT: [[I24_3]] = add nuw nsw i32 [[I4_05]], 2
; CHECK-NEXT: [[I11_NOT_3:%.*]] = icmp eq i32 [[I24_3]], [[I10]]
-; CHECK-NEXT: br i1 [[I11_NOT_3]], label %[[BB12_LOOPEXIT]], label %[[BB13]], !llvm.loop [[LOOP8:![0-9]+]]
+; CHECK-NEXT: br i1 [[I11_NOT_3]], label %[[BB12_LOOPEXIT]], label %[[BB14]], !llvm.loop [[LOOP10:![0-9]+]]
;
bb:
%i = alloca i32, align 4
@@ -168,10 +172,11 @@ attributes #3 = { nounwind }
!15 = !{!16, !16, i64 0}
!16 = !{!"long", !7, i64 0}
;.
-; CHECK: [[CHAR_TBAA5]] = !{[[META6:![0-9]+]], [[META6]], i64 0}
-; CHECK: [[META6]] = !{!"omnipotent char", [[META7:![0-9]+]], i64 0}
-; CHECK: [[META7]] = !{!"Simple C++ TBAA"}
-; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META9:![0-9]+]], [[META10:![0-9]+]]}
-; CHECK: [[META9]] = !{!"llvm.loop.mustprogress"}
-; CHECK: [[META10]] = !{!"llvm.loop.isvectorized", i32 1}
+; CHECK: [[INT_TBAA5]] = !{[[META6:![0-9]+]], [[META6]], i64 0}
+; CHECK: [[META6]] = !{!"int", [[META7:![0-9]+]], i64 0}
+; CHECK: [[META7]] = !{!"omnipotent char", [[META8:![0-9]+]], i64 0}
+; CHECK: [[META8]] = !{!"Simple C++ TBAA"}
+; CHECK: [[CHAR_TBAA9]] = !{[[META7]], [[META7]], i64 0}
+; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META11:![0-9]+]]}
+; CHECK: [[META11]] = !{!"llvm.loop.mustprogress"}
;.
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/earlycse-after-simplifycfg-two-entry-phi-node-folding.ll b/llvm/test/Transforms/PhaseOrdering/X86/earlycse-after-simplifycfg-two-entry-phi-node-folding.ll
index 4c57ba74782e5..c6316605f7c03 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/earlycse-after-simplifycfg-two-entry-phi-node-folding.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/earlycse-after-simplifycfg-two-entry-phi-node-folding.ll
@@ -11,7 +11,8 @@ define dso_local void @foo(ptr %in, i64 %lo, i64 %hi, i32 %ishi) #0 {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TOBOOL_NOT:%.*]] = icmp eq i32 [[ISHI:%.*]], 0
; CHECK-NEXT: [[LO_HI:%.*]] = select i1 [[TOBOOL_NOT]], i64 [[LO:%.*]], i64 [[HI:%.*]]
-; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[IN:%.*]], i64 [[LO_HI]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl nsw i64 [[LO_HI]], 2
+; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[IN:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[ARRAYVAL2:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4
; CHECK-NEXT: [[INC2:%.*]] = add nsw i32 [[ARRAYVAL2]], 1
; CHECK-NEXT: store i32 [[INC2]], ptr [[ARRAYIDX1]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll b/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll
index ffdff21f94f0e..1be66ac92d59e 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/excessive-unrolling.ll
@@ -175,17 +175,18 @@ define void @test_runtime_trip_count(i32 %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 nuw double, ptr @b, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP11:%.*]] = shl nuw nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds nuw i8, ptr @b, i64 [[TMP11]]
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw 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 nuw double, ptr @c, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr @c, i64 [[TMP11]]
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds nuw 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 nuw double, ptr @a, i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr @a, i64 [[TMP11]]
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw 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
@@ -200,12 +201,13 @@ define void @test_runtime_trip_count(i32 %N) {
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ], [ [[INDVARS_IV_PH]], [[FOR_BODY_PREHEADER7]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr @b, i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr @b, i64 [[TMP12]]
; CHECK-NEXT: [[TMP9:%.*]] = load double, ptr [[ARRAYIDX]], align 8
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw double, ptr @c, i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr @c, i64 [[TMP12]]
; CHECK-NEXT: [[TMP10:%.*]] = load double, ptr [[ARRAYIDX2]], align 8
; CHECK-NEXT: [[ADD:%.*]] = fadd double [[TMP9]], [[TMP10]]
-; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw double, ptr @a, i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr @a, i64 [[TMP12]]
; CHECK-NEXT: store double [[ADD]], ptr [[ARRAYIDX4]], align 8
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
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 7954ff051a33d..c8c7f8fdc8f72 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/hoist-load-of-baseptr.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/hoist-load-of-baseptr.ll
@@ -31,7 +31,8 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O1-NEXT: br i1 [[EXITCOND7_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_COND1_PREHEADER]], !llvm.loop [[LOOP0:![0-9]+]]
; O1: [[FOR_BODY4]]:
; O1-NEXT: [[J_05:%.*]] = phi i64 [ [[INC5:%.*]], %[[FOR_BODY4]] ], [ 0, %[[FOR_COND1_PREHEADER]] ]
-; O1-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 [[J_05]]
+; O1-NEXT: [[TMP2:%.*]] = shl nsw i64 [[J_05]], 2
+; O1-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP2]]
; O1-NEXT: [[TMP1:%.*]] = load i32, ptr [[ADD_PTR_I]], align 4, !tbaa [[INT_TBAA2:![0-9]+]]
; O1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP1]], 1
; O1-NEXT: store i32 [[INC]], ptr [[ADD_PTR_I]], align 4, !tbaa [[INT_TBAA2]]
@@ -55,7 +56,8 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O2-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[FOR_BODY4_PREHEADER9:.*]], label %[[VECTOR_BODY:.*]]
; 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: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; O2-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP8]]
; O2-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
; O2-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4, !tbaa [[INT_TBAA0:![0-9]+]]
; O2-NEXT: [[WIDE_LOAD8:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !tbaa [[INT_TBAA0]]
@@ -79,7 +81,8 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; 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]]
+; O2-NEXT: [[TMP7:%.*]] = shl nsw i64 [[J_05]], 2
+; O2-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP7]]
; O2-NEXT: [[TMP6:%.*]] = load i32, ptr [[ADD_PTR_I]], align 4, !tbaa [[INT_TBAA0]]
; O2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP6]], 1
; O2-NEXT: store i32 [[INC]], ptr [[ADD_PTR_I]], align 4, !tbaa [[INT_TBAA0]]
@@ -103,7 +106,8 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O3-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[FOR_BODY4_US_PREHEADER:.*]], label %[[VECTOR_BODY:.*]]
; 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: [[TMP8:%.*]] = shl nsw i64 [[INDEX]], 2
+; O3-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP8]]
; O3-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
; O3-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP1]], align 4, !tbaa [[INT_TBAA0:![0-9]+]]
; O3-NEXT: [[WIDE_LOAD9:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !tbaa [[INT_TBAA0]]
@@ -121,7 +125,8 @@ define dso_local void @_Z7computeRSt6vectorIiSaIiEEy(ptr noundef nonnull align 8
; O3-NEXT: br label %[[FOR_BODY4_US:.*]]
; O3: [[FOR_BODY4_US]]:
; O3-NEXT: [[J_05_US:%.*]] = phi i64 [ [[INC5_US:%.*]], %[[FOR_BODY4_US]] ], [ [[J_05_US_PH]], %[[FOR_BODY4_US_PREHEADER]] ]
-; O3-NEXT: [[ADD_PTR_I_US:%.*]] = getelementptr inbounds i32, ptr [[TMP0]], i64 [[J_05_US]]
+; O3-NEXT: [[TMP7:%.*]] = shl nsw i64 [[J_05_US]], 2
+; O3-NEXT: [[ADD_PTR_I_US:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP7]]
; O3-NEXT: [[TMP6:%.*]] = load i32, ptr [[ADD_PTR_I_US]], align 4, !tbaa [[INT_TBAA0]]
; O3-NEXT: [[INC_US:%.*]] = add nsw i32 [[TMP6]], 1
; O3-NEXT: store i32 [[INC_US]], ptr [[ADD_PTR_I_US]], align 4, !tbaa [[INT_TBAA0]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/merge-functions2.ll b/llvm/test/Transforms/PhaseOrdering/X86/merge-functions2.ll
index b77abb7e8f203..834f77567afe8 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/merge-functions2.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/merge-functions2.ll
@@ -11,8 +11,9 @@ define i32 @f(i32 noundef %x) {
; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[X]], 8
; CHECK-NEXT: br i1 [[TMP0]], label %[[SWITCH_LOOKUP:.*]], label %[[SW_EPILOG:.*]]
; CHECK: [[SWITCH_LOOKUP]]:
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[X]] to i64
-; CHECK-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i32, ptr @switch.table.g, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[X]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CHECK-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i8, ptr @switch.table.g, i64 [[TMP2]]
; CHECK-NEXT: [[SWITCH_LOAD:%.*]] = load i32, ptr [[SWITCH_GEP]], align 4
; CHECK-NEXT: br label %[[SW_EPILOG]]
; CHECK: [[SW_EPILOG]]:
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/merge-functions3.ll b/llvm/test/Transforms/PhaseOrdering/X86/merge-functions3.ll
index 71700276398ff..7f1688567129f 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/merge-functions3.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/merge-functions3.ll
@@ -14,8 +14,9 @@ define range(i32 0, 2) i32 @f(i32 noundef %x) local_unnamed_addr {
; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[X]], 8
; CHECK-NEXT: br i1 [[TMP0]], label %[[SWITCH_LOOKUP:.*]], label %[[SW_EPILOG:.*]]
; CHECK: [[SWITCH_LOOKUP]]:
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[X]] to i64
-; CHECK-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i32, ptr @switch.table.g, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[X]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CHECK-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i8, ptr @switch.table.g, i64 [[TMP2]]
; CHECK-NEXT: [[SWITCH_LOAD:%.*]] = load i32, ptr [[SWITCH_GEP]], align 4
; CHECK-NEXT: br label %[[SW_EPILOG]]
; CHECK: [[SW_EPILOG]]:
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll b/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
index 9177f279217fa..13bd91aacfec8 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pixel-splat.ll
@@ -42,7 +42,8 @@ define void @loop_or(ptr noalias %pIn, ptr noalias %pOut, i32 %s) {
; CHECK-NEXT: [[TMP5:%.*]] = mul nuw nsw <4 x i32> [[TMP3]], splat (i32 65793)
; CHECK-NEXT: [[TMP6:%.*]] = or disjoint <4 x i32> [[TMP4]], splat (i32 -16777216)
; CHECK-NEXT: [[TMP7:%.*]] = or disjoint <4 x i32> [[TMP5]], splat (i32 -16777216)
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i32, ptr [[POUT:%.*]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP12:%.*]] = shl nuw nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[POUT:%.*]], i64 [[TMP12]]
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw 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
@@ -62,7 +63,8 @@ define void @loop_or(ptr noalias %pIn, ptr noalias %pOut, i32 %s) {
; CHECK-NEXT: [[CONV:%.*]] = zext i8 [[TMP11]] to i32
; CHECK-NEXT: [[OR2:%.*]] = mul nuw nsw i32 [[CONV]], 65793
; CHECK-NEXT: [[OR3:%.*]] = or disjoint i32 [[OR2]], -16777216
-; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i32, ptr [[POUT]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP13:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i8, ptr [[POUT]], i64 [[TMP13]]
; CHECK-NEXT: store i32 [[OR3]], ptr [[ARRAYIDX5]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll b/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll
index 482907d295706..e73e0dcfe3c6e 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pr88239.ll
@@ -11,13 +11,14 @@ define void @foo(ptr noalias noundef %0, ptr noalias noundef %1) optsize {
; CHECK-NEXT: br label [[TMP4:%.*]]
; CHECK: vector.body:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ 0, [[TMP2:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[TMP4]] ]
-; CHECK-NEXT: [[TMP3:%.*]] = sub nuw nsw i64 255, [[INDVARS_IV]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP0]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP8:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = sub nuw nsw i64 1020, [[TMP8]]
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP0]], i64 [[TMP3]]
; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i64 -28
; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = load <8 x i32>, ptr [[GEP]], align 4
; CHECK-NEXT: [[TMP5:%.*]] = add nsw <8 x i32> [[WIDE_MASKED_GATHER]], splat (i32 5)
; CHECK-NEXT: [[TMP6:%.*]] = shufflevector <8 x i32> [[TMP5]], <8 x i32> poison, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP1]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 [[TMP8]]
; CHECK-NEXT: store <8 x i32> [[TMP6]], ptr [[TMP10]], align 4
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw i64 [[INDVARS_IV]], 8
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], 256
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll b/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll
index d9803697a7265..9824405203148 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/preserve-access-group.ll
@@ -14,7 +14,6 @@ define void @test(i32 noundef %nface, i32 noundef %ncell, ptr noalias noundef %f
; CHECK-NEXT: br i1 [[CMP8]], label %[[FOR_BODY_PREHEADER:.*]], label %[[FOR_COND_CLEANUP:.*]]
; CHECK: [[FOR_BODY_PREHEADER]]:
; CHECK-NEXT: [[TMP0:%.*]] = zext nneg i32 [[NFACE]] to i64
-; CHECK-NEXT: [[INVARIANT_GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[FACE_CELL]], i64 [[TMP0]]
; CHECK-NEXT: [[TMP1:%.*]] = icmp ult i32 [[NFACE]], 4
; CHECK-NEXT: br i1 [[TMP1]], label %[[FOR_BODY_PREHEADER14:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
@@ -22,19 +21,24 @@ define void @test(i32 noundef %nface, i32 noundef %ncell, ptr noalias noundef %f
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDVARS_IV_EPIL:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw i32, ptr [[FACE_CELL]], i64 [[INDVARS_IV_EPIL]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP10]], align 4, !tbaa [[INT_TBAA0:![0-9]+]], !llvm.access.group [[ACC_GRP4:![0-9]+]]
-; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i32, ptr [[INVARIANT_GEP]], i64 [[INDVARS_IV_EPIL]]
-; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !tbaa [[INT_TBAA0]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: [[TMP6:%.*]] = shl nuw nsw i64 [[INDVARS_IV_EPIL]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[FACE_CELL]], i64 [[TMP6]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4, !tbaa [[INT_TBAA0:![0-9]+]], !llvm.access.group [[ACC_GRP4:![0-9]+]]
+; CHECK-NEXT: [[TMP12:%.*]] = add nuw nsw i64 [[INDVARS_IV_EPIL]], [[TMP0]]
+; CHECK-NEXT: [[TMP4:%.*]] = shl nuw nsw i64 [[TMP12]], 2
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[FACE_CELL]], i64 [[TMP4]]
+; CHECK-NEXT: [[WIDE_LOAD12:%.*]] = load <4 x i32>, ptr [[TMP13]], align 4, !tbaa [[INT_TBAA0]], !llvm.access.group [[ACC_GRP4]]
; CHECK-NEXT: [[TMP3:%.*]] = sext <4 x i32> [[WIDE_LOAD]] to <4 x i64>
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds double, ptr [[Y]], <4 x i64> [[TMP3]]
+; CHECK-NEXT: [[TMP14:%.*]] = shl nsw <4 x i64> [[TMP3]], splat (i64 3)
+; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[Y]], <4 x i64> [[TMP14]]
; CHECK-NEXT: [[TMP5:%.*]] = sext <4 x i32> [[WIDE_LOAD12]] to <4 x i64>
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds double, ptr [[X]], <4 x i64> [[TMP5]]
-; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = tail call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 8 [[TMP4]], <4 x i1> splat (i1 true), <4 x double> poison), !tbaa [[DOUBLE_TBAA5:![0-9]+]], !llvm.access.group [[ACC_GRP4]]
-; CHECK-NEXT: [[WIDE_MASKED_GATHER13:%.*]] = tail call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 8 [[TMP6]], <4 x i1> splat (i1 true), <4 x double> poison), !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: [[TMP10:%.*]] = shl nsw <4 x i64> [[TMP5]], splat (i64 3)
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[X]], <4 x i64> [[TMP10]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = tail call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 8 [[TMP16]], <4 x i1> splat (i1 true), <4 x double> poison), !tbaa [[DOUBLE_TBAA5:![0-9]+]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER13:%.*]] = tail call <4 x double> @llvm.masked.gather.v4f64.v4p0(<4 x ptr> align 8 [[TMP11]], <4 x i1> splat (i1 true), <4 x double> poison), !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
; CHECK-NEXT: [[TMP7:%.*]] = fcmp fast olt <4 x double> [[WIDE_MASKED_GATHER]], [[WIDE_MASKED_GATHER13]]
; CHECK-NEXT: [[TMP8:%.*]] = select <4 x i1> [[TMP7]], <4 x double> [[WIDE_MASKED_GATHER13]], <4 x double> [[WIDE_MASKED_GATHER]]
-; CHECK-NEXT: tail call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> [[TMP8]], <4 x ptr> align 8 [[TMP4]], <4 x i1> splat (i1 true)), !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: tail call void @llvm.masked.scatter.v4f64.v4p0(<4 x double> [[TMP8]], <4 x ptr> align 8 [[TMP16]], <4 x i1> splat (i1 true)), !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDVARS_IV_EPIL]], 4
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[UNROLL_ITER]]
; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
@@ -48,19 +52,24 @@ define void @test(i32 noundef %nface, i32 noundef %ncell, ptr noalias noundef %f
; CHECK-NEXT: ret void
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[INDVARS_IV_NEXT_2:%.*]] = phi i64 [ [[INDVARS_IV_NEXT:%.*]], %[[FOR_BODY]] ], [ [[INDVARS_IV_PH]], %[[FOR_BODY_PREHEADER14]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[FACE_CELL]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-NEXT: [[TMP15:%.*]] = shl nuw nsw i64 [[INDVARS_IV_NEXT_2]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[FACE_CELL]], i64 [[TMP15]]
; CHECK-NEXT: [[TMP22:%.*]] = load i32, ptr [[ARRAYIDX]], align 4, !tbaa [[INT_TBAA0]], !llvm.access.group [[ACC_GRP4]]
-; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds nuw i32, ptr [[INVARIANT_GEP]], i64 [[INDVARS_IV_NEXT_2]]
-; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[GEP]], align 4, !tbaa [[INT_TBAA0]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: [[TMP17:%.*]] = add nuw nsw i64 [[INDVARS_IV_NEXT_2]], [[TMP0]]
+; CHECK-NEXT: [[TMP18:%.*]] = shl nuw nsw i64 [[TMP17]], 2
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[FACE_CELL]], i64 [[TMP18]]
+; CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4, !tbaa [[INT_TBAA0]], !llvm.access.group [[ACC_GRP4]]
; CHECK-NEXT: [[IDXPROM3_3:%.*]] = sext i32 [[TMP22]] to i64
-; CHECK-NEXT: [[ARRAYIDX4_3:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[IDXPROM3_3]]
+; CHECK-NEXT: [[TMP20:%.*]] = shl nsw i64 [[IDXPROM3_3]], 3
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, ptr [[Y]], i64 [[TMP20]]
; CHECK-NEXT: [[IDXPROM5_3:%.*]] = sext i32 [[TMP23]] to i64
-; CHECK-NEXT: [[ARRAYIDX6_3:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[IDXPROM5_3]]
-; CHECK-NEXT: [[TMP24:%.*]] = load double, ptr [[ARRAYIDX4_3]], align 8, !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
-; CHECK-NEXT: [[TMP25:%.*]] = load double, ptr [[ARRAYIDX6_3]], align 8, !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: [[TMP21:%.*]] = shl nsw i64 [[IDXPROM5_3]], 3
+; CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i8, ptr [[X]], i64 [[TMP21]]
+; CHECK-NEXT: [[TMP24:%.*]] = load double, ptr [[ARRAYIDX4]], align 8, !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: [[TMP25:%.*]] = load double, ptr [[ARRAYIDX6]], align 8, !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
; CHECK-NEXT: [[CMP_I_3:%.*]] = fcmp fast olt double [[TMP24]], [[TMP25]]
; CHECK-NEXT: [[TMP26:%.*]] = select i1 [[CMP_I_3]], double [[TMP25]], double [[TMP24]]
-; CHECK-NEXT: store double [[TMP26]], ptr [[ARRAYIDX4_3]], align 8, !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
+; CHECK-NEXT: store double [[TMP26]], ptr [[ARRAYIDX4]], align 8, !tbaa [[DOUBLE_TBAA5]], !llvm.access.group [[ACC_GRP4]]
; CHECK-NEXT: [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV_NEXT_2]], 1
; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[TMP0]]
; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label %[[FOR_COND_CLEANUP]], label %[[FOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/simplifycfg-late.ll b/llvm/test/Transforms/PhaseOrdering/X86/simplifycfg-late.ll
index b8b166bb0e0e3..8e1abecf40c4d 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/simplifycfg-late.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/simplifycfg-late.ll
@@ -13,8 +13,9 @@ define i32 @f(i32 %c) {
; CHECK-NEXT: [[TMP0:%.*]] = icmp ult i32 [[SWITCH_TABLEIDX]], 7
; CHECK-NEXT: br i1 [[TMP0]], label [[SWITCH_LOOKUP:%.*]], label [[RETURN:%.*]]
; CHECK: switch.lookup:
-; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[SWITCH_TABLEIDX]] to i64
-; CHECK-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i32, ptr @switch.table.f, i64 [[TMP1]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[SWITCH_TABLEIDX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = zext nneg i32 [[TMP1]] to i64
+; CHECK-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i8, ptr @switch.table.f, i64 [[TMP2]]
; CHECK-NEXT: [[SWITCH_LOAD:%.*]] = load i32, ptr [[SWITCH_GEP]], align 4
; CHECK-NEXT: br label [[RETURN]]
; CHECK: return:
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll b/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll
index fcdb68353311d..d8a1ab461625b 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/speculation-vs-tbaa.ll
@@ -19,7 +19,8 @@ define void @licm(ptr align 8 dereferenceable(8) %_M_start.i, i64 %numElem) {
; O1-NEXT: br label %[[FOR_BODY:.*]]
; O1: [[FOR_BODY]]:
; O1-NEXT: [[K_02:%.*]] = phi i64 [ 0, %[[FOR_BODY_LR_PH]] ], [ [[INC:%.*]], %[[FOR_BODY]] ]
-; O1-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds double, ptr [[TMP0]], i64 [[K_02]]
+; O1-NEXT: [[TMP1:%.*]] = shl nsw i64 [[K_02]], 3
+; O1-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP1]]
; O1-NEXT: store double 2.000000e+00, ptr [[ADD_PTR_I]], align 8, !tbaa [[DOUBLE_TBAA8:![0-9]+]]
; O1-NEXT: [[INC]] = add nuw i64 [[K_02]], 1
; O1-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[NUMELEM]]
@@ -41,7 +42,8 @@ define void @licm(ptr align 8 dereferenceable(8) %_M_start.i, i64 %numElem) {
; O23-NEXT: br label %[[VECTOR_BODY:.*]]
; 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: [[TMP4:%.*]] = shl nsw i64 [[INDEX]], 3
+; O23-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP4]]
; O23-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 16
; O23-NEXT: store <2 x double> splat (double 2.000000e+00), ptr [[TMP1]], align 8, !tbaa [[DOUBLE_TBAA8:![0-9]+]]
; O23-NEXT: store <2 x double> splat (double 2.000000e+00), ptr [[TMP2]], align 8, !tbaa [[DOUBLE_TBAA8]]
@@ -56,7 +58,8 @@ define void @licm(ptr align 8 dereferenceable(8) %_M_start.i, i64 %numElem) {
; O23-NEXT: br label %[[FOR_BODY:.*]]
; O23: [[FOR_BODY]]:
; O23-NEXT: [[K_02:%.*]] = phi i64 [ [[INC:%.*]], %[[FOR_BODY]] ], [ [[K_02_PH]], %[[FOR_BODY_PREHEADER]] ]
-; O23-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds double, ptr [[TMP0]], i64 [[K_02]]
+; O23-NEXT: [[TMP5:%.*]] = shl nsw i64 [[K_02]], 3
+; O23-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds i8, ptr [[TMP0]], i64 [[TMP5]]
; O23-NEXT: store double 2.000000e+00, ptr [[ADD_PTR_I]], align 8, !tbaa [[DOUBLE_TBAA8]]
; O23-NEXT: [[INC]] = add nuw i64 [[K_02]], 1
; O23-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[NUMELEM]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll b/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll
index 574132c18d263..e35fb10863f2d 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/spurious-peeling.ll
@@ -17,26 +17,22 @@ define dso_local void @_Z13vecIncFromPtrP12FloatVecPair(ptr %FVP) {
; O1-SAME: ptr readonly captures(none) [[FVP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; O1-NEXT: [[ENTRY:.*:]]
; O1-NEXT: [[VSRC23_I:%.*]] = getelementptr inbounds nuw i8, ptr [[FVP]], i64 16
-; O1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[VSRC23_I]], align 8, !tbaa [[ANYPTR_TBAA0:![0-9]+]]
-; O1-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0:%.*]], ptr [[TMP0]], i64 undef
+; O1-NEXT: [[ARRAYIDX_I_I:%.*]] = load ptr, ptr [[VSRC23_I]], align 8, !tbaa [[ANYPTR_TBAA0:![0-9]+]]
; O1-NEXT: [[SIZE4_I:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX_I_I]], i64 8
