diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/masked_gather_scatter.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/masked_gather_scatter.ll | 426 |
1 files changed, 310 insertions, 116 deletions
diff --git a/llvm/test/CodeGen/X86/masked_gather_scatter.ll b/llvm/test/CodeGen/X86/masked_gather_scatter.ll index 85cfc6234fb..61dc7c7724e 100644 --- a/llvm/test/CodeGen/X86/masked_gather_scatter.ll +++ b/llvm/test/CodeGen/X86/masked_gather_scatter.ll @@ -1,3 +1,4 @@ +; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py ; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f < %s | FileCheck %s --check-prefix=KNL_64 ; RUN: llc -mtriple=i386-unknown-linux-gnu -mattr=+avx512f < %s | FileCheck %s --check-prefix=KNL_32 ; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512vl -mattr=+avx512dq < %s | FileCheck %s --check-prefix=SKX @@ -38,6 +39,14 @@ define <16 x float> @test1(float* %base, <16 x i32> %ind) { ; SKX-NEXT: vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1} ; SKX-NEXT: vmovaps %zmm1, %zmm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test1: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: kxnorw %k0, %k0, %k1 +; SKX_32-NEXT: vgatherdps (%eax,%zmm0,4), %zmm1 {%k1} +; SKX_32-NEXT: vmovaps %zmm1, %zmm0 +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0 %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer @@ -87,6 +96,14 @@ define <16 x float> @test2(float* %base, <16 x i32> %ind, i16 %mask) { ; SKX-NEXT: vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1} ; SKX-NEXT: vmovaps %zmm1, %zmm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test2: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: kmovw {{[0-9]+}}(%esp), %k1 +; SKX_32-NEXT: vgatherdps (%eax,%zmm0,4), %zmm1 {%k1} +; SKX_32-NEXT: vmovaps %zmm1, %zmm0 +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0 %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer @@ -120,6 +137,14 @@ define <16 x i32> @test3(i32* %base, <16 x i32> %ind, i16 %mask) { ; SKX-NEXT: vpgatherdd (%rdi,%zmm0,4), %zmm1 {%k1} ; SKX-NEXT: vmovdqa64 %zmm1, %zmm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test3: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: kmovw {{[0-9]+}}(%esp), %k1 +; SKX_32-NEXT: vpgatherdd (%eax,%zmm0,4), %zmm1 {%k1} +; SKX_32-NEXT: vmovdqa64 %zmm1, %zmm0 +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <16 x i32*> undef, i32* %base, i32 0 %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> undef, <16 x i32> zeroinitializer @@ -163,6 +188,17 @@ define <16 x i32> @test4(i32* %base, <16 x i32> %ind, i16 %mask) { ; SKX-NEXT: vpgatherdd (%rdi,%zmm0,4), %zmm2 {%k1} ; SKX-NEXT: vpaddd %zmm2, %zmm1, %zmm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test4: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: kmovw {{[0-9]+}}(%esp), %k1 +; SKX_32-NEXT: kmovw %k1, %k2 +; SKX_32-NEXT: vpgatherdd (%eax,%zmm0,4), %zmm1 {%k2} +; SKX_32-NEXT: vmovdqa64 %zmm1, %zmm2 +; SKX_32-NEXT: vpgatherdd (%eax,%zmm0,4), %zmm2 {%k1} +; SKX_32-NEXT: vpaddd %zmm2, %zmm1, %zmm0 +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <16 x i32*> undef, i32* %base, i32 0 %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> undef, <16 x i32> zeroinitializer @@ -215,6 +251,15 @@ define void @test5(i32* %base, <16 x i32> %ind, i16 %mask, <16 x i32>%val) { ; SKX-NEXT: vpscatterdd %zmm1, (%rdi,%zmm0,4) {%k2} ; SKX-NEXT: vpscatterdd %zmm1, (%rdi,%zmm0,4) {%k1} ; SKX-NEXT: retq +; +; SKX_32-LABEL: test5: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: kmovw {{[0-9]+}}(%esp), %k1 +; SKX_32-NEXT: kmovw %k1, %k2 +; SKX_32-NEXT: vpscatterdd %zmm1, (%eax,%zmm0,4) {%k2} +; SKX_32-NEXT: vpscatterdd %zmm1, (%eax,%zmm0,4) {%k1} +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <16 x i32*> undef, i32* %base, i32 0 %broadcast.splat = shufflevector <16 x i32*> %broadcast.splatinsert, <16 x i32*> undef, <16 x i32> zeroinitializer @@ -267,6 +312,15 @@ define <8 x i32> @test6(<8 x i32>%a1, <8 x i32*> %ptr) { ; SKX-NEXT: vpscatterqd %ymm0, (,%zmm1) {%k1} ; SKX-NEXT: vmovdqa64 %ymm2, %ymm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test6: +; SKX_32: # BB#0: +; SKX_32-NEXT: kxnorw %k0, %k0, %k1 +; SKX_32-NEXT: kxnorw %k0, %k0, %k2 +; SKX_32-NEXT: vpgatherdd (,%ymm1), %ymm2 {%k2} +; SKX_32-NEXT: vpscatterdd %ymm0, (,%ymm1) {%k1} +; SKX_32-NEXT: vmovdqa64 %ymm2, %ymm0 +; SKX_32-NEXT: retl %a = call <8 x i32> @llvm.masked.gather.v8i32(<8 x i32*> %ptr, i32 4, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <8 x i32> undef) @@ -309,6 +363,17 @@ define <8 x i32> @test7(i32* %base, <8 x i32> %ind, i8 %mask) { ; SKX-NEXT: vpgatherdd (%rdi,%ymm0,4), %ymm2 {%k1} ; SKX-NEXT: vpaddd %ymm2, %ymm1, %ymm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test7: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: kmovb {{[0-9]+}}(%esp), %k1 +; SKX_32-NEXT: kmovw %k1, %k2 +; SKX_32-NEXT: vpgatherdd (%eax,%ymm0,4), %ymm1 {%k2} +; SKX_32-NEXT: vmovdqa64 %ymm1, %ymm2 +; SKX_32-NEXT: vpgatherdd (%eax,%ymm0,4), %ymm2 {%k1} +; SKX_32-NEXT: vpaddd %ymm2, %ymm1, %ymm0 +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <8 x i32*> undef, i32* %base, i32 0 %broadcast.splat = shufflevector <8 x i32*> %broadcast.splatinsert, <8 x i32*> undef, <8 x i32> zeroinitializer @@ -415,13 +480,13 @@ define <8 x i32> @test9(%struct.ST* %base, <8 x i64> %ind1, <8 x i32>%ind5) { ; KNL_32-LABEL: test9: ; KNL_32: # BB#0: # %entry ; KNL_32-NEXT: vpbroadcastd {{[0-9]+}}(%esp), %ymm2 -; KNL_32-NEXT: vpbroadcastd .LCPI8_0, %ymm3 +; KNL_32-NEXT: vpbroadcastd {{\.LCPI.