From a72162cc529cbfb6587176332b9421fbfb638936 Mon Sep 17 00:00:00 2001 From: Guozhi Wei Date: Fri, 31 Mar 2023 16:58:41 +0000 Subject: [PATCH] [AARCH64] Enable STORE of v4i8 to help more vectorization opportunities For the attached test case, currently llvm generates instructions to load/or/store the bytes one by one. Although NEON doesn't support v4i8 natively, we can promote it to v4i16 and operate on v4i16 vectors. So this patch override getStoreMinimumVF and specify the minimum VF for i8 vector is v4i8. Differential Revision: https://reviews.llvm.org/D145614 --- .../Target/AArch64/AArch64TargetTransformInfo.h | 9 ++ .../Transforms/SLPVectorizer/AArch64/32-bit.ll | 56 +++++++++ .../AArch64/vectorizable-selects-uniform-cmps.ll | 128 ++++++++------------- 3 files changed, 116 insertions(+), 77 deletions(-) create mode 100644 llvm/test/Transforms/SLPVectorizer/AArch64/32-bit.ll diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h index 287c582..df69c37 100644 --- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h +++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h @@ -397,6 +397,15 @@ public: /// @} bool enableSelectOptimize() { return ST->enableSelectOptimize(); } + + unsigned getStoreMinimumVF(unsigned VF, Type *ScalarMemTy, + Type *ScalarValTy) const { + // We can vectorize store v4i8. + if (ScalarMemTy->isIntegerTy(8) && isPowerOf2_32(VF) && VF >= 4) + return 4; + + return BaseT::getStoreMinimumVF(VF, ScalarMemTy, ScalarValTy); + } }; } // end namespace llvm diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/32-bit.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/32-bit.ll new file mode 100644 index 0000000..5d91e03 --- /dev/null +++ b/llvm/test/Transforms/SLPVectorizer/AArch64/32-bit.ll @@ -0,0 +1,56 @@ +; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2 +; RUN: opt -passes=slp-vectorizer -S < %s | FileCheck %s + +target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128" +target triple = "aarch64-unknown-linux-gnu" + +%S = type { i8, i8, i8, i8 } + +define ptr @foo(ptr %this, ptr %rhs) { +; CHECK-LABEL: define ptr @foo +; CHECK-SAME: (ptr [[THIS:%.*]], ptr [[RHS:%.*]]) { +; CHECK-NEXT: entry: +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[RHS]], align 1, !tbaa [[TBAA0:![0-9]+]] +; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i8>, ptr [[THIS]], align 1, !tbaa [[TBAA0]] +; CHECK-NEXT: [[TMP2:%.*]] = or <4 x i8> [[TMP0]], [[TMP1]] +; CHECK-NEXT: store <4 x i8> [[TMP2]], ptr [[THIS]], align 1, !tbaa [[TBAA0]] +; CHECK-NEXT: ret ptr [[THIS]] +; +entry: + %right1 = load i8, ptr %rhs, align 1, !tbaa !6, !range !11, !noundef !12 + %left1 = load i8, ptr %this, align 1, !tbaa !6, !range !11, !noundef !12 + %res1 = or i8 %right1, %left1 + store i8 %res1, ptr %this, align 1, !tbaa !6 + %b = getelementptr inbounds %S, ptr %rhs, i64 0, i32 1 + %right2 = load i8, ptr %b, align 1, !tbaa !13, !range !11, !noundef !12 + %b8 = getelementptr inbounds %S, ptr %this, i64 0, i32 1 + %left2 = load i8, ptr %b8, align 1, !tbaa !13, !range !11, !noundef !12 + %res2 = or i8 %right2, %left2 + store i8 %res2, ptr %b8, align 1, !tbaa !13 + %c = getelementptr inbounds %S, ptr %rhs, i64 0, i32 2 + %right3 = load i8, ptr %c, align 1, !tbaa !14, !range !11, !noundef !12 + %c16 = getelementptr inbounds %S, ptr %this, i64 0, i32 2 + %left3 = load i8, ptr %c16, align 1, !tbaa !14, !range !11, !noundef !12 + %res3 = or i8 %right3, %left3 + store i8 %res3, ptr %c16, align 1, !tbaa !14 + %d = getelementptr inbounds %S, ptr %rhs, i64 0, i32 3 + %right4 = load i8, ptr %d, align 1, !tbaa !15, !range !11, !noundef !12 + %d24 = getelementptr inbounds %S, ptr %this, i64 0, i32 3 + %left4 = load i8, ptr %d24, align 1, !tbaa !15, !range !11, !noundef !12 + %res4 = or i8 %right4, %left4 + store i8 %res4, ptr %d24, align 1, !tbaa !15 + ret ptr %this +} + + +!6 = !