; O1-NEXT: [[TMP1:%.*]] = load i32, ptr [[SIZE4_I]], align 8, !tbaa [[INT_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: [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX_I_I]], align 8, !tbaa [[ANYPTR_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 [[ANYPTR_TBAA0]]
-; O1-NEXT: [[ARRAYIDX_I4_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0]], ptr [[TMP3]], i64 undef
-; O1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[ARRAYIDX_I4_I]], align 8, !tbaa [[ANYPTR_TBAA8]]
-; O1-NEXT: [[ARRAYIDX_I5_I:%.*]] = getelementptr inbounds float, ptr [[TMP4]], i64 undef
+; O1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8, !tbaa [[ANYPTR_TBAA8]]
; O1-NEXT: br label %[[FOR_BODY7_I:.*]]
; O1: [[FOR_BODY7_I]]:
; O1-NEXT: [[J_07_I:%.*]] = phi i32 [ 0, %[[FOR_BODY7_LR_PH_I]] ], [ [[INC_I:%.*]], %[[FOR_BODY7_I]] ]
-; O1-NEXT: [[TMP5:%.*]] = load float, ptr [[ARRAYIDX_I3_I]], align 4, !tbaa [[FLOAT_TBAA9:![0-9]+]]
-; O1-NEXT: [[TMP6:%.*]] = load float, ptr [[ARRAYIDX_I5_I]], align 4, !tbaa [[FLOAT_TBAA9]]
+; O1-NEXT: [[TMP5:%.*]] = load float, ptr [[TMP2]], align 4, !tbaa [[FLOAT_TBAA9:![0-9]+]]
+; O1-NEXT: [[TMP6:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[FLOAT_TBAA9]]
; O1-NEXT: [[ADD_I:%.*]] = fadd float [[TMP5]], [[TMP6]]
-; O1-NEXT: store float [[ADD_I]], ptr [[ARRAYIDX_I5_I]], align 4, !tbaa [[FLOAT_TBAA9]]
+; O1-NEXT: store float [[ADD_I]], ptr [[TMP4]], align 4, !tbaa [[FLOAT_TBAA9]]
; O1-NEXT: [[INC_I]] = add nuw i32 [[J_07_I]], 1
; O1-NEXT: [[EXITCOND_NOT_I:%.*]] = icmp eq i32 [[INC_I]], [[TMP1]]
; O1-NEXT: br i1 [[EXITCOND_NOT_I]], label %[[_ZN12FLOATVECPAIR6VECINCEV_EXIT]], label %[[FOR_BODY7_I]], !llvm.loop [[LOOP11:![0-9]+]]
@@ -47,27 +43,23 @@ define dso_local void @_Z13vecIncFromPtrP12FloatVecPair(ptr %FVP) {
; O23-SAME: ptr readonly captures(none) [[FVP:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; O23-NEXT: [[ENTRY:.*:]]
; O23-NEXT: [[VSRC23_I:%.*]] = getelementptr inbounds nuw i8, ptr [[FVP]], i64 16
-; O23-NEXT: [[TMP0:%.*]] = load ptr, ptr [[VSRC23_I]], align 8, !tbaa [[ANYPTR_TBAA0:![0-9]+]]
-; O23-NEXT: [[ARRAYIDX_I_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0:%.*]], ptr [[TMP0]], i64 undef
+; O23-NEXT: [[ARRAYIDX_I_I:%.*]] = load ptr, ptr [[VSRC23_I]], align 8, !tbaa [[ANYPTR_TBAA0:![0-9]+]]
; O23-NEXT: [[SIZE4_I:%.*]] = getelementptr inbounds nuw i8, ptr [[ARRAYIDX_I_I]], i64 8
; O23-NEXT: [[TMP1:%.*]] = load i32, ptr [[SIZE4_I]], align 8, !tbaa [[INT_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: [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX_I_I]], align 8, !tbaa [[ANYPTR_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 [[ANYPTR_TBAA0]]
-; O23-NEXT: [[ARRAYIDX_I4_I:%.*]] = getelementptr inbounds [[CLASS_HOMEMADEVECTOR_0]], ptr [[TMP3]], i64 undef
-; O23-NEXT: [[TMP4:%.*]] = load ptr, ptr [[ARRAYIDX_I4_I]], align 8, !tbaa [[ANYPTR_TBAA8]]
-; O23-NEXT: [[ARRAYIDX_I5_I:%.*]] = getelementptr inbounds float, ptr [[TMP4]], i64 undef
-; O23-NEXT: [[DOTPRE_I:%.*]] = load float, ptr [[ARRAYIDX_I5_I]], align 4, !tbaa [[FLOAT_TBAA9:![0-9]+]]
+; O23-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8, !tbaa [[ANYPTR_TBAA8]]
+; O23-NEXT: [[DOTPRE_I:%.*]] = load float, ptr [[TMP4]], align 4, !tbaa [[FLOAT_TBAA9:![0-9]+]]
; O23-NEXT: br label %[[FOR_BODY7_I:.*]]
; O23: [[FOR_BODY7_I]]:
; O23-NEXT: [[TMP5:%.*]] = phi float [ [[DOTPRE_I]], %[[FOR_BODY7_LR_PH_I]] ], [ [[ADD_I:%.*]], %[[FOR_BODY7_I]] ]
; O23-NEXT: [[J_07_I:%.*]] = phi i32 [ 0, %[[FOR_BODY7_LR_PH_I]] ], [ [[INC_I:%.*]], %[[FOR_BODY7_I]] ]
-; O23-NEXT: [[TMP6:%.*]] = load float, ptr [[ARRAYIDX_I3_I]], align 4, !tbaa [[FLOAT_TBAA9]]
+; O23-NEXT: [[TMP6:%.*]] = load float, ptr [[TMP2]], align 4, !tbaa [[FLOAT_TBAA9]]
; O23-NEXT: [[ADD_I]] = fadd float [[TMP5]], [[TMP6]]
-; O23-NEXT: store float [[ADD_I]], ptr [[ARRAYIDX_I5_I]], align 4, !tbaa [[FLOAT_TBAA9]]
+; O23-NEXT: store float [[ADD_I]], ptr [[TMP4]], align 4, !tbaa [[FLOAT_TBAA9]]
; O23-NEXT: [[INC_I]] = add nuw i32 [[J_07_I]], 1
; O23-NEXT: [[EXITCOND_NOT_I:%.*]] = icmp eq i32 [[INC_I]], [[TMP1]]
; O23-NEXT: br i1 [[EXITCOND_NOT_I]], label %[[_ZN12FLOATVECPAIR6VECINCEV_EXIT]], label %[[FOR_BODY7_I]], !llvm.loop [[LOOP11:![0-9]+]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll b/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll
index 57637d6077f28..dd61170e5826c 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/vdiv-nounroll.ll
@@ -24,10 +24,11 @@ define void @vdiv(ptr %a, float %b) {
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds nuw float, ptr [[A]], i64 [[INDEX]]
-; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP1]], align 4, !tbaa [[FLOAT_TBAA3:![0-9]+]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP1]]
+; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, ptr [[TMP2]], align 4, !tbaa [[FLOAT_TBAA3:![0-9]+]]
; CHECK-NEXT: [[TMP3:%.*]] = fmul fast <4 x float> [[WIDE_LOAD]], [[TMP0]]
-; CHECK-NEXT: store <4 x float> [[TMP3]], ptr [[TMP1]], align 4, !tbaa [[FLOAT_TBAA3]]
+; CHECK-NEXT: store <4 x float> [[TMP3]], ptr [[TMP2]], align 4, !tbaa [[FLOAT_TBAA3]]
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT: [[TMP5:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
; CHECK-NEXT: br i1 [[TMP5]], label %[[FOR_COND_CLEANUP:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll b/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
index 66a4d3f3dc7dd..e13034ab39cca 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/vdiv.ll
@@ -41,7 +41,8 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH1]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP44:%.*]] = shl nuw nsw i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP44]]
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP5]], i64 32
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP5]], i64 64
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP5]], i64 96
@@ -53,7 +54,7 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: [[TMP10:%.*]] = fmul fast <4 x double> [[WIDE_LOAD6]], [[TMP2]]
; 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 nuw double, ptr [[X]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP44]]
; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP13]], i64 32
; CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP13]], i64 64
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP13]], i64 96
@@ -69,7 +70,7 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_END]], label %[[VEC_EPILOG_ITER_CHECK:.*]]
; CHECK: [[VEC_EPILOG_ITER_CHECK]]:
; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp eq i64 [[N_VEC_REMAINING]], 0
-; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[FOR_BODY_PREHEADER9]], label %[[VEC_EPILOG_PH]]
+; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label %[[FOR_BODY_PREHEADER9]], label %[[VEC_EPILOG_PH]], !prof [[PROF10:![0-9]+]]
; CHECK: [[VEC_EPILOG_PH]]:
; CHECK-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[VEC_EPILOG_ITER_CHECK]] ], [ 0, %[[VECTOR_PH]] ]
; CHECK-NEXT: [[N_VEC11:%.*]] = and i64 [[WIDE_TRIP_COUNT]], 2147483644
@@ -79,14 +80,15 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: br label %[[VEC_EPILOG_VECTOR_BODY:.*]]
; CHECK: [[VEC_EPILOG_VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX12:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], %[[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT16:%.*]], %[[VEC_EPILOG_VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDEX12]]
-; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <4 x double>, ptr [[TMP39]], align 8, !tbaa [[DOUBLE_TBAA3]]
+; CHECK-NEXT: [[TMP46:%.*]] = shl nuw nsw i64 [[INDEX12]], 3
+; CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP46]]
+; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <4 x double>, ptr [[TMP48]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP40:%.*]] = fmul fast <4 x double> [[WIDE_LOAD13]], [[TMP38]]
-; CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDEX12]]
-; CHECK-NEXT: store <4 x double> [[TMP40]], ptr [[TMP41]], align 8, !tbaa [[DOUBLE_TBAA3]]
+; CHECK-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP46]]
+; CHECK-NEXT: store <4 x double> [[TMP40]], ptr [[TMP50]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[INDEX_NEXT16]] = add nuw i64 [[INDEX12]], 4
; CHECK-NEXT: [[TMP42:%.*]] = icmp eq i64 [[INDEX_NEXT16]], [[N_VEC11]]
-; CHECK-NEXT: br i1 [[TMP42]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; CHECK-NEXT: br i1 [[TMP42]], label %[[VEC_EPILOG_MIDDLE_BLOCK:.*]], label %[[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
; CHECK: [[VEC_EPILOG_MIDDLE_BLOCK]]:
; CHECK-NEXT: [[CMP_N17:%.*]] = icmp eq i64 [[N_VEC11]], [[WIDE_TRIP_COUNT]]
; CHECK-NEXT: br i1 [[CMP_N17]], label %[[FOR_END]], label %[[FOR_BODY_PREHEADER9]]
@@ -102,15 +104,16 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK: [[FOR_BODY_PROL]]:
; CHECK-NEXT: [[INDVARS_IV_PROL:%.*]] = phi i64 [ [[INDVARS_IV_NEXT_PROL:%.*]], %[[FOR_BODY_PROL]] ], [ [[INDVARS_IV_PH]], %[[FOR_BODY_PROL_PREHEADER]] ]
; CHECK-NEXT: [[PROL_ITER:%.*]] = phi i64 [ [[PROL_ITER_NEXT:%.*]], %[[FOR_BODY_PROL]] ], [ 0, %[[FOR_BODY_PROL_PREHEADER]] ]
-; CHECK-NEXT: [[ARRAYIDX_PROL:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_PROL]]
+; CHECK-NEXT: [[TMP52:%.*]] = shl nuw nsw i64 [[INDVARS_IV_PROL]], 3
+; CHECK-NEXT: [[ARRAYIDX_PROL:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP52]]
; CHECK-NEXT: [[T0_PROL:%.*]] = load double, ptr [[ARRAYIDX_PROL]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP19:%.*]] = fmul fast double [[T0_PROL]], [[TMP18]]
-; CHECK-NEXT: [[ARRAYIDX2_PROL:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_PROL]]
+; CHECK-NEXT: [[ARRAYIDX2_PROL:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP52]]
; CHECK-NEXT: store double [[TMP19]], ptr [[ARRAYIDX2_PROL]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_PROL]] = add nuw nsw i64 [[INDVARS_IV_PROL]], 1
; CHECK-NEXT: [[PROL_ITER_NEXT]] = add i64 [[PROL_ITER]], 1
; CHECK-NEXT: [[PROL_ITER_CMP_NOT:%.*]] = icmp eq i64 [[PROL_ITER_NEXT]], [[XTRAITER]]
-; CHECK-NEXT: br i1 [[PROL_ITER_CMP_NOT]], label %[[FOR_BODY_PROL_LOOPEXIT]], label %[[FOR_BODY_PROL]], !llvm.loop [[LOOP11:![0-9]+]]
+; CHECK-NEXT: br i1 [[PROL_ITER_CMP_NOT]], label %[[FOR_BODY_PROL_LOOPEXIT]], label %[[FOR_BODY_PROL]], !llvm.loop [[LOOP12:![0-9]+]]
; CHECK: [[FOR_BODY_PROL_LOOPEXIT]]:
; CHECK-NEXT: [[INDVARS_IV_UNR:%.*]] = phi i64 [ [[INDVARS_IV_PH]], %[[FOR_BODY_PREHEADER9]] ], [ [[INDVARS_IV_NEXT_PROL]], %[[FOR_BODY_PROL]] ]
; CHECK-NEXT: [[TMP20:%.*]] = sub nsw i64 [[INDVARS_IV_PH]], [[WIDE_TRIP_COUNT]]
@@ -128,56 +131,64 @@ define void @vdiv(ptr %x, ptr %y, double %a, i32 %N) #0 {
; CHECK-NEXT: br label %[[FOR_BODY:.*]]
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[INDVARS_IV:%.*]] = phi i64 [ [[INDVARS_IV_UNR]], %[[FOR_BODY_PREHEADER9_NEW]] ], [ [[INDVARS_IV_NEXT_7:%.*]], %[[FOR_BODY]] ]
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[TMP39:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP39]]
; CHECK-NEXT: [[T0:%.*]] = load double, ptr [[ARRAYIDX]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP30:%.*]] = fmul fast double [[T0]], [[TMP22]]
-; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV]]
+; CHECK-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP39]]
; CHECK-NEXT: store double [[TMP30]], ptr [[ARRAYIDX2]], align 8, !tbaa [[DOUBLE_TBAA3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP41:%.*]] = add i64 [[INDVARS_IV_NEXT]], 8
+; CHECK-NEXT: [[ARRAYIDX_1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP41]]
; CHECK-NEXT: [[T0_1:%.*]] = load double, ptr [[ARRAYIDX_1]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP31:%.*]] = fmul fast double [[T0_1]], [[TMP23]]
-; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_NEXT]]
+; CHECK-NEXT: [[ARRAYIDX2_1:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP41]]
; CHECK-NEXT: store double [[TMP31]], ptr [[ARRAYIDX2_1]], align 8, !tbaa [[DOUBLE_TBAA3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_1:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 2
-; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_NEXT_1]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_1:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP54:%.*]] = add i64 [[INDVARS_IV_NEXT_1]], 16
+; CHECK-NEXT: [[ARRAYIDX_2:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP54]]
; CHECK-NEXT: [[T0_2:%.*]] = load double, ptr [[ARRAYIDX_2]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP32:%.*]] = fmul fast double [[T0_2]], [[TMP24]]
-; CHECK-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_NEXT_1]]
+; CHECK-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP54]]
; CHECK-NEXT: store double [[TMP32]], ptr [[ARRAYIDX2_2]], align 8, !tbaa [[DOUBLE_TBAA3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_2:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 3
-; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_2:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP45:%.*]] = add i64 [[INDVARS_IV_NEXT_2]], 24
+; CHECK-NEXT: [[ARRAYIDX_3:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP45]]
; CHECK-NEXT: [[T0_3:%.*]] = load double, ptr [[ARRAYIDX_3]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP33:%.*]] = fmul fast double [[T0_3]], [[TMP25]]
-; CHECK-NEXT: [[ARRAYIDX2_3:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_NEXT_2]]
+; CHECK-NEXT: [[ARRAYIDX2_3:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP45]]
; CHECK-NEXT: store double [[TMP33]], ptr [[ARRAYIDX2_3]], align 8, !tbaa [[DOUBLE_TBAA3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_3:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 4
-; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_NEXT_3]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_3:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP47:%.*]] = add i64 [[INDVARS_IV_NEXT_3]], 32
+; CHECK-NEXT: [[ARRAYIDX_4:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP47]]
; CHECK-NEXT: [[T0_4:%.*]] = load double, ptr [[ARRAYIDX_4]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP34:%.*]] = fmul fast double [[T0_4]], [[TMP26]]
-; CHECK-NEXT: [[ARRAYIDX2_4:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_NEXT_3]]
+; CHECK-NEXT: [[ARRAYIDX2_4:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP47]]
; CHECK-NEXT: store double [[TMP34]], ptr [[ARRAYIDX2_4]], align 8, !tbaa [[DOUBLE_TBAA3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_4:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 5
-; CHECK-NEXT: [[ARRAYIDX_5:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_NEXT_4]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_4:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP49:%.*]] = add i64 [[INDVARS_IV_NEXT_4]], 40
+; CHECK-NEXT: [[ARRAYIDX_5:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP49]]
; CHECK-NEXT: [[T0_5:%.*]] = load double, ptr [[ARRAYIDX_5]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP35:%.*]] = fmul fast double [[T0_5]], [[TMP27]]
-; CHECK-NEXT: [[ARRAYIDX2_5:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_NEXT_4]]
+; CHECK-NEXT: [[ARRAYIDX2_5:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP49]]
; CHECK-NEXT: store double [[TMP35]], ptr [[ARRAYIDX2_5]], align 8, !tbaa [[DOUBLE_TBAA3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_5:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 6
-; CHECK-NEXT: [[ARRAYIDX_6:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_NEXT_5]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_5:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP51:%.*]] = add i64 [[INDVARS_IV_NEXT_5]], 48
+; CHECK-NEXT: [[ARRAYIDX_6:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP51]]
; CHECK-NEXT: [[T0_6:%.*]] = load double, ptr [[ARRAYIDX_6]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP36:%.*]] = fmul fast double [[T0_6]], [[TMP28]]
-; CHECK-NEXT: [[ARRAYIDX2_6:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_NEXT_5]]
+; CHECK-NEXT: [[ARRAYIDX2_6:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP51]]
; CHECK-NEXT: store double [[TMP36]], ptr [[ARRAYIDX2_6]], align 8, !tbaa [[DOUBLE_TBAA3]]
-; CHECK-NEXT: [[INDVARS_IV_NEXT_6:%.*]] = add nuw nsw i64 [[INDVARS_IV]], 7
-; CHECK-NEXT: [[ARRAYIDX_7:%.*]] = getelementptr inbounds nuw double, ptr [[Y]], i64 [[INDVARS_IV_NEXT_6]]
+; CHECK-NEXT: [[INDVARS_IV_NEXT_6:%.*]] = shl i64 [[INDVARS_IV]], 3
+; CHECK-NEXT: [[TMP53:%.*]] = add i64 [[INDVARS_IV_NEXT_6]], 56
+; CHECK-NEXT: [[ARRAYIDX_7:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i64 [[TMP53]]
; CHECK-NEXT: [[T0_7:%.*]] = load double, ptr [[ARRAYIDX_7]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[TMP37:%.*]] = fmul fast double [[T0_7]], [[TMP29]]
-; CHECK-NEXT: [[ARRAYIDX2_7:%.*]] = getelementptr inbounds nuw double, ptr [[X]], i64 [[INDVARS_IV_NEXT_6]]
+; CHECK-NEXT: [[ARRAYIDX2_7:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i64 [[TMP53]]
; CHECK-NEXT: store double [[TMP37]], ptr [[ARRAYIDX2_7]], align 8, !tbaa [[DOUBLE_TBAA3]]
; CHECK-NEXT: [[INDVARS_IV_NEXT_7]] = add nuw nsw i64 [[INDVARS_IV]], 8
; CHECK-NEXT: [[EXITCOND_NOT_7:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT_7]], [[WIDE_TRIP_COUNT]]
-; CHECK-NEXT: br i1 [[EXITCOND_NOT_7]], label %[[FOR_END]], label %[[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
+; CHECK-NEXT: br i1 [[EXITCOND_NOT_7]], label %[[FOR_END]], label %[[FOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
; CHECK: [[FOR_END]]:
; CHECK-NEXT: ret void
;
@@ -232,8 +243,9 @@ attributes #0 = { nounwind ssp uwtable "correctly-rounded-divide-sqrt-fp-math"="
; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META8:![0-9]+]], [[META9:![0-9]+]]}
; CHECK: [[META8]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META9]] = !{!"llvm.loop.unroll.runtime.disable"}
-; CHECK: [[LOOP10]] = distinct !{[[LOOP10]], [[META8]], [[META9]]}
-; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META12:![0-9]+]]}
-; CHECK: [[META12]] = !{!"llvm.loop.unroll.disable"}
-; CHECK: [[LOOP13]] = distinct !{[[LOOP13]], [[META8]]}
+; CHECK: [[PROF10]] = !{!"branch_weights", i32 4, i32 12}
+; CHECK: [[LOOP11]] = distinct !{[[LOOP11]], [[META8]], [[META9]]}
+; CHECK: [[LOOP12]] = distinct !{[[LOOP12]], [[META13:![0-9]+]]}
+; CHECK: [[META13]] = !{!"llvm.loop.unroll.disable"}
+; CHECK: [[LOOP14]] = distinct !{[[LOOP14]], [[META8]]}
;.
diff --git a/llvm/test/Transforms/PhaseOrdering/gvn-replacement-vs-hoist.ll b/llvm/test/Transforms/PhaseOrdering/gvn-replacement-vs-hoist.ll
index c6d1cbd421436..0f97fab19e1e0 100644
--- a/llvm/test/Transforms/PhaseOrdering/gvn-replacement-vs-hoist.ll
+++ b/llvm/test/Transforms/PhaseOrdering/gvn-replacement-vs-hoist.ll
@@ -5,9 +5,9 @@ define void @test(ptr noundef %a, i32 noundef %beam) {
; CHECK-LABEL: define void @test
; CHECK-SAME: (ptr noundef writeonly captures(none) [[A:%.*]], i32 noundef [[BEAM:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[MUL:%.*]] = shl nuw nsw i32 [[BEAM]], 1
-; CHECK-NEXT: [[IDXPROM:%.*]] = zext nneg i32 [[MUL]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[MUL:%.*]] = shl nuw nsw i32 [[BEAM]], 3
+; CHECK-NEXT: [[TMP0:%.*]] = zext nneg i32 [[MUL]] to i64
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP0]]
; CHECK-NEXT: br label [[FOR_BODY:%.*]]
; CHECK: for.cond.cleanup:
; CHECK-NEXT: ret void
@@ -19,9 +19,9 @@ define void @test(ptr noundef %a, i32 noundef %beam) {
; CHECK-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: br label [[FOR_INC]]
; CHECK: if.else:
-; CHECK-NEXT: [[MUL2:%.*]] = shl nuw nsw i32 [[I_06]], 1
-; CHECK-NEXT: [[IDXPROM3:%.*]] = zext nneg i32 [[MUL2]] to i64
-; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i32, ptr [[A]], i64 [[IDXPROM3]]
+; CHECK-NEXT: [[MUL2:%.*]] = shl nuw nsw i32 [[I_06]], 3
+; CHECK-NEXT: [[TMP1:%.*]] = zext nneg i32 [[MUL2]] to i64
+; CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[TMP1]]
; CHECK-NEXT: store i32 1, ptr [[ARRAYIDX4]], align 4
; CHECK-NEXT: br label [[FOR_INC]]
; CHECK: for.inc:
diff --git a/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll b/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll
index 5253c42d9c6d2..ed83e449cc98b 100644
--- a/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll
+++ b/llvm/test/Transforms/PhaseOrdering/loop-access-checks.ll
@@ -37,7 +37,10 @@ define void @test_fill_with_foreach([2 x i64] %elems.coerce) {
; CHECK-NEXT: tail call void @error()
; CHECK-NEXT: br label %[[COMMON_RET]]
; CHECK: [[FOR_BODY]]:
-; CHECK-NEXT: [[__BEGIN1_SROA_0_03:%.*]] = phi ptr [ [[INCDEC_PTR_I:%.*]], %[[FOR_BODY]] ], [ [[TMP0]], %[[FOR_COND_PREHEADER_SPLIT]] ]
+; CHECK-NEXT: [[__BEGIN1_SROA_0_03:%.*]] = phi ptr [ [[INCDEC_PTR_I:%.*]], %[[FOR_LATCH:.*]] ], [ [[TMP0]], %[[FOR_COND_PREHEADER_SPLIT]] ]
+; CHECK-NEXT: [[CMP2_I_I:%.*]] = icmp ult ptr [[__BEGIN1_SROA_0_03]], [[ADD_PTR_I]]
+; CHECK-NEXT: br i1 [[CMP2_I_I]], label %[[FOR_LATCH]], label %[[ERROR]]
+; CHECK: [[FOR_LATCH]]:
; CHECK-NEXT: tail call void @use(ptr noundef nonnull align 4 dereferenceable(4) [[__BEGIN1_SROA_0_03]])
; CHECK-NEXT: [[INCDEC_PTR_I]] = getelementptr inbounds nuw i8, ptr [[__BEGIN1_SROA_0_03]], i64 4
; CHECK-NEXT: [[CMP_I_NOT:%.*]] = icmp eq ptr [[INCDEC_PTR_I]], [[ADD_PTR_I]]
@@ -147,7 +150,8 @@ define void @foo(ptr noundef nonnull align 8 dereferenceable(24) noalias %vec) #
; CHECK-NEXT: ret void
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[I_010:%.*]] = phi i64 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[ENTRY]] ]
-; CHECK-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds double, ptr [[TMP1]], i64 [[I_010]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[I_010]], 3
+; CHECK-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = load double, ptr [[ADD_PTR_I]], align 8
; CHECK-NEXT: [[ADD:%.*]] = fadd double [[TMP2]], 1.000000e+00
; CHECK-NEXT: store double [[ADD]], ptr [[ADD_PTR_I]], align 8
@@ -286,7 +290,8 @@ define void @loop_with_signed_induction(ptr noundef nonnull align 8 dereferencea
; CHECK-NEXT: ret void
; CHECK: [[FOR_BODY]]:
; CHECK-NEXT: [[I_010:%.*]] = phi i64 [ [[INC:%.*]], %[[FOR_BODY]] ], [ 0, %[[ENTRY]] ]
-; CHECK-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds nuw double, ptr [[TMP1]], i64 [[I_010]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nsw i64 [[I_010]], 3
+; CHECK-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP1]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP2:%.*]] = load double, ptr [[ADD_PTR_I]], align 8, !tbaa [[DOUBLE_TBAA6:![0-9]+]]
; CHECK-NEXT: [[ADD:%.*]] = fadd double [[TMP2]], 1.000000e+00
; CHECK-NEXT: store double [[ADD]], ptr [[ADD_PTR_I]], align 8, !tbaa [[DOUBLE_TBAA6]]
@@ -360,7 +365,8 @@ define void @monkey(ptr noundef %arr, i32 noundef %len) {
; CHECK: [[FOR_BODY4]]:
; CHECK-NEXT: [[K_07:%.*]] = phi i32 [ [[DEC:%.*]], %[[FOR_BODY4]] ], [ [[I_09]], %[[FOR_BODY4_PREHEADER]] ]
; CHECK-NEXT: [[IDX_EXT_I:%.*]] = zext i32 [[K_07]] to i64
-; CHECK-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds nuw i32, ptr [[ARR]], i64 [[IDX_EXT_I]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDX_EXT_I]], 2
+; CHECK-NEXT: [[ADD_PTR_I:%.*]] = getelementptr inbounds nuw i8, ptr [[ARR]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ADD_PTR_I]], align 4
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP0]], 1
; CHECK-NEXT: store i32 [[ADD]], ptr [[ADD_PTR_I]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/loop-vectorize-bfi.ll b/llvm/test/Transforms/PhaseOrdering/loop-vectorize-bfi.ll
index 6183a2679c3aa..264fe9158e61a 100644
--- a/llvm/test/Transforms/PhaseOrdering/loop-vectorize-bfi.ll
+++ b/llvm/test/Transforms/PhaseOrdering/loop-vectorize-bfi.ll
@@ -19,7 +19,8 @@ define void @f(i1 %x) !prof !0 {
; CHECK-NEXT: [[EVL_BASED_IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[INDEX_EVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[AVL:%.*]] = phi i64 [ 65, %[[ENTRY]] ], [ [[AVL_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP3:%.*]] = call i32 @llvm.experimental.get.vector.length.i64(i64 [[AVL]], i32 2, i1 true)
-; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i64, ptr null, i64 [[EVL_BASED_IV]]
+; CHECK-NEXT: [[TMP7:%.*]] = shl i64 [[EVL_BASED_IV]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr i8, ptr null, i64 [[TMP7]]
