diff options
| -rw-r--r-- | llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll | 845 |
1 files changed, 278 insertions, 567 deletions
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll b/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll index 122c77aefcc..0cb586ff101 100644 --- a/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll +++ b/llvm/test/CodeGen/X86/vector-shuffle-combining-avx2.ll @@ -1,8 +1,8 @@ ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py -; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX2 -; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx512f | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX512 -; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX2 -; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx512f | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX512 +; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,X86,AVX2,X86-AVX2 +; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,X86,AVX512,X86-AVX512 +; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,X64,AVX2,X64-AVX2 +; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,X64,AVX512,X64-AVX512 declare <8 x i32> @llvm.x86.avx2.permd(<8 x i32>, <8 x i32>) declare <8 x float> @llvm.x86.avx2.permps(<8 x float>, <8 x i32>) @@ -10,45 +10,30 @@ declare <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8>, <16 x i8>) declare <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8>, <32 x i8>) define <32 x i8> @combine_pshufb_pslldq(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_pslldq: -; X32: # %bb.0: -; X32-NEXT: vxorps %xmm0, %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_pslldq: -; X64: # %bb.0: -; X64-NEXT: vxorps %xmm0, %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_pslldq: +; CHECK: # %bb.0: +; CHECK-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7>) %2 = shufflevector <32 x i8> %1, <32 x i8> zeroinitializer, <32 x i32> <i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23> ret <32 x i8> %2 } define <32 x i8> @combine_pshufb_psrldq(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_psrldq: -; X32: # %bb.0: -; X32-NEXT: vxorps %xmm0, %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_psrldq: -; X64: # %bb.0: -; X64-NEXT: vxorps %xmm0, %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_psrldq: +; CHECK: # %bb.0: +; CHECK-NEXT: vxorps %xmm0, %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128>) %2 = shufflevector <32 x i8> %1, <32 x i8> zeroinitializer, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32> ret <32 x i8> %2 } define <32 x i8> @combine_pshufb_vpermd(<8 x i32> %a) { -; X32-LABEL: combine_pshufb_vpermd: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_vpermd: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_vpermd: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18] +; CHECK-NEXT: ret{{[l|q]}} %tmp0 = call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 4>) %tmp1 = bitcast <8 x i32> %tmp0 to <32 x i8> %tmp2 = shufflevector <32 x i8> %tmp1, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 30> @@ -56,15 +41,10 @@ define <32 x i8> @combine_pshufb_vpermd(<8 x i32> %a) { } define <32 x i8> @combine_pshufb_vpermps(<8 x float> %a) { -; X32-LABEL: combine_pshufb_vpermps: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_vpermps: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_vpermps: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,16,17,18,18] +; CHECK-NEXT: ret{{[l|q]}} %tmp0 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 4>) %tmp1 = bitcast <8 x float> %tmp0 to <32 x i8> %tmp2 = shufflevector <32 x i8> %tmp1, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 30> @@ -72,45 +52,33 @@ define <32 x i8> @combine_pshufb_vpermps(<8 x float> %a) { } define <32 x i8> @combine_and_pshufb(<32 x i8> %a0) { -; X32-LABEL: combine_and_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X32-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7],ymm0[8],ymm1[9,10,11],ymm0[12],ymm1[13,14,15] -; X32-NEXT: retl -; -; X64-LABEL: combine_and_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X64-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7],ymm0[8],ymm1[9,10,11],ymm0[12],ymm1[13,14,15] -; X64-NEXT: retq +; CHECK-LABEL: combine_and_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1 +; CHECK-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7],ymm0[8],ymm1[9,10,11],ymm0[12],ymm1[13,14,15] +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <32 x i8> %a0, <32 x i8> zeroinitializer, <32 x i32> <i32 0, i32 1, i32 32, i32 32, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31> %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> <i8 