{!7, !8, i64 0} +!7 = !{!"S", !8, i64 0, !8, i64 1, !8, i64 2, !8, i64 3} +!8 = !{!"bool", !9, i64 0} +!9 = !{!"omnipotent char", !10, i64 0} +!10 = !{!"Simple C++ TBAA"} +!11 = !{i8 0, i8 2} +!12 = !{} +!13 = !{!7, !8, i64 1} +!14 = !{!7, !8, i64 2} +!15 = !{!7, !8, i64 3} + diff --git a/llvm/test/Transforms/SLPVectorizer/AArch64/vectorizable-selects-uniform-cmps.ll b/llvm/test/Transforms/SLPVectorizer/AArch64/vectorizable-selects-uniform-cmps.ll index 02468f3..69d0e62 100644 --- a/llvm/test/Transforms/SLPVectorizer/AArch64/vectorizable-selects-uniform-cmps.ll +++ b/llvm/test/Transforms/SLPVectorizer/AArch64/vectorizable-selects-uniform-cmps.ll @@ -106,25 +106,12 @@ entry: define void @select_uniform_ugt_7xi8(ptr %ptr, i8 %x) { ; CHECK-LABEL: @select_uniform_ugt_7xi8( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[L_0:%.*]] = load i8, ptr [[PTR:%.*]], align 1 -; CHECK-NEXT: [[CMP_0:%.*]] = icmp ugt i8 [[L_0]], -1 -; CHECK-NEXT: [[S_0:%.*]] = select i1 [[CMP_0]], i8 [[L_0]], i8 [[X:%.*]] -; CHECK-NEXT: store i8 [[S_0]], ptr [[PTR]], align 2 -; CHECK-NEXT: [[GEP_1:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 1 -; CHECK-NEXT: [[L_1:%.*]] = load i8, ptr [[GEP_1]], align 1 -; CHECK-NEXT: [[CMP_1:%.*]] = icmp ugt i8 [[L_1]], -1 -; CHECK-NEXT: [[S_1:%.*]] = select i1 [[CMP_1]], i8 [[L_1]], i8 [[X]] -; CHECK-NEXT: store i8 [[S_1]], ptr [[GEP_1]], align 2 -; CHECK-NEXT: [[GEP_2:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 2 -; CHECK-NEXT: [[L_2:%.*]] = load i8, ptr [[GEP_2]], align 1 -; CHECK-NEXT: [[CMP_2:%.*]] = icmp ugt i8 [[L_2]], -1 -; CHECK-NEXT: [[S_2:%.*]] = select i1 [[CMP_2]], i8 [[L_2]], i8 [[X]] -; CHECK-NEXT: store i8 [[S_2]], ptr [[GEP_2]], align 2 -; CHECK-NEXT: [[GEP_3:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 3 -; CHECK-NEXT: [[L_3:%.*]] = load i8, ptr [[GEP_3]], align 1 -; CHECK-NEXT: [[CMP_3:%.*]] = icmp ugt i8 [[L_3]], -1 -; CHECK-NEXT: [[S_3:%.*]] = select i1 [[CMP_3]], i8 [[L_3]], i8 [[X]] -; CHECK-NEXT: store i8 [[S_3]], ptr [[GEP_3]], align 2 +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i8>, ptr [[PTR:%.*]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt <4 x i8> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i8> poison, i8 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x i8> [[TMP2]], <4 x i8> poison, <4 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> [[TMP1]], <4 x i8> [[TMP0]], <4 x i8> [[TMP3]] +; CHECK-NEXT: store <4 x i8> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: [[GEP_4:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 4 ; CHECK-NEXT: [[L_4:%.*]] = load i8, ptr [[GEP_4]], align 1 ; CHECK-NEXT: [[CMP_4:%.*]] = icmp ugt i8 [[L_4]], -1 @@ -193,11 +180,11 @@ entry: define void @select_uniform_ugt_8xi8(ptr %ptr, i8 %x) { ; CHECK-LABEL: @select_uniform_ugt_8xi8( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP1:%.*]] = load <8 x i8>, ptr [[PTR:%.