; CHECK-NEXT: call void @llvm.vp.store.nxv2i64.p0(<vscale x 2 x i64> poison, ptr align 8 [[TMP4]], <vscale x 2 x i1> [[TMP2]], i32 [[TMP3]])
; CHECK-NEXT: call void @llvm.vp.store.nxv2i64.p0(<vscale x 2 x i64> poison, ptr align 8 [[TMP4]], <vscale x 2 x i1> [[TMP2]], i32 [[TMP3]])
; CHECK-NEXT: [[TMP5:%.*]] = zext nneg i32 [[TMP3]] to i64
diff --git a/llvm/test/Transforms/PhaseOrdering/lto-argpromotion-ipsccp.ll b/llvm/test/Transforms/PhaseOrdering/lto-argpromotion-ipsccp.ll
index bc42b0b354ab7..5c95ea22ba4f8 100644
--- a/llvm/test/Transforms/PhaseOrdering/lto-argpromotion-ipsccp.ll
+++ b/llvm/test/Transforms/PhaseOrdering/lto-argpromotion-ipsccp.ll
@@ -28,7 +28,8 @@ define internal void @child(ptr %p, ptr %n, ptr %c) noinline {
; CHECK-NEXT: br i1 [[CMP_NOT]], label %[[FOR_END:.*]], label %[[FOR_INC]]
; CHECK: [[FOR_INC]]:
; CHECK-NEXT: [[IDXPROM:%.*]] = zext nneg i32 [[I_0]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[P]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP1:%.*]] = shl nuw nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[P]], i64 [[TMP1]]
; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[MUL:%.*]] = mul i32 [[TMP0]], 5
; CHECK-NEXT: store i32 [[MUL]], ptr [[ARRAYIDX]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/lto-licm.ll b/llvm/test/Transforms/PhaseOrdering/lto-licm.ll
index 8b51f9f465587..09e201481fbf9 100644
--- a/llvm/test/Transforms/PhaseOrdering/lto-licm.ll
+++ b/llvm/test/Transforms/PhaseOrdering/lto-licm.ll
@@ -12,7 +12,8 @@ define void @hoist_fdiv(ptr %a, float %b) {
; CHECK-NEXT: br i1 [[CMP_NOT]], label [[FOR_END:%.*]], label [[FOR_INC]]
; CHECK: for.inc:
; CHECK-NEXT: [[IDXPROM:%.*]] = zext nneg i32 [[I_0]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw float, ptr [[A:%.*]], i64 [[IDXPROM]]
+; CHECK-NEXT: [[TMP3:%.*]] = shl nuw nsw i64 [[IDXPROM]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[A:%.*]], i64 [[TMP3]]
; CHECK-NEXT: [[TMP1:%.*]] = load float, ptr [[ARRAYIDX]], align 4
; CHECK-NEXT: [[TMP2:%.*]] = fmul fast float [[TMP1]], [[TMP0]]
; CHECK-NEXT: store float [[TMP2]], ptr [[ARRAYIDX]], align 4
diff --git a/llvm/test/Transforms/PhaseOrdering/scev-custom-dl.ll b/llvm/test/Transforms/PhaseOrdering/scev-custom-dl.ll
index 625941a93335b..fba0ff503ed14 100644
--- a/llvm/test/Transforms/PhaseOrdering/scev-custom-dl.ll
+++ b/llvm/test/Transforms/PhaseOrdering/scev-custom-dl.ll
@@ -13,23 +13,263 @@ target datalayout = "e-m:m-p:40:64:64:32-i32:32-i16:16-i8:8-n32"
define void @test1(i32 %d, ptr %p) nounwind uwtable ssp {
; CHECK-LABEL: 'test1'
; CHECK-NEXT: Classifying expressions for: @test1
-; CHECK-NEXT: %div1 = lshr i32 %d, 2
-; CHECK-NEXT: --> (%d /u 4) U: [0,1073741824) S: [0,1073741824)
-; CHECK-NEXT: %i.03 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
-; CHECK-NEXT: --> {0,+,1}<nuw><nsw><%for.body> U: [0,64) S: [0,64) Exits: 63 LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %p.addr.02 = phi ptr [ %p, %entry ], [ %add.ptr1, %for.body ]
-; CHECK-NEXT: --> {%p,+,(8 * (%d /u 4))}<%for.body> U: full-set S: full-set Exits: ((504 * (%d /u 4)) + %p) LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %add.ptr = getelementptr inbounds nuw i32, ptr %p.addr.02, i32 %div1
-; CHECK-NEXT: --> {((4 * (%d /u 4))<nuw><nsw> + %p)<nuw>,+,(8 * (%d /u 4))}<%for.body> U: full-set S: full-set Exits: ((508 * (%d /u 4)) + %p) LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %add.ptr1 = getelementptr inbounds nuw i32, ptr %add.ptr, i32 %div1
-; CHECK-NEXT: --> {((8 * (%d /u 4)) + %p),+,(8 * (%d /u 4))}<%for.body> U: full-set S: full-set Exits: ((512 * (%d /u 4)) + %p) LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %inc = add nuw nsw i32 %i.03, 1
-; CHECK-NEXT: --> {1,+,1}<nuw><nsw><%for.body> U: [1,65) S: [1,65) Exits: 64 LoopDispositions: { %for.body: Computable }
+; CHECK-NEXT: %div1 = and i32 %d, -4
+; CHECK-NEXT: --> (4 * (%d /u 4))<nuw> U: [0,-3) S: [-2147483648,2147483645)
+; CHECK-NEXT: %add.ptr = getelementptr inbounds nuw i8, ptr %p, i32 %div1
+; CHECK-NEXT: --> ((4 * (%d /u 4))<nuw> + %p)<nuw> U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1 = getelementptr inbounds nuw i8, ptr %add.ptr, i32 %div1
+; CHECK-NEXT: --> ((8 * (%d /u 4)) + %p)<nuw> U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.1 = getelementptr inbounds nuw i8, ptr %add.ptr1, i32 %div1
+; CHECK-NEXT: --> ((12 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.1 = getelementptr inbounds nuw i8, ptr %add.ptr.1, i32 %div1
+; CHECK-NEXT: --> ((16 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.2 = getelementptr inbounds nuw i8, ptr %add.ptr1.1, i32 %div1
+; CHECK-NEXT: --> ((20 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.2 = getelementptr inbounds nuw i8, ptr %add.ptr.2, i32 %div1
+; CHECK-NEXT: --> ((24 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.3 = getelementptr inbounds nuw i8, ptr %add.ptr1.2, i32 %div1
+; CHECK-NEXT: --> ((28 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.3 = getelementptr inbounds nuw i8, ptr %add.ptr.3, i32 %div1
+; CHECK-NEXT: --> ((32 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.4 = getelementptr inbounds nuw i8, ptr %add.ptr1.3, i32 %div1
+; CHECK-NEXT: --> ((36 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.4 = getelementptr inbounds nuw i8, ptr %add.ptr.4, i32 %div1
+; CHECK-NEXT: --> ((40 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.5 = getelementptr inbounds nuw i8, ptr %add.ptr1.4, i32 %div1
+; CHECK-NEXT: --> ((44 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.5 = getelementptr inbounds nuw i8, ptr %add.ptr.5, i32 %div1
+; CHECK-NEXT: --> ((48 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.6 = getelementptr inbounds nuw i8, ptr %add.ptr1.5, i32 %div1
+; CHECK-NEXT: --> ((52 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.6 = getelementptr inbounds nuw i8, ptr %add.ptr.6, i32 %div1
+; CHECK-NEXT: --> ((56 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.7 = getelementptr inbounds nuw i8, ptr %add.ptr1.6, i32 %div1
+; CHECK-NEXT: --> ((60 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.7 = getelementptr inbounds nuw i8, ptr %add.ptr.7, i32 %div1
+; CHECK-NEXT: --> ((64 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.8 = getelementptr inbounds nuw i8, ptr %add.ptr1.7, i32 %div1
+; CHECK-NEXT: --> ((68 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.8 = getelementptr inbounds nuw i8, ptr %add.ptr.8, i32 %div1
+; CHECK-NEXT: --> ((72 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.9 = getelementptr inbounds nuw i8, ptr %add.ptr1.8, i32 %div1
+; CHECK-NEXT: --> ((76 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.9 = getelementptr inbounds nuw i8, ptr %add.ptr.9, i32 %div1
+; CHECK-NEXT: --> ((80 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.10 = getelementptr inbounds nuw i8, ptr %add.ptr1.9, i32 %div1
+; CHECK-NEXT: --> ((84 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.10 = getelementptr inbounds nuw i8, ptr %add.ptr.10, i32 %div1
+; CHECK-NEXT: --> ((88 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.11 = getelementptr inbounds nuw i8, ptr %add.ptr1.10, i32 %div1
+; CHECK-NEXT: --> ((92 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.11 = getelementptr inbounds nuw i8, ptr %add.ptr.11, i32 %div1
+; CHECK-NEXT: --> ((96 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.12 = getelementptr inbounds nuw i8, ptr %add.ptr1.11, i32 %div1
+; CHECK-NEXT: --> ((100 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.12 = getelementptr inbounds nuw i8, ptr %add.ptr.12, i32 %div1
+; CHECK-NEXT: --> ((104 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.13 = getelementptr inbounds nuw i8, ptr %add.ptr1.12, i32 %div1
+; CHECK-NEXT: --> ((108 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.13 = getelementptr inbounds nuw i8, ptr %add.ptr.13, i32 %div1
+; CHECK-NEXT: --> ((112 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.14 = getelementptr inbounds nuw i8, ptr %add.ptr1.13, i32 %div1
+; CHECK-NEXT: --> ((116 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.14 = getelementptr inbounds nuw i8, ptr %add.ptr.14, i32 %div1
+; CHECK-NEXT: --> ((120 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.15 = getelementptr inbounds nuw i8, ptr %add.ptr1.14, i32 %div1
+; CHECK-NEXT: --> ((124 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.15 = getelementptr inbounds nuw i8, ptr %add.ptr.15, i32 %div1
+; CHECK-NEXT: --> ((128 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.16 = getelementptr inbounds nuw i8, ptr %add.ptr1.15, i32 %div1
+; CHECK-NEXT: --> ((132 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.16 = getelementptr inbounds nuw i8, ptr %add.ptr.16, i32 %div1
+; CHECK-NEXT: --> ((136 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.17 = getelementptr inbounds nuw i8, ptr %add.ptr1.16, i32 %div1
+; CHECK-NEXT: --> ((140 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.17 = getelementptr inbounds nuw i8, ptr %add.ptr.17, i32 %div1
+; CHECK-NEXT: --> ((144 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.18 = getelementptr inbounds nuw i8, ptr %add.ptr1.17, i32 %div1
+; CHECK-NEXT: --> ((148 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.18 = getelementptr inbounds nuw i8, ptr %add.ptr.18, i32 %div1
+; CHECK-NEXT: --> ((152 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.19 = getelementptr inbounds nuw i8, ptr %add.ptr1.18, i32 %div1
+; CHECK-NEXT: --> ((156 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.19 = getelementptr inbounds nuw i8, ptr %add.ptr.19, i32 %div1
+; CHECK-NEXT: --> ((160 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.20 = getelementptr inbounds nuw i8, ptr %add.ptr1.19, i32 %div1
+; CHECK-NEXT: --> ((164 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.20 = getelementptr inbounds nuw i8, ptr %add.ptr.20, i32 %div1
+; CHECK-NEXT: --> ((168 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.21 = getelementptr inbounds nuw i8, ptr %add.ptr1.20, i32 %div1
+; CHECK-NEXT: --> ((172 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.21 = getelementptr inbounds nuw i8, ptr %add.ptr.21, i32 %div1
+; CHECK-NEXT: --> ((176 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.22 = getelementptr inbounds nuw i8, ptr %add.ptr1.21, i32 %div1
+; CHECK-NEXT: --> ((180 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.22 = getelementptr inbounds nuw i8, ptr %add.ptr.22, i32 %div1
+; CHECK-NEXT: --> ((184 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.23 = getelementptr inbounds nuw i8, ptr %add.ptr1.22, i32 %div1
+; CHECK-NEXT: --> ((188 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.23 = getelementptr inbounds nuw i8, ptr %add.ptr.23, i32 %div1
+; CHECK-NEXT: --> ((192 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.24 = getelementptr inbounds nuw i8, ptr %add.ptr1.23, i32 %div1
+; CHECK-NEXT: --> ((196 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.24 = getelementptr inbounds nuw i8, ptr %add.ptr.24, i32 %div1
+; CHECK-NEXT: --> ((200 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.25 = getelementptr inbounds nuw i8, ptr %add.ptr1.24, i32 %div1
+; CHECK-NEXT: --> ((204 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.25 = getelementptr inbounds nuw i8, ptr %add.ptr.25, i32 %div1
+; CHECK-NEXT: --> ((208 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.26 = getelementptr inbounds nuw i8, ptr %add.ptr1.25, i32 %div1
+; CHECK-NEXT: --> ((212 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.26 = getelementptr inbounds nuw i8, ptr %add.ptr.26, i32 %div1
+; CHECK-NEXT: --> ((216 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.27 = getelementptr inbounds nuw i8, ptr %add.ptr1.26, i32 %div1
+; CHECK-NEXT: --> ((220 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.27 = getelementptr inbounds nuw i8, ptr %add.ptr.27, i32 %div1
+; CHECK-NEXT: --> ((224 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.28 = getelementptr inbounds nuw i8, ptr %add.ptr1.27, i32 %div1
+; CHECK-NEXT: --> ((228 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.28 = getelementptr inbounds nuw i8, ptr %add.ptr.28, i32 %div1