*}}, %ymm3 ; KNL_32-NEXT: vpmulld %ymm3, %ymm1, %ymm1 ; KNL_32-NEXT: vpmovqd %zmm0, %ymm0 -; KNL_32-NEXT: vpbroadcastd .LCPI8_1, %ymm3 +; KNL_32-NEXT: vpbroadcastd {{\.LCPI.*}}, %ymm3 ; KNL_32-NEXT: vpmulld %ymm3, %ymm0, %ymm0 ; KNL_32-NEXT: vpaddd %ymm1, %ymm0, %ymm0 -; KNL_32-NEXT: vpbroadcastd .LCPI8_2, %ymm1 +; KNL_32-NEXT: vpbroadcastd {{\.LCPI.*}}, %ymm1 ; KNL_32-NEXT: vpaddd %ymm1, %ymm0, %ymm0 ; KNL_32-NEXT: vpaddd %ymm0, %ymm2, %ymm0 ; KNL_32-NEXT: vpmovsxdq %ymm0, %zmm1 @@ -441,6 +506,18 @@ define <8 x i32> @test9(%struct.ST* %base, <8 x i64> %ind1, <8 x i32>%ind5) { ; SKX-NEXT: kxnorw %k0, %k0, %k1 ; SKX-NEXT: vpgatherqd (,%zmm1), %ymm0 {%k1} ; SKX-NEXT: retq +; +; SKX_32-LABEL: test9: +; SKX_32: # BB#0: # %entry +; SKX_32-NEXT: vpmulld {{\.LCPI.*}}{1to8}, %ymm1, %ymm1 +; SKX_32-NEXT: vpmovqd %zmm0, %ymm0 +; SKX_32-NEXT: vpmulld {{\.LCPI.*}}{1to8}, %ymm0, %ymm0 +; SKX_32-NEXT: vpaddd {{[0-9]+}}(%esp){1to8}, %ymm0, %ymm0 +; SKX_32-NEXT: vpaddd %ymm1, %ymm0, %ymm0 +; SKX_32-NEXT: vpaddd {{\.LCPI.*}}{1to8}, %ymm0, %ymm1 +; SKX_32-NEXT: kxnorw %k0, %k0, %k1 +; SKX_32-NEXT: vpgatherdd (,%ymm1), %ymm0 {%k1} +; SKX_32-NEXT: retl entry: %broadcast.splatinsert = insertelement <8 x %struct.ST*> undef, %struct.ST* %base, i32 0 %broadcast.splat = shufflevector <8 x %struct.ST*> %broadcast.splatinsert, <8 x %struct.ST*> undef, <8 x i32> zeroinitializer @@ -477,13 +554,13 @@ define <8 x i32> @test10(%struct.ST* %base, <8 x i64> %i1, <8 x i32>%ind5) { ; KNL_32-LABEL: test10: ; KNL_32: # BB#0: # %entry ; KNL_32-NEXT: vpbroadcastd {{[0-9]+}}(%esp), %ymm2 -; KNL_32-NEXT: vpbroadcastd .LCPI9_0, %ymm3 +; KNL_32-NEXT: vpbroadcastd {{\.LCPI.*}}, %ymm3 ; KNL_32-NEXT: vpmulld %ymm3, %ymm1, %ymm1 ; KNL_32-NEXT: vpmovqd %zmm0, %ymm0 -; KNL_32-NEXT: vpbroadcastd .LCPI9_1, %ymm3 +; KNL_32-NEXT: vpbroadcastd {{\.LCPI.*}}, %ymm3 ; KNL_32-NEXT: vpmulld %ymm3, %ymm0, %ymm0 ; KNL_32-NEXT: vpaddd %ymm1, %ymm0, %ymm0 -; KNL_32-NEXT: vpbroadcastd .LCPI9_2, %ymm1 +; KNL_32-NEXT: vpbroadcastd {{\.LCPI.*}}, %ymm1 ; KNL_32-NEXT: vpaddd %ymm1, %ymm0, %ymm0 ; KNL_32-NEXT: vpaddd %ymm0, %ymm2, %ymm0 ; KNL_32-NEXT: vpmovsxdq %ymm0, %zmm1 @@ -503,6 +580,18 @@ define <8 x i32> @test10(%struct.ST* %base, <8 x i64> %i1, <8 x i32>%ind5) { ; SKX-NEXT: kxnorw %k0, %k0, %k1 ; SKX-NEXT: vpgatherqd (,%zmm1), %ymm0 {%k1} ; SKX-NEXT: retq +; +; SKX_32-LABEL: test10: +; SKX_32: # BB#0: # %entry +; SKX_32-NEXT: vpmulld {{\.LCPI.*}}{1to8}, %ymm1, %ymm1 +; SKX_32-NEXT: vpmovqd %zmm0, %ymm0 +; SKX_32-NEXT: vpmulld {{\.LCPI.*}}{1to8}, %ymm0, %ymm0 +; SKX_32-NEXT: vpaddd {{[0-9]+}}(%esp){1to8}, %ymm0, %ymm0 +; SKX_32-NEXT: vpaddd %ymm1, %ymm0, %ymm0 +; SKX_32-NEXT: vpaddd {{\.LCPI.*}}{1to8}, %ymm0, %ymm1 +; SKX_32-NEXT: kxnorw %k0, %k0, %k1 +; SKX_32-NEXT: vpgatherdd (,%ymm1), %ymm0 {%k1} +; SKX_32-NEXT: retl entry: %broadcast.splatinsert = insertelement <8 x %struct.ST*> undef, %struct.ST* %base, i32 0 %broadcast.splat = shufflevector <8 x %struct.ST*> %broadcast.splatinsert, <8 x %struct.ST*> undef, <8 x i32> zeroinitializer @@ -535,6 +624,14 @@ define <16 x float> @test11(float* %base, i32 %ind) { ; SKX-NEXT: kxnorw %k0, %k0, %k1 ; SKX-NEXT: vgatherdps (%rdi,%zmm1,4), %zmm0 {%k1} ; SKX-NEXT: retq +; +; SKX_32-LABEL: test11: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: vpbroadcastd {{[0-9]+}}(%esp), %zmm1 +; SKX_32-NEXT: kxnorw %k0, %k0, %k1 +; SKX_32-NEXT: vgatherdps (%eax,%zmm1,4), %zmm0 {%k1} +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0 %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer @@ -568,6 +665,14 @@ define <16 x float> @test12(float* %base, <16 x i32> %ind) { ; SKX-NEXT: vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1} ; SKX-NEXT: vmovaps %zmm1, %zmm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test12: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: kxnorw %k0, %k0, %k1 +; SKX_32-NEXT: vgatherdps (%eax,%zmm0,4), %zmm1 {%k1} +; SKX_32-NEXT: vmovaps %zmm1, %zmm0 +; SKX_32-NEXT: retl %sext_ind = sext <16 x i32> %ind to <16 x i64> %gep.random = getelementptr float, float *%base, <16 x i64> %sext_ind @@ -596,6 +701,13 @@ define <16 x float> @test13(float* %base, <16 x i32> %ind) { ; SKX-NEXT: vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1} ; SKX-NEXT: vmovaps %zmm1, %zmm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test13: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: vgatherdps (%eax,%zmm0,4), %zmm1 {%k1} +; SKX_32-NEXT: vmovaps %zmm1, %zmm0 +; SKX_32-NEXT: retl %sext_ind = sext <16 x i32> %ind to <16 x i64> %gep.random = getelementptr float, float *%base, <16 x i64> %sext_ind @@ -675,25 +787,29 @@ define <4 x float> @test15(float* %base, <4 x i32> %ind, <4 x i1> %mask) { ; ; KNL_64-LABEL: test15: ; KNL_64: # BB#0: -; KNL_64: vpxor %ymm2, %ymm2, %ymm2 +; KNL_64-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_64-NEXT: vpxor %ymm2, %ymm2, %ymm2 ; KNL_64-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5,6,7] ; KNL_64-NEXT: vpmovsxdq %ymm0, %zmm2 ; KNL_64-NEXT: vpslld $31, %ymm1, %ymm0 ; KNL_64-NEXT: vptestmd %zmm0, %zmm0, %k1 ; KNL_64-NEXT: vgatherqps (%rdi,%zmm2,4), %ymm0 {%k1} -; KNL_64-NEXT: # kill +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill> ; KNL_64-NEXT: retq ; ; KNL_32-LABEL: test15: ; KNL_32: # BB#0: -; KNL_32: vpxor %ymm2, %ymm2, %ymm2 +; KNL_32-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_32-NEXT: vpxor %ymm2, %ymm2, %ymm2 ; KNL_32-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm2[4,5,6,7] ; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax ; KNL_32-NEXT: vpmovsxdq %ymm0, %zmm2 ; KNL_32-NEXT: vpslld $31, %ymm1, %ymm0 ; KNL_32-NEXT: vptestmd %zmm0, %zmm0, %k1 ; KNL_32-NEXT: vgatherqps (%eax,%zmm2,4), %ymm0 {%k1} -; KNL_32-NEXT: # kill +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill> ; KNL_32-NEXT: retl ; ; SKX-LABEL: test15: @@ -724,7 +840,9 @@ define <4 x double> @test16(double* %base, <4 x i32> %ind, <4 x i1> %mask, <4 x ; ; KNL_64-LABEL: test16: ; KNL_64: # BB#0: -; KNL_64: vpslld $31, %xmm1, %xmm1 +; KNL_64-NEXT: # kill: %YMM2<def> %YMM2<kill> %ZMM2<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_64-NEXT: vpslld $31, %xmm1, %xmm1 ; KNL_64-NEXT: vpsrad $31, %xmm1, %xmm1 ; KNL_64-NEXT: vpmovsxdq %xmm1, %ymm1 ; KNL_64-NEXT: vpxord %zmm3, %zmm3, %zmm3 @@ -738,14 +856,16 @@ define <4 x double> @test16(double* %base, <4 x i32> %ind, <4 x i1> %mask, <4 x ; ; KNL_32-LABEL: test16: ; KNL_32: # BB#0: -; KNL_32: vpslld $31, %xmm1, %xmm1 +; KNL_32-NEXT: # kill: %YMM2<def> %YMM2<kill> %ZMM2<def> +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_32-NEXT: vpslld $31, %xmm1, %xmm1 ; KNL_32-NEXT: vpsrad $31, %xmm1, %xmm1 ; KNL_32-NEXT: vpmovsxdq %xmm1, %ymm1 ; KNL_32-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_32-NEXT: vinserti64x4 $0, %ymm1, %zmm3, %zmm1 ; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax ; KNL_32-NEXT: vpmovsxdq %ymm0, %zmm0 -; KNL_32-NEXT: vpsllvq .LCPI15_0, %zmm1, %zmm1 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm1, %zmm1 ; KNL_32-NEXT: vptestmq %zmm1, %zmm1, %k1 ; KNL_32-NEXT: vgatherqpd (%eax,%zmm0,8), %zmm2 {%k1} ; KNL_32-NEXT: vmovapd %zmm2, %zmm0 @@ -778,7 +898,9 @@ define <2 x double> @test17(double* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x ; ; KNL_64-LABEL: test17: ; KNL_64: # BB#0: -; KNL_64: vpxord %zmm3, %zmm3, %zmm3 +; KNL_64-NEXT: # kill: %XMM2<def> %XMM2<kill> %ZMM2<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_64-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_64-NEXT: vinserti32x4 $0, %xmm1, %zmm3, %zmm1 ; KNL_64-NEXT: vpsllq $63, %zmm1, %zmm1 ; KNL_64-NEXT: vptestmq %zmm1, %zmm1, %k1 @@ -788,10 +910,12 @@ define <2 x double> @test17(double* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x ; ; KNL_32-LABEL: test17: ; KNL_32: # BB#0: -; KNL_32: vpxord %zmm3, %zmm3, %zmm3 +; KNL_32-NEXT: # kill: %XMM2<def> %XMM2<kill> %ZMM2<def> +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_32-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_32-NEXT: vinserti32x4 $0, %xmm1, %zmm3, %zmm1 ; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax -; KNL_32-NEXT: vpsllvq .LCPI16_0, %zmm1, %zmm1 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm1, %zmm1 ; KNL_32-NEXT: vptestmq %zmm1, %zmm1, %k1 ; KNL_32-NEXT: vgatherqpd (%eax,%zmm0,8), %zmm2 {%k1} ; KNL_32-NEXT: vmovapd %zmm2, %zmm0 @@ -830,7 +954,10 @@ define void @test18(<4 x i32>%a1, <4 x i32*> %ptr, <4 x i1>%mask) { ; ; KNL_64-LABEL: test18: ; KNL_64: # BB#0: -; KNL_64: vpxor %ymm3, %ymm3, %ymm3 +; KNL_64-NEXT: # kill: %XMM2<def> %XMM2<kill> %YMM2<def> +; KNL_64-NEXT: # kill: %YMM1<def> %YMM1<kill> %ZMM1<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_64-NEXT: vpxor %ymm3, %ymm3, %ymm3 ; KNL_64-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7] ; KNL_64-NEXT: vpslld $31, %ymm2, %ymm2 ; KNL_64-NEXT: vptestmd %zmm2, %zmm2, %k1 @@ -839,7 +966,10 @@ define void @test18(<4 x i32>%a1, <4 x i32*> %ptr, <4 x i1>%mask) { ; ; KNL_32-LABEL: test18: ; KNL_32: # BB#0: -; KNL_32: vpxor %ymm3, %ymm3, %ymm3 +; KNL_32-NEXT: # kill: %XMM2<def> %XMM2<kill> %YMM2<def> +; KNL_32-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_32-NEXT: vpxor %ymm3, %ymm3, %ymm3 ; KNL_32-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7] ; KNL_32-NEXT: vpmovsxdq %ymm1, %zmm1 ; KNL_32-NEXT: vpslld $31, %ymm2, %ymm2 @@ -868,7 +998,9 @@ define void @test19(<4 x double>%a1, double* %ptr, <4 x i1>%mask, <4 x i64> %ind ; ; KNL_64-LABEL: test19: ; KNL_64: # BB#0: -; KNL_64: vpslld $31, %xmm1, %xmm1 +; KNL_64-NEXT: # kill: %YMM2<def> %YMM2<kill> %ZMM2<def> +; KNL_64-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def> +; KNL_64-NEXT: vpslld $31, %xmm1, %xmm1 ; KNL_64-NEXT: vpsrad $31, %xmm1, %xmm1 ; KNL_64-NEXT: vpmovsxdq %xmm1, %ymm1 ; KNL_64-NEXT: vpxord %zmm3, %zmm3, %zmm3 @@ -880,13 +1012,15 @@ define void @test19(<4 x double>%a1, double* %ptr, <4 x i1>%mask, <4 x i64> %ind ; ; KNL_32-LABEL: test19: ; KNL_32: # BB#0: -; KNL_32: vpslld $31, %xmm1, %xmm1 +; KNL_32-NEXT: # kill: %YMM2<def> %YMM2<kill> %ZMM2<def> +; KNL_32-NEXT: # kill: %YMM0<def> %YMM0<kill> %ZMM0<def> +; KNL_32-NEXT: vpslld $31, %xmm1, %xmm1 ; KNL_32-NEXT: vpsrad $31, %xmm1, %xmm1 ; KNL_32-NEXT: vpmovsxdq %xmm1, %ymm1 ; KNL_32-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_32-NEXT: vinserti64x4 $0, %ymm1, %zmm3, %zmm1 ; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax -; KNL_32-NEXT: vpsllvq .LCPI18_0, %zmm1, %zmm1 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm1, %zmm1 ; KNL_32-NEXT: vptestmq %zmm1, %zmm1, %k1 ; KNL_32-NEXT: vscatterqpd %zmm0, (%eax,%zmm2,8) {%k1} ; KNL_32-NEXT: retl @@ -915,7 +1049,9 @@ define void @test20(<2 x float>%a1, <2 x float*> %ptr, <2 x i1> %mask) { ; ; KNL_64-LABEL: test20: ; KNL_64: # BB#0: -; KNL_64: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3] +; KNL_64-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_64-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3] ; KNL_64-NEXT: vmovq {{.*#+}} xmm2 = xmm2[0],zero ; KNL_64-NEXT: vpxor %ymm3, %ymm3, %ymm3 ; KNL_64-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7] @@ -926,7 +1062,8 @@ define void @test20(<2 x float>%a1, <2 x float*> %ptr, <2 x i1> %mask) { ; ; KNL_32-LABEL: test20: ; KNL_32: # BB#0: -; KNL_32: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3] +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<def> +; KNL_32-NEXT: vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3] ; KNL_32-NEXT: vmovq {{.*#+}} xmm2 = xmm2[0],zero ; KNL_32-NEXT: vpxor %ymm3, %ymm3, %ymm3 ; KNL_32-NEXT: vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3],ymm3[4,5,6,7] @@ -939,7 +1076,8 @@ define void @test20(<2 x float>%a1, <2 x float*> %ptr, <2 x i1> %mask) { ; ; SKX-LABEL: test20: ; SKX: # BB#0: -; SKX: vpsllq $63, %xmm2, %xmm2 +; SKX-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; SKX-NEXT: vpsllq $63, %xmm2, %xmm2 ; SKX-NEXT: vptestmq %xmm2, %xmm2, %k0 ; SKX-NEXT: kshiftlb $6, %k0, %k0 ; SKX-NEXT: kshiftrb $6, %k0, %k1 @@ -964,7 +1102,8 @@ define void @test21(<2 x i32>%a1, <2 x i32*> %ptr, <2 x i1>%mask) { ; ; KNL_64-LABEL: test21: ; KNL_64: # BB#0: -; KNL_64: vpxord %zmm3, %zmm3, %zmm3 +; KNL_64-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def> +; KNL_64-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_64-NEXT: vinserti32x4 $0, %xmm2, %zmm3, %zmm2 ; KNL_64-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; KNL_64-NEXT: vpsllq $63, %zmm2, %zmm2 @@ -974,17 +1113,19 @@ define void @test21(<2 x i32>%a1, <2 x i32*> %ptr, <2 x i1>%mask) { ; ; KNL_32-LABEL: test21: ; KNL_32: # BB#0: -; KNL_32: vpxord %zmm3, %zmm3, %zmm3 +; KNL_32-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def> +; KNL_32-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_32-NEXT: vinserti32x4 $0, %xmm2, %zmm3, %zmm2 ; KNL_32-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] -; KNL_32-NEXT: vpsllvq .LCPI20_0, %zmm2, %zmm2 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm2, %zmm2 ; KNL_32-NEXT: vptestmq %zmm2, %zmm2, %k1 ; KNL_32-NEXT: vpscatterqd %ymm0, (,%zmm1) {%k1} ; KNL_32-NEXT: retl ; ; SKX-LABEL: test21: ; SKX: # BB#0: -; SKX: vpsllq $63, %xmm2, %xmm2 +; SKX-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; SKX-NEXT: vpsllq $63, %xmm2, %xmm2 ; SKX-NEXT: vptestmq %xmm2, %xmm2, %k0 ; SKX-NEXT: kshiftlb $6, %k0, %k0 ; SKX-NEXT: kshiftrb $6, %k0, %k1 @@ -994,7 +1135,8 @@ define void @test21(<2 x i32>%a1, <2 x i32*> %ptr, <2 x i1>%mask) { ; ; SKX_32-LABEL: test21: ; SKX_32: # BB#0: -; SKX_32: vpsllq $63, %xmm2, %xmm2 +; SKX_32-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; SKX_32-NEXT: vpsllq $63, %xmm2, %xmm2 ; SKX_32-NEXT: vptestmq %xmm2, %xmm2, %k0 ; SKX_32-NEXT: kshiftlb $6, %k0, %k0 ; SKX_32-NEXT: kshiftrb $6, %k0, %k1 @@ -1013,7 +1155,8 @@ define <2 x float> @test22(float* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x fl ; ; KNL_64-LABEL: test22: ; KNL_64: # BB#0: -; KNL_64: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] +; KNL_64-NEXT: # kill: %XMM2<def> %XMM2<kill> %YMM2<def> +; KNL_64-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] ; KNL_64-NEXT: vmovq {{.