*]], align 1 -; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt <8 x i8> [[TMP1]], -; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x i8> poison, i8 [[X:%.*]], i32 0 -; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> poison, <8 x i32> zeroinitializer -; CHECK-NEXT: [[TMP4:%.*]] = select <8 x i1> [[TMP2]], <8 x i8> [[TMP1]], <8 x i8> [[SHUFFLE]] +; CHECK-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[PTR:%.*]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt <8 x i8> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x i8> poison, i8 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <8 x i8> [[TMP2]], <8 x i8> poison, <8 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <8 x i1> [[TMP1]], <8 x i8> [[TMP0]], <8 x i8> [[TMP3]] ; CHECK-NEXT: store <8 x i8> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: ret void ; @@ -254,39 +241,26 @@ entry: define void @select_uniform_ugt_16xi8(ptr %ptr, i8 %x) { ; CHECK-LABEL: @select_uniform_ugt_16xi8( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP1:%.*]] = load <8 x i8>, ptr [[PTR:%.*]], align 1 -; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt <8 x i8> [[TMP1]], -; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x i8> poison, i8 [[X:%.*]], i32 0 -; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> poison, <8 x i32> zeroinitializer -; CHECK-NEXT: [[TMP4:%.*]] = select <8 x i1> [[TMP2]], <8 x i8> [[TMP1]], <8 x i8> [[SHUFFLE]] +; CHECK-NEXT: [[TMP0:%.*]] = load <8 x i8>, ptr [[PTR:%.*]], align 1 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt <8 x i8> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x i8> poison, i8 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <8 x i8> [[TMP2]], <8 x i8> poison, <8 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <8 x i1> [[TMP1]], <8 x i8> [[TMP0]], <8 x i8> [[TMP3]] ; CHECK-NEXT: store <8 x i8> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: [[GEP_8:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 8 ; CHECK-NEXT: [[L_8:%.*]] = load i8, ptr [[GEP_8]], align 1 ; CHECK-NEXT: [[CMP_8:%.*]] = icmp ugt i8 [[L_8]], -1 -; CHECK-NEXT: [[TMP6:%.*]] = extractelement <8 x i8> [[TMP1]], i32 0 -; CHECK-NEXT: [[S_8:%.*]] = select i1 [[CMP_8]], i8 [[TMP6]], i8 [[X]] -; CHECK-NEXT: [[TMP7:%.*]] = extractelement <8 x i8> [[TMP4]], i32 0 -; CHECK-NEXT: store i8 [[TMP7]], ptr [[GEP_8]], align 2 +; CHECK-NEXT: [[TMP5:%.*]] = extractelement <8 x i8> [[TMP0]], i32 0 +; CHECK-NEXT: [[S_8:%.*]] = select i1 [[CMP_8]], i8 [[TMP5]], i8 [[X]] +; CHECK-NEXT: [[TMP6:%.*]] = extractelement <8 x i8> [[TMP4]], i32 0 +; CHECK-NEXT: store i8 [[TMP6]], ptr [[GEP_8]], align 2 ; CHECK-NEXT: [[GEP_9:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 9 -; CHECK-NEXT: [[L_9:%.*]] = load i8, ptr [[GEP_9]], align 1 -; CHECK-NEXT: [[CMP_9:%.*]] = icmp ugt i8 [[L_9]], -1 -; CHECK-NEXT: [[S_9:%.*]] = select i1 [[CMP_9]], i8 [[L_9]], i8 [[X]] -; CHECK-NEXT: store i8 [[S_9]], ptr [[GEP_9]], align 2 -; CHECK-NEXT: [[GEP_10:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 10 -; CHECK-NEXT: [[L_10:%.*]] = load i8, ptr [[GEP_10]], align 1 -; CHECK-NEXT: [[CMP_10:%.*]] = icmp ugt i8 [[L_10]], -1 -; CHECK-NEXT: [[S_10:%.*]] = select i1 [[CMP_10]], i8 [[L_10]], i8 [[X]] -; CHECK-NEXT: store i8 [[S_10]], ptr [[GEP_10]], align 2 -; CHECK-NEXT: [[GEP_11:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 11 -; CHECK-NEXT: [[L_11:%.