+; CHECK-NEXT: --> ((232 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.29 = getelementptr inbounds nuw i8, ptr %add.ptr1.28, i32 %div1
+; CHECK-NEXT: --> ((236 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.29 = getelementptr inbounds nuw i8, ptr %add.ptr.29, i32 %div1
+; CHECK-NEXT: --> ((240 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.30 = getelementptr inbounds nuw i8, ptr %add.ptr1.29, i32 %div1
+; CHECK-NEXT: --> ((244 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.30 = getelementptr inbounds nuw i8, ptr %add.ptr.30, i32 %div1
+; CHECK-NEXT: --> ((248 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.31 = getelementptr inbounds nuw i8, ptr %add.ptr1.30, i32 %div1
+; CHECK-NEXT: --> ((252 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.31 = getelementptr inbounds nuw i8, ptr %add.ptr.31, i32 %div1
+; CHECK-NEXT: --> ((256 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.32 = getelementptr inbounds nuw i8, ptr %add.ptr1.31, i32 %div1
+; CHECK-NEXT: --> ((260 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.32 = getelementptr inbounds nuw i8, ptr %add.ptr.32, i32 %div1
+; CHECK-NEXT: --> ((264 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.33 = getelementptr inbounds nuw i8, ptr %add.ptr1.32, i32 %div1
+; CHECK-NEXT: --> ((268 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.33 = getelementptr inbounds nuw i8, ptr %add.ptr.33, i32 %div1
+; CHECK-NEXT: --> ((272 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.34 = getelementptr inbounds nuw i8, ptr %add.ptr1.33, i32 %div1
+; CHECK-NEXT: --> ((276 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.34 = getelementptr inbounds nuw i8, ptr %add.ptr.34, i32 %div1
+; CHECK-NEXT: --> ((280 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.35 = getelementptr inbounds nuw i8, ptr %add.ptr1.34, i32 %div1
+; CHECK-NEXT: --> ((284 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.35 = getelementptr inbounds nuw i8, ptr %add.ptr.35, i32 %div1
+; CHECK-NEXT: --> ((288 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.36 = getelementptr inbounds nuw i8, ptr %add.ptr1.35, i32 %div1
+; CHECK-NEXT: --> ((292 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.36 = getelementptr inbounds nuw i8, ptr %add.ptr.36, i32 %div1
+; CHECK-NEXT: --> ((296 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.37 = getelementptr inbounds nuw i8, ptr %add.ptr1.36, i32 %div1
+; CHECK-NEXT: --> ((300 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.37 = getelementptr inbounds nuw i8, ptr %add.ptr.37, i32 %div1
+; CHECK-NEXT: --> ((304 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.38 = getelementptr inbounds nuw i8, ptr %add.ptr1.37, i32 %div1
+; CHECK-NEXT: --> ((308 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.38 = getelementptr inbounds nuw i8, ptr %add.ptr.38, i32 %div1
+; CHECK-NEXT: --> ((312 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.39 = getelementptr inbounds nuw i8, ptr %add.ptr1.38, i32 %div1
+; CHECK-NEXT: --> ((316 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.39 = getelementptr inbounds nuw i8, ptr %add.ptr.39, i32 %div1
+; CHECK-NEXT: --> ((320 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.40 = getelementptr inbounds nuw i8, ptr %add.ptr1.39, i32 %div1
+; CHECK-NEXT: --> ((324 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.40 = getelementptr inbounds nuw i8, ptr %add.ptr.40, i32 %div1
+; CHECK-NEXT: --> ((328 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.41 = getelementptr inbounds nuw i8, ptr %add.ptr1.40, i32 %div1
+; CHECK-NEXT: --> ((332 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.41 = getelementptr inbounds nuw i8, ptr %add.ptr.41, i32 %div1
+; CHECK-NEXT: --> ((336 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.42 = getelementptr inbounds nuw i8, ptr %add.ptr1.41, i32 %div1
+; CHECK-NEXT: --> ((340 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.42 = getelementptr inbounds nuw i8, ptr %add.ptr.42, i32 %div1
+; CHECK-NEXT: --> ((344 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.43 = getelementptr inbounds nuw i8, ptr %add.ptr1.42, i32 %div1
+; CHECK-NEXT: --> ((348 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.43 = getelementptr inbounds nuw i8, ptr %add.ptr.43, i32 %div1
+; CHECK-NEXT: --> ((352 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.44 = getelementptr inbounds nuw i8, ptr %add.ptr1.43, i32 %div1
+; CHECK-NEXT: --> ((356 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.44 = getelementptr inbounds nuw i8, ptr %add.ptr.44, i32 %div1
+; CHECK-NEXT: --> ((360 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.45 = getelementptr inbounds nuw i8, ptr %add.ptr1.44, i32 %div1
+; CHECK-NEXT: --> ((364 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.45 = getelementptr inbounds nuw i8, ptr %add.ptr.45, i32 %div1
+; CHECK-NEXT: --> ((368 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.46 = getelementptr inbounds nuw i8, ptr %add.ptr1.45, i32 %div1
+; CHECK-NEXT: --> ((372 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.46 = getelementptr inbounds nuw i8, ptr %add.ptr.46, i32 %div1
+; CHECK-NEXT: --> ((376 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.47 = getelementptr inbounds nuw i8, ptr %add.ptr1.46, i32 %div1
+; CHECK-NEXT: --> ((380 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.47 = getelementptr inbounds nuw i8, ptr %add.ptr.47, i32 %div1
+; CHECK-NEXT: --> ((384 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.48 = getelementptr inbounds nuw i8, ptr %add.ptr1.47, i32 %div1
+; CHECK-NEXT: --> ((388 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.48 = getelementptr inbounds nuw i8, ptr %add.ptr.48, i32 %div1
+; CHECK-NEXT: --> ((392 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.49 = getelementptr inbounds nuw i8, ptr %add.ptr1.48, i32 %div1
+; CHECK-NEXT: --> ((396 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.49 = getelementptr inbounds nuw i8, ptr %add.ptr.49, i32 %div1
+; CHECK-NEXT: --> ((400 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.50 = getelementptr inbounds nuw i8, ptr %add.ptr1.49, i32 %div1
+; CHECK-NEXT: --> ((404 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.50 = getelementptr inbounds nuw i8, ptr %add.ptr.50, i32 %div1
+; CHECK-NEXT: --> ((408 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.51 = getelementptr inbounds nuw i8, ptr %add.ptr1.50, i32 %div1
+; CHECK-NEXT: --> ((412 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.51 = getelementptr inbounds nuw i8, ptr %add.ptr.51, i32 %div1
+; CHECK-NEXT: --> ((416 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.52 = getelementptr inbounds nuw i8, ptr %add.ptr1.51, i32 %div1
+; CHECK-NEXT: --> ((420 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.52 = getelementptr inbounds nuw i8, ptr %add.ptr.52, i32 %div1
+; CHECK-NEXT: --> ((424 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.53 = getelementptr inbounds nuw i8, ptr %add.ptr1.52, i32 %div1
+; CHECK-NEXT: --> ((428 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.53 = getelementptr inbounds nuw i8, ptr %add.ptr.53, i32 %div1
+; CHECK-NEXT: --> ((432 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.54 = getelementptr inbounds nuw i8, ptr %add.ptr1.53, i32 %div1
+; CHECK-NEXT: --> ((436 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.54 = getelementptr inbounds nuw i8, ptr %add.ptr.54, i32 %div1
+; CHECK-NEXT: --> ((440 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.55 = getelementptr inbounds nuw i8, ptr %add.ptr1.54, i32 %div1
+; CHECK-NEXT: --> ((444 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.55 = getelementptr inbounds nuw i8, ptr %add.ptr.55, i32 %div1
+; CHECK-NEXT: --> ((448 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.56 = getelementptr inbounds nuw i8, ptr %add.ptr1.55, i32 %div1
+; CHECK-NEXT: --> ((452 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.56 = getelementptr inbounds nuw i8, ptr %add.ptr.56, i32 %div1
+; CHECK-NEXT: --> ((456 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.57 = getelementptr inbounds nuw i8, ptr %add.ptr1.56, i32 %div1
+; CHECK-NEXT: --> ((460 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.57 = getelementptr inbounds nuw i8, ptr %add.ptr.57, i32 %div1
+; CHECK-NEXT: --> ((464 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.58 = getelementptr inbounds nuw i8, ptr %add.ptr1.57, i32 %div1
+; CHECK-NEXT: --> ((468 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.58 = getelementptr inbounds nuw i8, ptr %add.ptr.58, i32 %div1
+; CHECK-NEXT: --> ((472 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.59 = getelementptr inbounds nuw i8, ptr %add.ptr1.58, i32 %div1
+; CHECK-NEXT: --> ((476 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.59 = getelementptr inbounds nuw i8, ptr %add.ptr.59, i32 %div1
+; CHECK-NEXT: --> ((480 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.60 = getelementptr inbounds nuw i8, ptr %add.ptr1.59, i32 %div1
+; CHECK-NEXT: --> ((484 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.60 = getelementptr inbounds nuw i8, ptr %add.ptr.60, i32 %div1
+; CHECK-NEXT: --> ((488 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.61 = getelementptr inbounds nuw i8, ptr %add.ptr1.60, i32 %div1
+; CHECK-NEXT: --> ((492 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.61 = getelementptr inbounds nuw i8, ptr %add.ptr.61, i32 %div1
+; CHECK-NEXT: --> ((496 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.62 = getelementptr inbounds nuw i8, ptr %add.ptr1.61, i32 %div1
+; CHECK-NEXT: --> ((500 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.62 = getelementptr inbounds nuw i8, ptr %add.ptr.62, i32 %div1
+; CHECK-NEXT: --> ((504 * (%d /u 4)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.63 = getelementptr inbounds nuw i8, ptr %add.ptr1.62, i32 %div1
+; CHECK-NEXT: --> ((508 * (%d /u 4)) + %p) U: full-set S: full-set
; CHECK-NEXT: Determining loop execution counts for: @test1
-; CHECK-NEXT: Loop %for.body: backedge-taken count is i32 63
-; CHECK-NEXT: Loop %for.body: constant max backedge-taken count is i32 63
-; CHECK-NEXT: Loop %for.body: symbolic max backedge-taken count is i32 63
-; CHECK-NEXT: Loop %for.body: Trip multiple is 64
;
entry:
%div = udiv i32 %d, 4
@@ -60,23 +300,265 @@ for.end: ; preds = %for.cond
define void @test1a(i32 %d, ptr %p) nounwind uwtable ssp {
; CHECK-LABEL: 'test1a'
; CHECK-NEXT: Classifying expressions for: @test1a
-; CHECK-NEXT: %div1 = lshr i32 %d, 1
-; CHECK-NEXT: --> (%d /u 2) U: [0,-2147483648) S: [0,-2147483648)
-; CHECK-NEXT: %i.03 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
-; CHECK-NEXT: --> {0,+,1}<nuw><nsw><%for.body> U: [0,64) S: [0,64) Exits: 63 LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %p.addr.02 = phi ptr [ %p, %entry ], [ %add.ptr1, %for.body ]
-; CHECK-NEXT: --> {%p,+,(8 * (%d /u 2))}<%for.body> U: full-set S: full-set Exits: ((504 * (%d /u 2)) + %p) LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %add.ptr = getelementptr inbounds nuw i32, ptr %p.addr.02, i32 %div1
-; CHECK-NEXT: --> {((4 * (%d /u 2))<nuw><nsw> + %p)<nuw>,+,(8 * (%d /u 2))}<%for.body> U: full-set S: full-set Exits: ((508 * (%d /u 2)) + %p) LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %add.ptr1 = getelementptr inbounds nuw i32, ptr %add.ptr, i32 %div1
-; CHECK-NEXT: --> {((8 * (%d /u 2)) + %p),+,(8 * (%d /u 2))}<%for.body> U: full-set S: full-set Exits: ((512 * (%d /u 2)) + %p) LoopDispositions: { %for.body: Computable }
-; CHECK-NEXT: %inc = add nuw nsw i32 %i.03, 1
-; CHECK-NEXT: --> {1,+,1}<nuw><nsw><%for.body> U: [1,65) S: [1,65) Exits: 64 LoopDispositions: { %for.body: Computable }
+; CHECK-NEXT: %0 = shl nuw nsw i32 %d, 1
+; CHECK-NEXT: --> (2 * %d)<nuw><nsw> U: [0,-1) S: [-2147483648,2147483647)
+; CHECK-NEXT: %1 = and i32 %0, 2147483644
+; CHECK-NEXT: --> (4 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32))<nuw><nsw> U: [0,2147483645) S: [0,2147483645)
+; CHECK-NEXT: %add.ptr = getelementptr inbounds nuw i8, ptr %p, i32 %1