*#+}} xmm1 = xmm1[0],zero ; KNL_64-NEXT: vpxor %ymm3, %ymm3, %ymm3 ; KNL_64-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7] @@ -1027,7 +1170,8 @@ define <2 x float> @test22(float* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x fl ; ; KNL_32-LABEL: test22: ; KNL_32: # BB#0: -; KNL_32: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] +; KNL_32-NEXT: # kill: %XMM2<def> %XMM2<kill> %YMM2<def> +; KNL_32-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] ; KNL_32-NEXT: vmovq {{.*#+}} xmm1 = xmm1[0],zero ; KNL_32-NEXT: vpxor %ymm3, %ymm3, %ymm3 ; KNL_32-NEXT: vpblendd {{.*#+}} ymm1 = ymm1[0,1,2,3],ymm3[4,5,6,7] @@ -1075,7 +1219,9 @@ define <2 x i32> @test23(i32* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x i32> % ; ; KNL_64-LABEL: test23: ; KNL_64: # BB#0: -; KNL_64: vpxord %zmm3, %zmm3, %zmm3 +; KNL_64-NEXT: # kill: %XMM2<def> %XMM2<kill> %ZMM2<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_64-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_64-NEXT: vinserti32x4 $0, %xmm1, %zmm3, %zmm1 ; KNL_64-NEXT: vpsllq $63, %zmm1, %zmm1 ; KNL_64-NEXT: vptestmq %zmm1, %zmm1, %k1 @@ -1085,10 +1231,12 @@ define <2 x i32> @test23(i32* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x i32> % ; ; KNL_32-LABEL: test23: ; KNL_32: # BB#0: -; KNL_32: vpxord %zmm3, %zmm3, %zmm3 +; KNL_32-NEXT: # kill: %XMM2<def> %XMM2<kill> %ZMM2<def> +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_32-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_32-NEXT: vinserti32x4 $0, %xmm1, %zmm3, %zmm1 ; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax -; KNL_32-NEXT: vpsllvq .LCPI22_0, %zmm1, %zmm1 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm1, %zmm1 ; KNL_32-NEXT: vptestmq %zmm1, %zmm1, %k1 ; KNL_32-NEXT: vpgatherqq (%eax,%zmm0,8), %zmm2 {%k1} ; KNL_32-NEXT: vmovdqa64 %zmm2, %zmm0 @@ -1119,7 +1267,8 @@ define <2 x i32> @test23(i32* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x i32> % define <2 x i32> @test24(i32* %base, <2 x i32> %ind) { ; KNL_64-LABEL: test24: ; KNL_64: # BB#0: -; KNL_64: movb $3, %al +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_64-NEXT: movb $3, %al ; KNL_64-NEXT: kmovw %eax, %k1 ; KNL_64-NEXT: vpgatherqq (%rdi,%zmm0,8), %zmm1 {%k1} ; KNL_64-NEXT: vmovdqa64 %zmm1, %zmm0 @@ -1127,10 +1276,11 @@ define <2 x i32> @test24(i32* %base, <2 x i32> %ind) { ; ; KNL_32-LABEL: test24: ; KNL_32: # BB#0: -; KNL_32: movl {{[0-9]+}}(%esp), %eax +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax ; KNL_32-NEXT: vpxord %zmm1, %zmm1, %zmm1 -; KNL_32-NEXT: vinserti32x4 $0, .LCPI23_0, %zmm1, %zmm1 -; KNL_32-NEXT: vpsllvq .LCPI23_1, %zmm1, %zmm1 +; KNL_32-NEXT: vinserti32x4 $0, {{\.LCPI.*}}, %zmm1, %zmm1 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm1, %zmm1 ; KNL_32-NEXT: vptestmq %zmm1, %zmm1, %k1 ; KNL_32-NEXT: vpgatherqq (%eax,%zmm0,8), %zmm1 {%k1} ; KNL_32-NEXT: vmovdqa64 %zmm1, %zmm0 @@ -1160,7 +1310,9 @@ define <2 x i64> @test25(i64* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x i64> % ; ; KNL_64-LABEL: test25: ; KNL_64: # BB#0: -; KNL_64: vpxord %zmm3, %zmm3, %zmm3 +; KNL_64-NEXT: # kill: %XMM2<def> %XMM2<kill> %ZMM2<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_64-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_64-NEXT: vinserti32x4 $0, %xmm1, %zmm3, %zmm1 ; KNL_64-NEXT: vpsllq $63, %zmm1, %zmm1 ; KNL_64-NEXT: vptestmq %zmm1, %zmm1, %k1 @@ -1170,10 +1322,12 @@ define <2 x i64> @test25(i64* %base, <2 x i32> %ind, <2 x i1> %mask, <2 x i64> % ; ; KNL_32-LABEL: test25: ; KNL_32: # BB#0: -; KNL_32: vpxord %zmm3, %zmm3, %zmm3 +; KNL_32-NEXT: # kill: %XMM2<def> %XMM2<kill> %ZMM2<def> +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_32-NEXT: vpxord %zmm3, %zmm3, %zmm3 ; KNL_32-NEXT: vinserti32x4 $0, %xmm1, %zmm3, %zmm1 ; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax -; KNL_32-NEXT: vpsllvq .LCPI24_0, %zmm1, %zmm1 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm1, %zmm1 ; KNL_32-NEXT: vptestmq %zmm1, %zmm1, %k1 ; KNL_32-NEXT: vpgatherqq (%eax,%zmm0,8), %zmm2 {%k1} ; KNL_32-NEXT: vmovdqa64 %zmm2, %zmm0 @@ -1205,7 +1359,9 @@ define <2 x i64> @test26(i64* %base, <2 x i32> %ind, <2 x i64> %src0) { ; ; KNL_64-LABEL: test26: ; KNL_64: # BB#0: -; KNL_64: movb $3, %al +; KNL_64-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def> +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_64-NEXT: movb $3, %al ; KNL_64-NEXT: kmovw %eax, %k1 ; KNL_64-NEXT: vpgatherqq (%rdi,%zmm0,8), %zmm1 {%k1} ; KNL_64-NEXT: vmovdqa64 %zmm1, %zmm0 @@ -1213,10 +1369,12 @@ define <2 x i64> @test26(i64* %base, <2 x i32> %ind, <2 x i64> %src0) { ; ; KNL_32-LABEL: test26: ; KNL_32: # BB#0: -; KNL_32: movl {{[0-9]+}}(%esp), %eax +; KNL_32-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def> +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %ZMM0<def> +; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax ; KNL_32-NEXT: vpxord %zmm2, %zmm2, %zmm2 -; KNL_32-NEXT: vinserti32x4 $0, .LCPI25_0, %zmm2, %zmm2 -; KNL_32-NEXT: vpsllvq .LCPI25_1, %zmm2, %zmm2 +; KNL_32-NEXT: vinserti32x4 $0, {{\.LCPI.