*]] = load i8, ptr [[GEP_11]], align 1 -; CHECK-NEXT: [[CMP_11:%.*]] = icmp ugt i8 [[L_11]], -1 -; CHECK-NEXT: [[S_11:%.*]] = select i1 [[CMP_11]], i8 [[L_11]], i8 [[X]] -; CHECK-NEXT: store i8 [[S_11]], ptr [[GEP_11]], align 2 -; CHECK-NEXT: [[GEP_12:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 12 -; CHECK-NEXT: [[L_12:%.*]] = load i8, ptr [[GEP_12]], align 1 -; CHECK-NEXT: [[CMP_12:%.*]] = icmp ugt i8 [[L_12]], -1 -; CHECK-NEXT: [[S_12:%.*]] = select i1 [[CMP_12]], i8 [[L_12]], i8 [[X]] -; CHECK-NEXT: store i8 [[S_12]], ptr [[GEP_12]], align 2 +; CHECK-NEXT: [[TMP7:%.*]] = load <4 x i8>, ptr [[GEP_9]], align 1 +; CHECK-NEXT: [[TMP8:%.*]] = icmp ugt <4 x i8> [[TMP7]], +; CHECK-NEXT: [[TMP9:%.*]] = insertelement <4 x i8> poison, i8 [[X]], i32 0 +; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <4 x i8> [[TMP9]], <4 x i8> poison, <4 x i32> zeroinitializer +; CHECK-NEXT: [[TMP11:%.*]] = select <4 x i1> [[TMP8]], <4 x i8> [[TMP7]], <4 x i8> [[TMP10]] +; CHECK-NEXT: store <4 x i8> [[TMP11]], ptr [[GEP_9]], align 2 ; CHECK-NEXT: [[GEP_13:%.*]] = getelementptr inbounds i8, ptr [[PTR]], i8 13 ; CHECK-NEXT: [[L_13:%.*]] = load i8, ptr [[GEP_13]], align 1 ; CHECK-NEXT: [[CMP_13:%.*]] = icmp ugt i8 [[L_13]], -1 @@ -407,11 +381,11 @@ entry: define void @select_uniform_ugt_4xi16(ptr %ptr, i16 %x) { ; CHECK-LABEL: @select_uniform_ugt_4xi16( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i16>, ptr [[PTR:%.*]], align 2 -; CHECK-NEXT: [[TMP2:%.*]] = icmp ugt <4 x i16> [[TMP1]], -; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x i16> poison, i16 [[X:%.*]], i32 0 -; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <4 x i16> [[TMP3]], <4 x i16> poison, <4 x i32> zeroinitializer -; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> [[TMP2]], <4 x i16> [[TMP1]], <4 x i16> [[SHUFFLE]] +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i16>, ptr [[PTR:%.*]], align 2 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt <4 x i16> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i16> poison, i16 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x i16> [[TMP2]], <4 x i16> poison, <4 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> [[TMP1]], <4 x i16> [[TMP0]], <4 x i16> [[TMP3]] ; CHECK-NEXT: store <4 x i16> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: ret void ; @@ -445,11 +419,11 @@ entry: define void @select_uniform_ult_8xi16(ptr %ptr, i16 %x) { ; CHECK-LABEL: @select_uniform_ult_8xi16( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP1:%.*]] = load <8 x i16>, ptr [[PTR:%.*]], align 2 -; CHECK-NEXT: [[TMP2:%.*]] = icmp ult <8 x i16> [[TMP1]], -; CHECK-NEXT: [[TMP3:%.*]] = insertelement <8 x i16> poison, i16 [[X:%.*]], i32 0 -; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <8 x i16> [[TMP3]], <8 x i16> poison, <8 x i32> zeroinitializer -; CHECK-NEXT: [[TMP4:%.*]] = select <8 x i1> [[TMP2]], <8 x i16> [[TMP1]], <8 x i16> [[SHUFFLE]] +; CHECK-NEXT: [[TMP0:%.*]] = load <8 x i16>, ptr [[PTR:%.*]], align 2 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ult <8 x i16> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <8 x i16> poison, i16 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <8 x i16> [[TMP2]], <8 x i16> poison, <8 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <8 x i1> [[TMP1]], <8 x i16> [[TMP0]], <8 x i16> [[TMP3]] ; CHECK-NEXT: store <8 x i16> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: ret void ; @@ -506,11 +480,11 @@ entry: define void @select_uniform_eq_2xi32(ptr %ptr, i32 %x) { ; CHECK-LABEL: @select_uniform_eq_2xi32( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP1:%.