+; CHECK-NEXT: --> ((4 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32))<nuw><nsw> + %p)<nuw> U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1 = getelementptr inbounds nuw i8, ptr %add.ptr, i32 %1
+; CHECK-NEXT: --> ((8 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32))<nuw> + %p)<nuw> U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.1 = getelementptr inbounds nuw i8, ptr %add.ptr1, i32 %1
+; CHECK-NEXT: --> ((12 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.1 = getelementptr inbounds nuw i8, ptr %add.ptr.1, i32 %1
+; CHECK-NEXT: --> ((16 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.2 = getelementptr inbounds nuw i8, ptr %add.ptr1.1, i32 %1
+; CHECK-NEXT: --> ((20 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.2 = getelementptr inbounds nuw i8, ptr %add.ptr.2, i32 %1
+; CHECK-NEXT: --> ((24 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.3 = getelementptr inbounds nuw i8, ptr %add.ptr1.2, i32 %1
+; CHECK-NEXT: --> ((28 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.3 = getelementptr inbounds nuw i8, ptr %add.ptr.3, i32 %1
+; CHECK-NEXT: --> ((32 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.4 = getelementptr inbounds nuw i8, ptr %add.ptr1.3, i32 %1
+; CHECK-NEXT: --> ((36 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.4 = getelementptr inbounds nuw i8, ptr %add.ptr.4, i32 %1
+; CHECK-NEXT: --> ((40 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.5 = getelementptr inbounds nuw i8, ptr %add.ptr1.4, i32 %1
+; CHECK-NEXT: --> ((44 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.5 = getelementptr inbounds nuw i8, ptr %add.ptr.5, i32 %1
+; CHECK-NEXT: --> ((48 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.6 = getelementptr inbounds nuw i8, ptr %add.ptr1.5, i32 %1
+; CHECK-NEXT: --> ((52 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.6 = getelementptr inbounds nuw i8, ptr %add.ptr.6, i32 %1
+; CHECK-NEXT: --> ((56 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.7 = getelementptr inbounds nuw i8, ptr %add.ptr1.6, i32 %1
+; CHECK-NEXT: --> ((60 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.7 = getelementptr inbounds nuw i8, ptr %add.ptr.7, i32 %1
+; CHECK-NEXT: --> ((64 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.8 = getelementptr inbounds nuw i8, ptr %add.ptr1.7, i32 %1
+; CHECK-NEXT: --> ((68 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.8 = getelementptr inbounds nuw i8, ptr %add.ptr.8, i32 %1
+; CHECK-NEXT: --> ((72 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.9 = getelementptr inbounds nuw i8, ptr %add.ptr1.8, i32 %1
+; CHECK-NEXT: --> ((76 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.9 = getelementptr inbounds nuw i8, ptr %add.ptr.9, i32 %1
+; CHECK-NEXT: --> ((80 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.10 = getelementptr inbounds nuw i8, ptr %add.ptr1.9, i32 %1
+; CHECK-NEXT: --> ((84 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.10 = getelementptr inbounds nuw i8, ptr %add.ptr.10, i32 %1
+; CHECK-NEXT: --> ((88 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.11 = getelementptr inbounds nuw i8, ptr %add.ptr1.10, i32 %1
+; CHECK-NEXT: --> ((92 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.11 = getelementptr inbounds nuw i8, ptr %add.ptr.11, i32 %1
+; CHECK-NEXT: --> ((96 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.12 = getelementptr inbounds nuw i8, ptr %add.ptr1.11, i32 %1
+; CHECK-NEXT: --> ((100 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.12 = getelementptr inbounds nuw i8, ptr %add.ptr.12, i32 %1
+; CHECK-NEXT: --> ((104 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.13 = getelementptr inbounds nuw i8, ptr %add.ptr1.12, i32 %1
+; CHECK-NEXT: --> ((108 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.13 = getelementptr inbounds nuw i8, ptr %add.ptr.13, i32 %1
+; CHECK-NEXT: --> ((112 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.14 = getelementptr inbounds nuw i8, ptr %add.ptr1.13, i32 %1
+; CHECK-NEXT: --> ((116 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.14 = getelementptr inbounds nuw i8, ptr %add.ptr.14, i32 %1
+; CHECK-NEXT: --> ((120 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.15 = getelementptr inbounds nuw i8, ptr %add.ptr1.14, i32 %1
+; CHECK-NEXT: --> ((124 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.15 = getelementptr inbounds nuw i8, ptr %add.ptr.15, i32 %1
+; CHECK-NEXT: --> ((128 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.16 = getelementptr inbounds nuw i8, ptr %add.ptr1.15, i32 %1
+; CHECK-NEXT: --> ((132 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.16 = getelementptr inbounds nuw i8, ptr %add.ptr.16, i32 %1
+; CHECK-NEXT: --> ((136 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.17 = getelementptr inbounds nuw i8, ptr %add.ptr1.16, i32 %1
+; CHECK-NEXT: --> ((140 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.17 = getelementptr inbounds nuw i8, ptr %add.ptr.17, i32 %1
+; CHECK-NEXT: --> ((144 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.18 = getelementptr inbounds nuw i8, ptr %add.ptr1.17, i32 %1
+; CHECK-NEXT: --> ((148 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.18 = getelementptr inbounds nuw i8, ptr %add.ptr.18, i32 %1
+; CHECK-NEXT: --> ((152 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.19 = getelementptr inbounds nuw i8, ptr %add.ptr1.18, i32 %1
+; CHECK-NEXT: --> ((156 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.19 = getelementptr inbounds nuw i8, ptr %add.ptr.19, i32 %1
+; CHECK-NEXT: --> ((160 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.20 = getelementptr inbounds nuw i8, ptr %add.ptr1.19, i32 %1
+; CHECK-NEXT: --> ((164 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.20 = getelementptr inbounds nuw i8, ptr %add.ptr.20, i32 %1
+; CHECK-NEXT: --> ((168 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.21 = getelementptr inbounds nuw i8, ptr %add.ptr1.20, i32 %1
+; CHECK-NEXT: --> ((172 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.21 = getelementptr inbounds nuw i8, ptr %add.ptr.21, i32 %1
+; CHECK-NEXT: --> ((176 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.22 = getelementptr inbounds nuw i8, ptr %add.ptr1.21, i32 %1
+; CHECK-NEXT: --> ((180 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.22 = getelementptr inbounds nuw i8, ptr %add.ptr.22, i32 %1
+; CHECK-NEXT: --> ((184 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.23 = getelementptr inbounds nuw i8, ptr %add.ptr1.22, i32 %1
+; CHECK-NEXT: --> ((188 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.23 = getelementptr inbounds nuw i8, ptr %add.ptr.23, i32 %1
+; CHECK-NEXT: --> ((192 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.24 = getelementptr inbounds nuw i8, ptr %add.ptr1.23, i32 %1
+; CHECK-NEXT: --> ((196 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.24 = getelementptr inbounds nuw i8, ptr %add.ptr.24, i32 %1
+; CHECK-NEXT: --> ((200 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.25 = getelementptr inbounds nuw i8, ptr %add.ptr1.24, i32 %1
+; CHECK-NEXT: --> ((204 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.25 = getelementptr inbounds nuw i8, ptr %add.ptr.25, i32 %1
+; CHECK-NEXT: --> ((208 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.26 = getelementptr inbounds nuw i8, ptr %add.ptr1.25, i32 %1
+; CHECK-NEXT: --> ((212 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.26 = getelementptr inbounds nuw i8, ptr %add.ptr.26, i32 %1
+; CHECK-NEXT: --> ((216 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.27 = getelementptr inbounds nuw i8, ptr %add.ptr1.26, i32 %1
+; CHECK-NEXT: --> ((220 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.27 = getelementptr inbounds nuw i8, ptr %add.ptr.27, i32 %1
+; CHECK-NEXT: --> ((224 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.28 = getelementptr inbounds nuw i8, ptr %add.ptr1.27, i32 %1
+; CHECK-NEXT: --> ((228 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.28 = getelementptr inbounds nuw i8, ptr %add.ptr.28, i32 %1
+; CHECK-NEXT: --> ((232 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.29 = getelementptr inbounds nuw i8, ptr %add.ptr1.28, i32 %1
+; CHECK-NEXT: --> ((236 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.29 = getelementptr inbounds nuw i8, ptr %add.ptr.29, i32 %1
+; CHECK-NEXT: --> ((240 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.30 = getelementptr inbounds nuw i8, ptr %add.ptr1.29, i32 %1
+; CHECK-NEXT: --> ((244 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.30 = getelementptr inbounds nuw i8, ptr %add.ptr.30, i32 %1
+; CHECK-NEXT: --> ((248 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.31 = getelementptr inbounds nuw i8, ptr %add.ptr1.30, i32 %1
+; CHECK-NEXT: --> ((252 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.31 = getelementptr inbounds nuw i8, ptr %add.ptr.31, i32 %1
+; CHECK-NEXT: --> ((256 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.32 = getelementptr inbounds nuw i8, ptr %add.ptr1.31, i32 %1
+; CHECK-NEXT: --> ((260 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.32 = getelementptr inbounds nuw i8, ptr %add.ptr.32, i32 %1
+; CHECK-NEXT: --> ((264 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.33 = getelementptr inbounds nuw i8, ptr %add.ptr1.32, i32 %1
+; CHECK-NEXT: --> ((268 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.33 = getelementptr inbounds nuw i8, ptr %add.ptr.33, i32 %1
+; CHECK-NEXT: --> ((272 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.34 = getelementptr inbounds nuw i8, ptr %add.ptr1.33, i32 %1
+; CHECK-NEXT: --> ((276 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.34 = getelementptr inbounds nuw i8, ptr %add.ptr.34, i32 %1
+; CHECK-NEXT: --> ((280 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.35 = getelementptr inbounds nuw i8, ptr %add.ptr1.34, i32 %1
+; CHECK-NEXT: --> ((284 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.35 = getelementptr inbounds nuw i8, ptr %add.ptr.35, i32 %1
+; CHECK-NEXT: --> ((288 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.36 = getelementptr inbounds nuw i8, ptr %add.ptr1.35, i32 %1
+; CHECK-NEXT: --> ((292 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.36 = getelementptr inbounds nuw i8, ptr %add.ptr.36, i32 %1
+; CHECK-NEXT: --> ((296 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.37 = getelementptr inbounds nuw i8, ptr %add.ptr1.36, i32 %1
+; CHECK-NEXT: --> ((300 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.37 = getelementptr inbounds nuw i8, ptr %add.ptr.37, i32 %1
+; CHECK-NEXT: --> ((304 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.38 = getelementptr inbounds nuw i8, ptr %add.ptr1.37, i32 %1
+; CHECK-NEXT: --> ((308 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.38 = getelementptr inbounds nuw i8, ptr %add.ptr.38, i32 %1
+; CHECK-NEXT: --> ((312 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.39 = getelementptr inbounds nuw i8, ptr %add.ptr1.38, i32 %1
+; CHECK-NEXT: --> ((316 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.39 = getelementptr inbounds nuw i8, ptr %add.ptr.39, i32 %1
+; CHECK-NEXT: --> ((320 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.40 = getelementptr inbounds nuw i8, ptr %add.ptr1.39, i32 %1
+; CHECK-NEXT: --> ((324 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.40 = getelementptr inbounds nuw i8, ptr %add.ptr.40, i32 %1
+; CHECK-NEXT: --> ((328 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.41 = getelementptr inbounds nuw i8, ptr %add.ptr1.40, i32 %1
+; CHECK-NEXT: --> ((332 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.41 = getelementptr inbounds nuw i8, ptr %add.ptr.41, i32 %1
+; CHECK-NEXT: --> ((336 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.42 = getelementptr inbounds nuw i8, ptr %add.ptr1.41, i32 %1
+; CHECK-NEXT: --> ((340 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.42 = getelementptr inbounds nuw i8, ptr %add.ptr.42, i32 %1
+; CHECK-NEXT: --> ((344 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.43 = getelementptr inbounds nuw i8, ptr %add.ptr1.42, i32 %1
+; CHECK-NEXT: --> ((348 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.43 = getelementptr inbounds nuw i8, ptr %add.ptr.43, i32 %1