*}}, %zmm2, %zmm2 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm2, %zmm2 ; KNL_32-NEXT: vptestmq %zmm2, %zmm2, %k1 ; KNL_32-NEXT: vpgatherqq (%eax,%zmm0,8), %zmm1 {%k1} ; KNL_32-NEXT: vmovdqa64 %zmm1, %zmm0 @@ -1252,7 +1410,7 @@ define <2 x float> @test27(float* %base, <2 x i32> %ind) { ; KNL_64-NEXT: movb $3, %al ; KNL_64-NEXT: kmovw %eax, %k1 ; KNL_64-NEXT: vgatherqps (%rdi,%zmm1,4), %ymm0 {%k1} -; KNL_64-NEXT: # kill +; KNL_64-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill> ; KNL_64-NEXT: retq ; ; KNL_32-LABEL: test27: @@ -1263,7 +1421,7 @@ define <2 x float> @test27(float* %base, <2 x i32> %ind) { ; KNL_32-NEXT: movb $3, %cl ; KNL_32-NEXT: kmovw %ecx, %k1 ; KNL_32-NEXT: vgatherqps (%eax,%zmm1,4), %ymm0 {%k1} -; KNL_32-NEXT: # kill +; KNL_32-NEXT: # kill: %XMM0<def> %XMM0<kill> %YMM0<kill> ; KNL_32-NEXT: retl ; ; SKX-LABEL: test27: @@ -1273,6 +1431,15 @@ define <2 x float> @test27(float* %base, <2 x i32> %ind) { ; SKX-NEXT: kmovb %eax, %k1 ; SKX-NEXT: vgatherdps (%rdi,%xmm1,4), %xmm0 {%k1} ; SKX-NEXT: retq +; +; SKX_32-LABEL: test27: +; SKX_32: # BB#0: +; SKX_32-NEXT: vpshufd {{.*#+}} xmm1 = xmm0[0,2,2,3] +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: movb $3, %cl +; SKX_32-NEXT: kmovb %ecx, %k1 +; SKX_32-NEXT: vgatherdps (%eax,%xmm1,4), %xmm0 {%k1} +; SKX_32-NEXT: retl %sext_ind = sext <2 x i32> %ind to <2 x i64> %gep.random = getelementptr float, float* %base, <2 x i64> %sext_ind %res = call <2 x float> @llvm.masked.gather.v2f32(<2 x float*> %gep.random, i32 4, <2 x i1> <i1 true, i1 true>, <2 x float> undef) @@ -1285,7 +1452,8 @@ define void @test28(<2 x i32>%a1, <2 x i32*> %ptr) { ; ; KNL_64-LABEL: test28: ; KNL_64: # BB#0: -; KNL_64: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] +; KNL_64-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def> +; KNL_64-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; KNL_64-NEXT: movb $3, %al ; KNL_64-NEXT: kmovw %eax, %k1 ; KNL_64-NEXT: vpscatterqd %ymm0, (,%zmm1) {%k1} @@ -1293,17 +1461,19 @@ define void @test28(<2 x i32>%a1, <2 x i32*> %ptr) { ; ; KNL_32-LABEL: test28: ; KNL_32: # BB#0: -; KNL_32: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] +; KNL_32-NEXT: # kill: %XMM1<def> %XMM1<kill> %ZMM1<def> +; KNL_32-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; KNL_32-NEXT: vpxord %zmm2, %zmm2, %zmm2 -; KNL_32-NEXT: vinserti32x4 $0, .LCPI27_0, %zmm2, %zmm2 -; KNL_32-NEXT: vpsllvq .LCPI27_1, %zmm2, %zmm2 +; KNL_32-NEXT: vinserti32x4 $0, {{\.LCPI.*}}, %zmm2, %zmm2 +; KNL_32-NEXT: vpsllvq {{\.LCPI.*}}, %zmm2, %zmm2 ; KNL_32-NEXT: vptestmq %zmm2, %zmm2, %k1 ; KNL_32-NEXT: vpscatterqd %ymm0, (,%zmm1) {%k1} ; KNL_32-NEXT: retl ; ; SKX-LABEL: test28: ; SKX: # BB#0: -; SKX: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] +; SKX-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; SKX-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; SKX-NEXT: movb $3, %al ; SKX-NEXT: kmovb %eax, %k1 ; SKX-NEXT: vpscatterqd %xmm0, (,%ymm1) {%k1} @@ -1311,7 +1481,8 @@ define void @test28(<2 x i32>%a1, <2 x i32*> %ptr) { ; ; SKX_32-LABEL: test28: ; SKX_32: # BB#0: -; SKX_32: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] +; SKX_32-NEXT: # kill: %XMM1<def> %XMM1<kill> %YMM1<def> +; SKX_32-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] ; SKX_32-NEXT: movb $3, %al ; SKX_32-NEXT: kmovb %eax, %k1 ; SKX_32-NEXT: vpscatterqd %xmm0, (,%ymm1) {%k1} @@ -1353,6 +1524,15 @@ define <16 x float> @test29(float* %base, <16 x i32> %ind) { ; SKX-NEXT: vgatherdps (%rdi,%zmm0,4), %zmm1 {%k1} ; SKX-NEXT: vmovaps %zmm1, %zmm0 ; SKX-NEXT: retq +; +; SKX_32-LABEL: test29: +; SKX_32: # BB#0: +; SKX_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; SKX_32-NEXT: movw $44, %cx +; SKX_32-NEXT: kmovw %ecx, %k1 +; SKX_32-NEXT: vgatherdps (%eax,%zmm0,4), %zmm1 {%k1} +; SKX_32-NEXT: vmovaps %zmm1, %zmm0 +; SKX_32-NEXT: retl %broadcast.splatinsert = insertelement <16 x float*> undef, float* %base, i32 0 %broadcast.splat = shufflevector <16 x float*> %broadcast.splatinsert, <16 x float*> undef, <16 x i32> zeroinitializer @@ -1370,9 +1550,12 @@ define <3 x i32> @test30(<3 x i32*> %base, <3 x i32> %ind, <3 x i1> %mask, <3 x ; KNL_64-LABEL: test30: ; KNL_64: # BB#0: ; KNL_64-NEXT: andl $1, %edx +; KNL_64-NEXT: kmovw %edx, %k1 ; KNL_64-NEXT: andl $1, %esi +; KNL_64-NEXT: kmovw %esi, %k2 ; KNL_64-NEXT: movl %edi, %eax ; KNL_64-NEXT: andl $1, %eax +; KNL_64-NEXT: kmovw %eax, %k0 ; KNL_64-NEXT: vpmovsxdq %xmm1, %ymm1 ; KNL_64-NEXT: vpsllq $2, %ymm1, %ymm1 ; KNL_64-NEXT: vpaddq %ymm1, %ymm0, %ymm1 @@ -1380,76 +1563,81 @@ define <3 x i32> @test30(<3 x i32*> %base, <3 x i32> %ind, <3 x i1> %mask, <3 x ; KNL_64-NEXT: testb $1, %dil ; KNL_64-NEXT: je .LBB29_2 ; KNL_64-NEXT: # BB#1: # %cond.load -; KNL_64-NEXT: vmovq %xmm1, %rcx +; KNL_64-NEXT: vmovq %xmm1, %rax ; KNL_64-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero ; KNL_64-NEXT: .LBB29_2: # %else -; KNL_64-NEXT: testb %sil, %sil +; KNL_64-NEXT: kmovw %k2, %eax +; KNL_64-NEXT: movl %eax, %ecx +; KNL_64-NEXT: andl $1, %ecx +; KNL_64-NEXT: testb %cl, %cl ; KNL_64-NEXT: je .LBB29_4 ; KNL_64-NEXT: # BB#3: # %cond.load1 ; KNL_64-NEXT: vpextrq $1, %xmm1, %rcx ; KNL_64-NEXT: vpinsrd $1, (%rcx), %xmm0, %xmm0 ; KNL_64-NEXT: .LBB29_4: # %else2 +; KNL_64-NEXT: kmovw %k1, %ecx +; KNL_64-NEXT: movl %ecx, %edx +; KNL_64-NEXT: andl $1, %edx ; KNL_64-NEXT: testb %dl, %dl ; KNL_64-NEXT: je .LBB29_6 ; KNL_64-NEXT: # BB#5: # %cond.load4 ; KNL_64-NEXT: vextracti128 $1, %ymm1, %xmm1 -; KNL_64-NEXT: vmovq %xmm1, %rcx -; KNL_64-NEXT: vpinsrd $2, (%rcx), %xmm0, %xmm0 +; KNL_64-NEXT: vmovq %xmm1, %rdx +; KNL_64-NEXT: vpinsrd $2, (%rdx), %xmm0, %xmm0 ; KNL_64-NEXT: .LBB29_6: # %else5 -; KNL_64-NEXT: vmovd %eax, %xmm1 -; KNL_64-NEXT: vpinsrd $1, %esi, %xmm1, %xmm1 -; KNL_64-NEXT: vpinsrd $2, %edx, %xmm1, %xmm1 +; KNL_64-NEXT: kmovw %k0, %edx +; KNL_64-NEXT: vmovd %edx, %xmm1 +; KNL_64-NEXT: vpinsrd $1, %eax, %xmm1, %xmm1 +; KNL_64-NEXT: vpinsrd $2, %ecx, %xmm1, %xmm1 ; KNL_64-NEXT: vpslld $31, %xmm1, %xmm1 ; KNL_64-NEXT: vblendvps %xmm1, %xmm0, %xmm2, %xmm0 ; KNL_64-NEXT: retq ; ; KNL_32-LABEL: test30: ; KNL_32: # BB#0: -; KNL_32-NEXT: pushl %ebx -; KNL_32-NEXT: .Ltmp0: -; KNL_32-NEXT: .cfi_def_cfa_offset 8 -; KNL_32-NEXT: pushl %esi -; KNL_32-NEXT: .Ltmp1: -; KNL_32-NEXT: .cfi_def_cfa_offset 12 -; KNL_32-NEXT: .Ltmp2: -; KNL_32-NEXT: .cfi_offset %esi, -12 -; KNL_32-NEXT: .Ltmp3: -; KNL_32-NEXT: .cfi_offset %ebx, -8 ; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax ; KNL_32-NEXT: andl $1, %eax -; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %ecx +; KNL_32-NEXT: kmovw %eax, %k1 +; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; KNL_32-NEXT: andl $1, %eax +; KNL_32-NEXT: kmovw %eax, %k2 +; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %eax +; KNL_32-NEXT: movl %eax, %ecx ; KNL_32-NEXT: andl $1, %ecx -; KNL_32-NEXT: movl {{[0-9]+}}(%esp), %ebx -; KNL_32-NEXT: movl %ebx, %edx -; KNL_32-NEXT: andl $1, %edx +; KNL_32-NEXT: kmovw %ecx, %k0 ; KNL_32-NEXT: vpslld $2, %xmm1, %xmm1 ; KNL_32-NEXT: vpaddd %xmm1, %xmm0, %xmm1 ; KNL_32-NEXT: # implicit-def: %XMM0 -; KNL_32-NEXT: testb $1, %bl +; KNL_32-NEXT: testb $1, %al ; KNL_32-NEXT: je .LBB29_2 ; KNL_32-NEXT: # BB#1: # %cond.load -; KNL_32-NEXT: vmovd %xmm1, %esi +; KNL_32-NEXT: vmovd %xmm1, %eax ; KNL_32-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero ; KNL_32-NEXT: .LBB29_2: # %else +; KNL_32-NEXT: kmovw %k2, %eax +; KNL_32-NEXT: movl %eax, %ecx +; KNL_32-NEXT: andl $1, %ecx ; KNL_32-NEXT: testb %cl, %cl ; KNL_32-NEXT: je .LBB29_4 ; KNL_32-NEXT: # BB#3: # %cond.load1 -; KNL_32-NEXT: vpextrd $1, %xmm1, %esi -; KNL_32-NEXT: vpinsrd $1, (%esi), %xmm0, %xmm0 +; KNL_32-NEXT: vpextrd $1, %xmm1, %ecx +; KNL_32-NEXT: vpinsrd $1, (%ecx), %xmm0, %xmm0 ; KNL_32-NEXT: .LBB29_4: # %else2 -; KNL_32-NEXT: testb %al, %al +; KNL_32-NEXT: kmovw %k1, %ecx +; KNL_32-NEXT: movl %ecx, %edx +; KNL_32-NEXT: andl $1, %edx +; KNL_32-NEXT: testb %dl, %dl ; KNL_32-NEXT: je .LBB29_6 ; KNL_32-NEXT: # BB#5: # %cond.load4 -; KNL_32-NEXT: vpextrd $2, %xmm1, %esi -; KNL_32-NEXT: vpinsrd $2, (%esi), %xmm0, %xmm0 +; KNL_32-NEXT: vpextrd $2, %xmm1, %edx +; KNL_32-NEXT: vpinsrd $2, (%edx), %xmm0, %xmm0 ; KNL_32-NEXT: .LBB29_6: # %else5 +; KNL_32-NEXT: kmovw %k0, %edx ; KNL_32-NEXT: vmovd %edx, %xmm1 -; KNL_32-NEXT: vpinsrd $1, %ecx, %xmm1, %xmm1 -; KNL_32-NEXT: vpinsrd $2, %eax, %xmm1, %xmm1 +; KNL_32-NEXT: vpinsrd $1, %eax, %xmm1, %xmm1 +; KNL_32-NEXT: vpinsrd $2, %ecx, %xmm1, %xmm1 ; KNL_32-NEXT: vpslld $31, %xmm1, %xmm1 ; KNL_32-NEXT: vblendvps %xmm1, %xmm0, %xmm2, %xmm0 -; KNL_32-NEXT: popl %esi -; KNL_32-NEXT: popl %ebx ; KNL_32-NEXT: retl ; ; SKX-LABEL: test30: @@ -1460,35 +1648,38 @@ define <3 x i32> @test30(<3 x i32*> %base, <3 x i32> %ind, <3 x i1> %mask, <3 x ; SKX-NEXT: kshiftrw $15, %k0, %k0 ; SKX-NEXT: vpmovsxdq %xmm1, %ymm1 ; SKX-NEXT: vpsllq $2, %ymm1, %ymm1 -; SKX-NEXT: vpaddq %ymm1, %ymm0, %ymm1 +; SKX-NEXT: vpaddq %ymm1, %ymm0, %ymm0 ; SKX-NEXT: kmovw %k0, %eax -; SKX-NEXT: # implicit-def: %XMM0 +; SKX-NEXT: andl $1, %eax +; SKX-NEXT: # implicit-def: %XMM1 ; SKX-NEXT: testb %al, %al ; SKX-NEXT: je .LBB29_2 ; SKX-NEXT: # BB#1: # %cond.load -; SKX-NEXT: vmovq %xmm1, %rax -; SKX-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; SKX-NEXT: vmovq %xmm0, %rax +; SKX-NEXT: vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero ; SKX-NEXT: .LBB29_2: # %else ; SKX-NEXT: kshiftlw $14, %k1, %k0 ; SKX-NEXT: kshiftrw $15, %k0, %k0 ; SKX-NEXT: kmovw %k0, %eax +; SKX-NEXT: andl $1, %eax ; SKX-NEXT: testb %al, %al ; SKX-NEXT: je .LBB29_4 ; SKX-NEXT: # BB#3: # %cond.load1 -; SKX-NEXT: vpextrq $1, %xmm1, %rax -; SKX-NEXT: vpinsrd $1, (%rax), %xmm0, %xmm0 +; SKX-NEXT: vpextrq $1, %xmm0, %rax +; SKX-NEXT: vpinsrd $1, (%rax), %xmm1, %xmm1 ; SKX-NEXT: .LBB29_4: # %else2 ; SKX-NEXT: kshiftlw $13, %k1, %k0 ; SKX-NEXT: kshiftrw $15, %k0, %k0 ; SKX-NEXT: kmovw %k0, %eax +; SKX-NEXT: andl $1, %eax ; SKX-NEXT: testb %al, %al ; SKX-NEXT: je .LBB29_6 ; SKX-NEXT: # BB#5: # %cond.load4 -; SKX-NEXT: vextracti64x2 $1, %ymm1, %xmm1 -; SKX-NEXT: vmovq %xmm1, %rax -; SKX-NEXT: vpinsrd $2, (%rax), %xmm0, %xmm0 +; SKX-NEXT: vextracti64x2 $1, %ymm0, %xmm0 +; SKX-NEXT: vmovq %xmm0, %rax +; SKX-NEXT: vpinsrd $2, (%rax), %xmm1, %xmm1 ; SKX-NEXT: .LBB29_6: # %else5 -; SKX-NEXT: vpblendmd %xmm0, %xmm3, %xmm0 {%k1} +; SKX-NEXT: vpblendmd %xmm1, %xmm3, %xmm0 {%k1} ; SKX-NEXT: retq ; ; SKX_32-LABEL: test30: @@ -1501,35 +1692,38 @@ define <3 x i32> @test30(<3 x i32*> %base, <3 x i32> %ind, <3 x i1> %mask, <3 x ; SKX_32-NEXT: kshiftlw $15, %k1, %k0 ; SKX_32-NEXT: kshiftrw $15, %k0, %k0 ; SKX_32-NEXT: vpslld $2, %xmm1, %xmm1 -; SKX_32-NEXT: vpaddd %xmm1, %xmm0, %xmm1 +; SKX_32-NEXT: vpaddd %xmm1, %xmm0, %xmm0 ; SKX_32-NEXT: kmovw %k0, %eax -; SKX_32-NEXT: # implicit-def: %XMM0 +; SKX_32-NEXT: andl $1, %eax +; SKX_32-NEXT: # implicit-def: %XMM1 ; SKX_32-NEXT: testb %al, %al ; SKX_32-NEXT: je .LBB29_2 ; SKX_32-NEXT: # BB#1: # %cond.load -; SKX_32-NEXT: vmovd %xmm1, %eax -; SKX_32-NEXT: vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero +; SKX_32-NEXT: vmovd %xmm0, %eax +; SKX_32-NEXT: vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero ; SKX_32-NEXT: .LBB29_2: # %else ; SKX_32-NEXT: kshiftlw $14, %k1, %k0 ; SKX_32-NEXT: kshiftrw $15, %k0, %k0 ; SKX_32-NEXT: kmovw %k0, %eax +; SKX_32-NEXT: andl $1, %eax ; SKX_32-NEXT: testb %al, %al ; SKX_32-NEXT: je .LBB29_4 ; SKX_32-NEXT: # BB#3: # %cond.load1 -; SKX_32-NEXT: vpextrd $1, %xmm1, %eax -; SKX_32-NEXT: vpinsrd $1, (%eax), %xmm0, %xmm0 +; SKX_32-NEXT: vpextrd $1, %xmm0, %eax +; SKX_32-NEXT: vpinsrd $1, (%eax), %xmm1, %xmm1 ; SKX_32-NEXT: .LBB29_4: # %else2 ; SKX_32-NEXT: vmovdqa32 {{[0-9]+}}(%esp), %xmm2 ; SKX_32-NEXT: kshiftlw $13, %k1, %k0 ; SKX_32-NEXT: kshiftrw $15, %k0, %k0 ; SKX_32-NEXT: kmovw %k0, %eax +; SKX_32-NEXT: andl $1, %eax ; SKX_32-NEXT: testb %al, %al ; SKX_32-NEXT: je .LBB29_6 ; SKX_32-NEXT: # BB#5: # %cond.load4 -; SKX_32-NEXT: vpextrd $2, %xmm1, %eax -; SKX_32-NEXT: vpinsrd $2, (%eax), %xmm0, %xmm0 +; SKX_32-NEXT: vpextrd $2, %xmm0, %eax +; SKX_32-NEXT: vpinsrd $2, (%eax), %xmm1, %xmm1 ; SKX_32-NEXT: .LBB29_6: # %else5 -; SKX_32-NEXT: vpblendmd %xmm0, %xmm2, %xmm0 {%k1} +; SKX_32-NEXT: vpblendmd %xmm1, %xmm2, %xmm0 {%k1} ; SKX_32-NEXT: addl $12, %esp ; SKX_32-NEXT: retl @@ -1646,12 +1840,12 @@ define <16 x i64> @test_gather_16i64(<16 x i64*> %ptrs, <16 x i1> %mask, <16 x i ; KNL_32-LABEL: test_gather_16i64: ; KNL_32: # BB#0: ; KNL_32-NEXT: pushl %ebp -; KNL_32-NEXT: .Ltmp4: +; KNL_32-NEXT: .Ltmp0: ; KNL_32-NEXT: .cfi_def_cfa_offset 8 -; KNL_32-NEXT: .Ltmp5: +; KNL_32-NEXT: .Ltmp1: ; KNL_32-NEXT: .cfi_offset %ebp, -8 ; KNL_32-NEXT: movl %esp, %ebp -; KNL_32-NEXT: .Ltmp6: +; KNL_32-NEXT: .Ltmp2: ; KNL_32-NEXT: .cfi_def_cfa_register %ebp ; KNL_32-NEXT: andl $-64, %esp ; KNL_32-NEXT: subl $64, %esp @@ -1769,12 +1963,12 @@ define <16 x double> @test_gather_16f64(<16 x double*> %ptrs, <16 x i1> %mask, < ; KNL_32-LABEL: test_gather_16f64: ; KNL_32: # BB#0: ; KNL_32-NEXT: pushl %ebp -; KNL_32-NEXT: .Ltmp7: +; KNL_32-NEXT: .Ltmp3: ; KNL_32-NEXT: .cfi_def_cfa_offset 8 -; KNL_32-NEXT: .Ltmp8: +; KNL_32-NEXT: .Ltmp4: ; KNL_32-NEXT: .cfi_offset %ebp, -8 ; KNL_32-NEXT: movl %esp, %ebp -; KNL_32-NEXT: .Ltmp9: +; KNL_32-NEXT: .Ltmp5: ; KNL_32-NEXT: .cfi_def_cfa_register %ebp ; KNL_32-NEXT: andl $-64, %esp ; KNL_32-NEXT: subl $64, %esp @@ -1886,12 +2080,12 @@ define void @test_scatter_16i64(<16 x i64*> %ptrs, <16 x i1> %mask, <16 x i64> % ; KNL_32-LABEL: test_scatter_16i64: ; KNL_32: # BB#0: ; KNL_32-NEXT: pushl %ebp -; KNL_32-NEXT: .Ltmp10: +; KNL_32-NEXT: .Ltmp6: ; KNL_32-NEXT: .cfi_def_cfa_offset 8 -; KNL_32-NEXT: .Ltmp11: +; KNL_32-NEXT: .Ltmp7: ; KNL_32-NEXT: .cfi_offset %ebp, -8 ; KNL_32-NEXT: movl %esp, %ebp -; KNL_32-NEXT: .Ltmp12: +; KNL_32-NEXT: .Ltmp8: ; KNL_32-NEXT: .cfi_def_cfa_register %ebp ; KNL_32-NEXT: andl $-64, %esp ; KNL_32-NEXT: subl $64, %esp @@ -2000,12 +2194,12 @@ define void @test_scatter_16f64(<16 x double*> %ptrs, <16 x i1> %mask, <16 x dou ; KNL_32-LABEL: test_scatter_16f64: ; KNL_32: # BB#0: ; KNL_32-NEXT: pushl %ebp -; KNL_32-NEXT: .Ltmp13: +; KNL_32-NEXT: .Ltmp9: ; KNL_32-NEXT: .cfi_def_cfa_offset 8 -; KNL_32-NEXT: .Ltmp14: +; KNL_32-NEXT: .Ltmp10: ; KNL_32-NEXT: .cfi_offset %ebp, -8 ; KNL_32-NEXT: movl %esp, %ebp -; KNL_32-NEXT: .Ltmp15: +; KNL_32-NEXT: .Ltmp11: ; KNL_32-NEXT: .cfi_def_cfa_register %ebp ; KNL_32-NEXT: andl $-64, %esp ; KNL_32-NEXT: subl $64, %esp |