*]] = load <2 x i32>, ptr [[PTR:%.*]], align 4 -; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <2 x i32> [[TMP1]], -; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x i32> poison, i32 [[X:%.*]], i32 0 -; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <2 x i32> [[TMP3]], <2 x i32> poison, <2 x i32> zeroinitializer -; CHECK-NEXT: [[TMP4:%.*]] = select <2 x i1> [[TMP2]], <2 x i32> [[TMP1]], <2 x i32> [[SHUFFLE]] +; CHECK-NEXT: [[TMP0:%.*]] = load <2 x i32>, ptr [[PTR:%.*]], align 4 +; CHECK-NEXT: [[TMP1:%.*]] = icmp eq <2 x i32> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <2 x i32> poison, i32 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <2 x i32> [[TMP2]], <2 x i32> poison, <2 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <2 x i1> [[TMP1]], <2 x i32> [[TMP0]], <2 x i32> [[TMP3]] ; CHECK-NEXT: store <2 x i32> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: ret void ; @@ -532,11 +506,11 @@ entry: define void @select_uniform_eq_4xi32(ptr %ptr, i32 %x) { ; CHECK-LABEL: @select_uniform_eq_4xi32( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i32>, ptr [[PTR:%.*]], align 4 -; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <4 x i32> [[TMP1]], -; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x i32> poison, i32 [[X:%.*]], i32 0 -; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <4 x i32> [[TMP3]], <4 x i32> poison, <4 x i32> zeroinitializer -; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> [[TMP2]], <4 x i32> [[TMP1]], <4 x i32> [[SHUFFLE]] +; CHECK-NEXT: [[TMP0:%.*]] = load <4 x i32>, ptr [[PTR:%.*]], align 4 +; CHECK-NEXT: [[TMP1:%.*]] = icmp eq <4 x i32> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> poison, i32 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> poison, <4 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <4 x i1> [[TMP1]], <4 x i32> [[TMP0]], <4 x i32> [[TMP3]] ; CHECK-NEXT: store <4 x i32> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: ret void ; @@ -569,11 +543,11 @@ entry: define void @select_uniform_ne_2xi64(ptr %ptr, i64 %x) { ; CHECK-LABEL: @select_uniform_ne_2xi64( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[TMP1:%.*]] = load <2 x i64>, ptr [[PTR:%.*]], align 8 -; CHECK-NEXT: [[TMP2:%.*]] = icmp ne <2 x i64> [[TMP1]], -; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x i64> poison, i64 [[X:%.*]], i32 0 -; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <2 x i64> [[TMP3]], <2 x i64> poison, <2 x i32> zeroinitializer -; CHECK-NEXT: [[TMP4:%.*]] = select <2 x i1> [[TMP2]], <2 x i64> [[TMP1]], <2 x i64> [[SHUFFLE]] +; CHECK-NEXT: [[TMP0:%.*]] = load <2 x i64>, ptr [[PTR:%.*]], align 8 +; CHECK-NEXT: [[TMP1:%.*]] = icmp ne <2 x i64> [[TMP0]], +; CHECK-NEXT: [[TMP2:%.*]] = insertelement <2 x i64> poison, i64 [[X:%.*]], i32 0 +; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <2 x i64> [[TMP2]], <2 x i64> poison, <2 x i32> zeroinitializer +; CHECK-NEXT: [[TMP4:%.*]] = select <2 x i1> [[TMP1]], <2 x i64> [[TMP0]], <2 x i64> [[TMP3]] ; CHECK-NEXT: store <2 x i64> [[TMP4]], ptr [[PTR]], align 2 ; CHECK-NEXT: ret void ; -- 2.7.4