+; CHECK-NEXT: --> ((352 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.44 = getelementptr inbounds nuw i8, ptr %add.ptr1.43, i32 %1
+; CHECK-NEXT: --> ((356 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.44 = getelementptr inbounds nuw i8, ptr %add.ptr.44, i32 %1
+; CHECK-NEXT: --> ((360 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.45 = getelementptr inbounds nuw i8, ptr %add.ptr1.44, i32 %1
+; CHECK-NEXT: --> ((364 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.45 = getelementptr inbounds nuw i8, ptr %add.ptr.45, i32 %1
+; CHECK-NEXT: --> ((368 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.46 = getelementptr inbounds nuw i8, ptr %add.ptr1.45, i32 %1
+; CHECK-NEXT: --> ((372 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.46 = getelementptr inbounds nuw i8, ptr %add.ptr.46, i32 %1
+; CHECK-NEXT: --> ((376 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.47 = getelementptr inbounds nuw i8, ptr %add.ptr1.46, i32 %1
+; CHECK-NEXT: --> ((380 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.47 = getelementptr inbounds nuw i8, ptr %add.ptr.47, i32 %1
+; CHECK-NEXT: --> ((384 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.48 = getelementptr inbounds nuw i8, ptr %add.ptr1.47, i32 %1
+; CHECK-NEXT: --> ((388 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.48 = getelementptr inbounds nuw i8, ptr %add.ptr.48, i32 %1
+; CHECK-NEXT: --> ((392 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.49 = getelementptr inbounds nuw i8, ptr %add.ptr1.48, i32 %1
+; CHECK-NEXT: --> ((396 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.49 = getelementptr inbounds nuw i8, ptr %add.ptr.49, i32 %1
+; CHECK-NEXT: --> ((400 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.50 = getelementptr inbounds nuw i8, ptr %add.ptr1.49, i32 %1
+; CHECK-NEXT: --> ((404 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.50 = getelementptr inbounds nuw i8, ptr %add.ptr.50, i32 %1
+; CHECK-NEXT: --> ((408 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.51 = getelementptr inbounds nuw i8, ptr %add.ptr1.50, i32 %1
+; CHECK-NEXT: --> ((412 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.51 = getelementptr inbounds nuw i8, ptr %add.ptr.51, i32 %1
+; CHECK-NEXT: --> ((416 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.52 = getelementptr inbounds nuw i8, ptr %add.ptr1.51, i32 %1
+; CHECK-NEXT: --> ((420 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.52 = getelementptr inbounds nuw i8, ptr %add.ptr.52, i32 %1
+; CHECK-NEXT: --> ((424 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.53 = getelementptr inbounds nuw i8, ptr %add.ptr1.52, i32 %1
+; CHECK-NEXT: --> ((428 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.53 = getelementptr inbounds nuw i8, ptr %add.ptr.53, i32 %1
+; CHECK-NEXT: --> ((432 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.54 = getelementptr inbounds nuw i8, ptr %add.ptr1.53, i32 %1
+; CHECK-NEXT: --> ((436 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.54 = getelementptr inbounds nuw i8, ptr %add.ptr.54, i32 %1
+; CHECK-NEXT: --> ((440 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.55 = getelementptr inbounds nuw i8, ptr %add.ptr1.54, i32 %1
+; CHECK-NEXT: --> ((444 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.55 = getelementptr inbounds nuw i8, ptr %add.ptr.55, i32 %1
+; CHECK-NEXT: --> ((448 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.56 = getelementptr inbounds nuw i8, ptr %add.ptr1.55, i32 %1
+; CHECK-NEXT: --> ((452 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.56 = getelementptr inbounds nuw i8, ptr %add.ptr.56, i32 %1
+; CHECK-NEXT: --> ((456 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.57 = getelementptr inbounds nuw i8, ptr %add.ptr1.56, i32 %1
+; CHECK-NEXT: --> ((460 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.57 = getelementptr inbounds nuw i8, ptr %add.ptr.57, i32 %1
+; CHECK-NEXT: --> ((464 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.58 = getelementptr inbounds nuw i8, ptr %add.ptr1.57, i32 %1
+; CHECK-NEXT: --> ((468 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.58 = getelementptr inbounds nuw i8, ptr %add.ptr.58, i32 %1
+; CHECK-NEXT: --> ((472 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.59 = getelementptr inbounds nuw i8, ptr %add.ptr1.58, i32 %1
+; CHECK-NEXT: --> ((476 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.59 = getelementptr inbounds nuw i8, ptr %add.ptr.59, i32 %1
+; CHECK-NEXT: --> ((480 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.60 = getelementptr inbounds nuw i8, ptr %add.ptr1.59, i32 %1
+; CHECK-NEXT: --> ((484 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.60 = getelementptr inbounds nuw i8, ptr %add.ptr.60, i32 %1
+; CHECK-NEXT: --> ((488 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.61 = getelementptr inbounds nuw i8, ptr %add.ptr1.60, i32 %1
+; CHECK-NEXT: --> ((492 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.61 = getelementptr inbounds nuw i8, ptr %add.ptr.61, i32 %1
+; CHECK-NEXT: --> ((496 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.62 = getelementptr inbounds nuw i8, ptr %add.ptr1.61, i32 %1
+; CHECK-NEXT: --> ((500 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr1.62 = getelementptr inbounds nuw i8, ptr %add.ptr.62, i32 %1
+; CHECK-NEXT: --> ((504 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
+; CHECK-NEXT: %add.ptr.63 = getelementptr inbounds nuw i8, ptr %add.ptr1.62, i32 %1
+; CHECK-NEXT: --> ((508 * (zext i29 (trunc i32 (%d /u 2) to i29) to i32)) + %p) U: full-set S: full-set
; CHECK-NEXT: Determining loop execution counts for: @test1a
-; CHECK-NEXT: Loop %for.body: backedge-taken count is i32 63
-; CHECK-NEXT: Loop %for.body: constant max backedge-taken count is i32 63
-; CHECK-NEXT: Loop %for.body: symbolic max backedge-taken count is i32 63
-; CHECK-NEXT: Loop %for.body: Trip multiple is 64
;
entry:
%div = udiv i32 %d, 2
@@ -111,7 +593,9 @@ define void @test_range_ref1a(i32 %x) {
; CHECK-NEXT: Classifying expressions for: @test_range_ref1a
; CHECK-NEXT: %i.01.0 = phi i32 [ 100, %entry ], [ %tmp4, %bb ]
; CHECK-NEXT: --> {100,+,-1}<nsw><%bb> U: [0,101) S: [0,101) Exits: 0 LoopDispositions: { %bb: Computable }
-; CHECK-NEXT: %tmp1 = getelementptr i32, ptr @array, i32 %i.01.0
+; CHECK-NEXT: %0 = shl i32 %i.01.0, 2
+; CHECK-NEXT: --> {400,+,-4}<nsw><%bb> U: [0,401) S: [0,401) Exits: 0 LoopDispositions: { %bb: Computable }
+; CHECK-NEXT: %tmp1 = getelementptr i8, ptr @array, i32 %0
; CHECK-NEXT: --> {(400 + @array)<nuw>,+,-4}<nw><%bb> U: [0,-3) S: [-2147483648,2147483645) Exits: @array LoopDispositions: { %bb: Computable }
; CHECK-NEXT: %tmp4 = add nsw i32 %i.01.0, -1
; CHECK-NEXT: --> {99,+,-1}<nsw><%bb> U: [-1,100) S: [-1,100) Exits: -1 LoopDispositions: { %bb: Computable }
@@ -141,13 +625,15 @@ define i32 @test_loop_idiom_recogize(i32 %x, i32 %y, ptr %lam, ptr %alp) nounwin
; CHECK-NEXT: Classifying expressions for: @test_loop_idiom_recogize
; CHECK-NEXT: %indvar = phi i32 [ 0, %bb1.thread ], [ %indvar.next, %bb1 ]
; CHECK-NEXT: --> {0,+,1}<nuw><nsw><%bb1> U: [0,256) S: [0,256) Exits: 255 LoopDispositions: { %bb1: Computable }
-; CHECK-NEXT: %i.0.reg2mem.0 = sub nuw nsw i32 255, %indvar
-; CHECK-NEXT: --> {255,+,-1}<nsw><%bb1> U: [0,256) S: [0,256) Exits: 0 LoopDispositions: { %bb1: Computable }
-; CHECK-NEXT: %0 = getelementptr i32, ptr %alp, i32 %i.0.reg2mem.0
+; CHECK-NEXT: %0 = shl i32 %indvar, 2
+; CHECK-NEXT: --> {0,+,4}<nuw><nsw><%bb1> U: [0,1021) S: [0,1021) Exits: 1020 LoopDispositions: { %bb1: Computable }
+; CHECK-NEXT: %1 = sub nuw nsw i32 1020, %0
+; CHECK-NEXT: --> {1020,+,-4}<nsw><%bb1> U: [0,1021) S: [0,1021) Exits: 0 LoopDispositions: { %bb1: Computable }
+; CHECK-NEXT: %2 = getelementptr i8, ptr %alp, i32 %1
; CHECK-NEXT: --> {(1020 + %alp),+,-4}<nw><%bb1> U: full-set S: full-set Exits: %alp LoopDispositions: { %bb1: Computable }
-; CHECK-NEXT: %1 = load i32, ptr %0, align 4
-; CHECK-NEXT: --> %1 U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %bb1: Variant }
-; CHECK-NEXT: %2 = getelementptr i32, ptr %lam, i32 %i.0.reg2mem.0
+; CHECK-NEXT: %3 = load i32, ptr %2, align 4
+; CHECK-NEXT: --> %3 U: full-set S: full-set Exits: <<Unknown>> LoopDispositions: { %bb1: Variant }
+; CHECK-NEXT: %4 = getelementptr i8, ptr %lam, i32 %1
; CHECK-NEXT: --> {(1020 + %lam),+,-4}<nw><%bb1> U: full-set S: full-set Exits: %lam LoopDispositions: { %bb1: Computable }
; CHECK-NEXT: %indvar.next = add nuw nsw i32 %indvar, 1
; CHECK-NEXT: --> {1,+,1}<nuw><nsw><%bb1> U: [1,257) S: [1,257) Exits: 256 LoopDispositions: { %bb1: Computable }
@@ -155,7 +641,7 @@ define i32 @test_loop_idiom_recogize(i32 %x, i32 %y, ptr %lam, ptr %alp) nounwin
; CHECK-NEXT: --> (255 * %x) U: full-set S: full-set
; CHECK-NEXT: %z.0.reg2mem.0 = add i32 %y, %x
; CHECK-NEXT: --> (%x + %y) U: full-set S: full-set
-; CHECK-NEXT: %3 = add i32 %z.0.reg2mem.0, %tmp10
+; CHECK-NEXT: %5 = add i32 %z.0.reg2mem.0, %tmp10
; CHECK-NEXT: --> ((256 * %x) + %y) U: full-set S: full-set
; CHECK-NEXT: Determining loop execution counts for: @test_loop_idiom_recogize
; CHECK-NEXT: Loop %bb1: backedge-taken count is i32 255
diff --git a/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll b/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll
index ba71299d6919e..482919f09cc3b 100644
--- a/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll
+++ b/llvm/test/Transforms/PhaseOrdering/simplifycfg-options.ll
@@ -70,8 +70,9 @@ declare void @foo()
define double @max_of_loads(ptr %x, ptr %y, i64 %i) {
; CHECK-LABEL: @max_of_loads(
; CHECK-NEXT: entry:
-; CHECK-NEXT: [[XI_PTR:%.*]] = getelementptr double, ptr [[X:%.*]], i64 [[I:%.*]]
-; CHECK-NEXT: [[YI_PTR:%.*]] = getelementptr double, ptr [[Y:%.*]], i64 [[I]]
+; CHECK-NEXT: [[TMP0:%.*]] = shl i64 [[I:%.*]], 3
+; CHECK-NEXT: [[XI_PTR:%.*]] = getelementptr i8, ptr [[X:%.*]], i64 [[TMP0]]
+; CHECK-NEXT: [[YI_PTR:%.*]] = getelementptr i8, ptr [[Y:%.*]], i64 [[TMP0]]
; CHECK-NEXT: [[XI:%.*]] = load double, ptr [[XI_PTR]], align 8
; CHECK-NEXT: [[YI:%.*]] = load double, ptr [[YI_PTR]], align 8
; CHECK-NEXT: [[CMP:%.*]] = fcmp ogt double [[XI]], [[YI]]
diff --git a/llvm/test/Transforms/SimpleLoopUnswitch/AMDGPU/uniform-unswitch.ll b/llvm/test/Transforms/SimpleLoopUnswitch/AMDGPU/uniform-unswitch.ll
index cdd0fcab59b7d..71aeec44f0b13 100644
--- a/llvm/test/Transforms/SimpleLoopUnswitch/AMDGPU/uniform-unswitch.ll
+++ b/llvm/test/Transforms/SimpleLoopUnswitch/AMDGPU/uniform-unswitch.ll
@@ -35,7 +35,8 @@ define amdgpu_kernel void @uniform_unswitch(ptr nocapture %out, i32 %n, i32 %x)
; CHECK-NEXT: br i1 [[CMP1]], label [[IF_THEN:%.*]], label [[FOR_INC]]
; CHECK: if.then:
; CHECK-NEXT: [[TMP0:%.*]] = zext nneg i32 [[I_07]] to i64
-; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr addrspace(1) [[OUT_GLOBAL]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP2:%.*]] = shl nuw nsw i64 [[TMP0]], 2
+; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr addrspace(1) [[OUT_GLOBAL]], i64 [[TMP2]]
; CHECK-NEXT: store i32 [[I_07]], ptr addrspace(1) [[ARRAYIDX]], align 4
; CHECK-NEXT: br label [[FOR_INC]]
; CHECK: for.inc:
diff --git a/llvm/test/Transforms/SimplifyCFG/Hexagon/switch-to-lookup-table.ll b/llvm/test/Transforms/SimplifyCFG/Hexagon/switch-to-lookup-table.ll
index 4f21c06e519ff..194eef1851065 100644
--- a/llvm/test/Transforms/SimplifyCFG/Hexagon/switch-to-lookup-table.ll
+++ b/llvm/test/Transforms/SimplifyCFG/Hexagon/switch-to-lookup-table.ll
@@ -13,7 +13,8 @@ define i32 @foo(i32 %x) section ".tcm_text" {
; ENABLE-NEXT: [[TMP0:%.*]] = icmp ult i32 [[X:%.*]], 6
; ENABLE-NEXT: br i1 [[TMP0]], label [[SWITCH_LOOKUP:%.*]], label [[RETURN:%.*]]
; ENABLE: switch.lookup:
-; ENABLE-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i32, ptr @switch.table.foo, i32 [[X]]
+; ENABLE-NEXT: [[TMP1:%.*]] = shl nuw nsw i32 [[X]], 2
+; ENABLE-NEXT: [[SWITCH_GEP:%.*]] = getelementptr inbounds nuw i8, ptr @switch.table.foo, i32 [[TMP1]]
; ENABLE-NEXT: [[SWITCH_LOAD:%.*]] = load i32, ptr [[SWITCH_GEP]], align 4
; ENABLE-NEXT: br label [[RETURN]]
; ENABLE: return:
More information about the llvm-commits
mailing list