0, i8 1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 8, i8 9, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 8, i8 9, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>) ret <32 x i8> %2 } define <32 x i8> @combine_pshufb_and(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_and: -; X32: # %bb.0: -; X32-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X32-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7],ymm0[8],ymm1[9,10,11],ymm0[12],ymm1[13,14,15] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_and: -; X64: # %bb.0: -; X64-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X64-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7],ymm0[8],ymm1[9,10,11],ymm0[12],ymm1[13,14,15] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_and: +; CHECK: # %bb.0: +; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1 +; CHECK-NEXT: vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7],ymm0[8],ymm1[9,10,11],ymm0[12],ymm1[13,14,15] +; CHECK-NEXT: ret{{[l|q]}} %1 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 0, i8 1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 8, i8 9, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 8, i8 9, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>) %2 = shufflevector <32 x i8> %1, <32 x i8> zeroinitializer, <32 x i32> <i32 0, i32 1, i32 32, i32 32, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31> ret <32 x i8> %2 } define <4 x i64> @combine_permq_pshufb_as_vperm2i128(<4 x i64> %a0) { -; X32-LABEL: combine_permq_pshufb_as_vperm2i128: -; X32: # %bb.0: -; X32-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero -; X32-NEXT: vpaddq {{\.LCPI.*}}, %ymm0, %ymm0 -; X32-NEXT: retl +; X86-LABEL: combine_permq_pshufb_as_vperm2i128: +; X86: # %bb.0: +; X86-NEXT: vperm2i128 {{.*#+}} ymm0 = ymm0[2,3],zero,zero +; X86-NEXT: vpaddq {{\.LCPI.*}}, %ymm0, %ymm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_permq_pshufb_as_vperm2i128: ; X64: # %bb.0: @@ -126,17 +94,11 @@ define <4 x i64> @combine_permq_pshufb_as_vperm2i128(<4 x i64> %a0) { } define <8 x i32> @combine_as_vpermd(<8 x i32> %a0) { -; X32-LABEL: combine_as_vpermd: -; X32: # %bb.0: -; X32-NEXT: vmovaps {{.*#+}} ymm1 = [4,5,4,5,6,7,0,7] -; X32-NEXT: vpermps %ymm0, %ymm1, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_as_vpermd: -; X64: # %bb.0: -; X64-NEXT: vmovaps {{.*#+}} ymm1 = [4,5,4,5,6,7,0,7] -; X64-NEXT: vpermps %ymm0, %ymm1, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_as_vpermd: +; CHECK: # %bb.0: +; CHECK-NEXT: vmovaps {{.*#+}} ymm1 = [4,5,4,5,6,7,0,7] +; CHECK-NEXT: vpermps %ymm0, %ymm1, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <8 x i32> %a0, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3> %2 = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a0, <8 x i32> <i32 5, i32 4, i32 3, i32 2, i32 1, i32 0, i32 7, i32 6>) %3 = shufflevector <8 x i32> %1, <8 x i32> %2, <8 x i32> <i32 0, i32 8, i32 9, i32 1, i32 15, i32 14, i32 4, i32 3> @@ -144,17 +106,11 @@ define <8 x i32> @combine_as_vpermd(<8 x i32> %a0) { } define <8 x float> @combine_as_vpermps(<8 x float> %a0) { -; X32-LABEL: combine_as_vpermps: -; X32: # %bb.0: -; X32-NEXT: vmovaps {{.*#+}} ymm1 = <6,4,7,5,1,u,4,7> -; X32-NEXT: vpermps %ymm0, %ymm1, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_as_vpermps: -; X64: # %bb.0: -; X64-NEXT: vmovaps {{.*#+}} ymm1 = <6,4,7,5,1,u,4,7> -; X64-NEXT: vpermps %ymm0, %ymm1, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_as_vpermps: +; CHECK: # %bb.0: +; CHECK-NEXT: vmovaps {{.*#+}} ymm1 = <6,4,7,5,1,u,4,7> +; CHECK-NEXT: vpermps %ymm0, %ymm1, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <8 x float> %a0, <8 x float> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7> %2 = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a0, <8 x i32> <i32 1, i32 undef, i32 3, i32 2, i32 5, i32 4, i32 7, i32 6>) %3 = shufflevector <8 x float> %1, <8 x float> %2, <8 x i32> <i32 15, i32 0, i32 14, i32 1, i32 8, i32 9, i32 4, i32 3> @@ -162,17 +118,11 @@ define <8 x float> @combine_as_vpermps(<8 x float> %a0) { } define <32 x i8> @combine_permq_pshufb_as_vpblendd(<4 x i64> %a0) { -; X32-LABEL: combine_permq_pshufb_as_vpblendd: -; X32: # %bb.0: -; X32-NEXT: vxorps %xmm1, %xmm1, %xmm1 -; X32-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7] -; X32-NEXT: retl -; -; X64-LABEL: combine_permq_pshufb_as_vpblendd: -; X64: # %bb.0: -; X64-NEXT: vxorps %xmm1, %xmm1, %xmm1 -; X64-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7] -; X64-NEXT: retq +; CHECK-LABEL: combine_permq_pshufb_as_vpblendd: +; CHECK: # %bb.0: +; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; CHECK-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7] +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <4 x i64> %a0, <4 x i64> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2> %2 = bitcast <4 x i64> %1 to <32 x i8> %3 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %2, <32 x i8> <i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255, i8 255>) @@ -180,31 +130,20 @@ define <32 x i8> @combine_permq_pshufb_as_vpblendd(<4 x i64> %a0) { } define <16 x i8> @combine_pshufb_as_vpbroadcastb128(<16 x i8> %a) { -; X32-LABEL: combine_pshufb_as_vpbroadcastb128: -; X32: # %bb.0: -; X32-NEXT: vpbroadcastb %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vpbroadcastb128: -; X64: # %bb.0: -; X64-NEXT: vpbroadcastb %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vpbroadcastb128: +; CHECK: # %bb.0: +; CHECK-NEXT: vpbroadcastb %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %a, <16 x i8> zeroinitializer) ret <16 x i8> %1 } define <32 x i8> @combine_pshufb_as_vpbroadcastb256(<2 x i64> %a) { -; X32-LABEL: combine_pshufb_as_vpbroadcastb256: -; X32: # %bb.0: -; X32-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X32-NEXT: vpbroadcastb %xmm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vpbroadcastb256: -; X64: # %bb.0: -; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X64-NEXT: vpbroadcastb %xmm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vpbroadcastb256: +; CHECK: # %bb.0: +; CHECK-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 +; CHECK-NEXT: vpbroadcastb %xmm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <2 x i64> %a, <2 x i64> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef> %2 = bitcast <4 x i64> %1 to <32 x i8> %3 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %2, <32 x i8> zeroinitializer) @@ -215,31 +154,20 @@ define <32 x i8> @combine_pshufb_as_vpbroadcastb256(<2 x i64> %a) { } define <16 x i8> @combine_pshufb_as_vpbroadcastw128(<16 x i8> %a) { -; X32-LABEL: combine_pshufb_as_vpbroadcastw128: -; X32: # %bb.0: -; X32-NEXT: vpbroadcastw %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vpbroadcastw128: -; X64: # %bb.0: -; X64-NEXT: vpbroadcastw %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vpbroadcastw128: +; CHECK: # %bb.0: +; CHECK-NEXT: vpbroadcastw %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %a, <16 x i8> <i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1>) ret <16 x i8> %1 } define <32 x i8> @combine_pshufb_as_vpbroadcastw256(<2 x i64> %a) { -; X32-LABEL: combine_pshufb_as_vpbroadcastw256: -; X32: # %bb.0: -; X32-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X32-NEXT: vpbroadcastw %xmm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vpbroadcastw256: -; X64: # %bb.0: -; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X64-NEXT: vpbroadcastw %xmm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vpbroadcastw256: +; CHECK: # %bb.0: +; CHECK-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 +; CHECK-NEXT: vpbroadcastw %xmm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <2 x i64> %a, <2 x i64> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef> %2 = bitcast <4 x i64> %1 to <32 x i8> %3 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %2, <32 x i8> <i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1, i8 0, i8 1>) @@ -250,11 +178,11 @@ define <32 x i8> @combine_pshufb_as_vpbroadcastw256(<2 x i64> %a) { } define <16 x i8> @combine_pshufb_as_vpbroadcastd128(<16 x i8> %a) { -; X32-LABEL: combine_pshufb_as_vpbroadcastd128: -; X32: # %bb.0: -; X32-NEXT: vpbroadcastd %xmm0, %xmm0 -; X32-NEXT: vpaddb {{\.LCPI.*}}, %xmm0, %xmm0 -; X32-NEXT: retl +; X86-LABEL: combine_pshufb_as_vpbroadcastd128: +; X86: # %bb.0: +; X86-NEXT: vpbroadcastd %xmm0, %xmm0 +; X86-NEXT: vpaddb {{\.LCPI.*}}, %xmm0, %xmm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_pshufb_as_vpbroadcastd128: ; X64: # %bb.0: @@ -267,12 +195,12 @@ define <16 x i8> @combine_pshufb_as_vpbroadcastd128(<16 x i8> %a) { } define <8 x i32> @combine_permd_as_vpbroadcastd256(<4 x i32> %a) { -; X32-LABEL: combine_permd_as_vpbroadcastd256: -; X32: # %bb.0: -; X32-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X32-NEXT: vpbroadcastd %xmm0, %ymm0 -; X32-NEXT: vpaddd {{\.LCPI.*}}, %ymm0, %ymm0 -; X32-NEXT: retl +; X86-LABEL: combine_permd_as_vpbroadcastd256: +; X86: # %bb.0: +; X86-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 +; X86-NEXT: vpbroadcastd %xmm0, %ymm0 +; X86-NEXT: vpaddd {{\.LCPI.*}}, %ymm0, %ymm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_permd_as_vpbroadcastd256: ; X64: # %bb.0: @@ -287,26 +215,21 @@ define <8 x i32> @combine_permd_as_vpbroadcastd256(<4 x i32> %a) { } define <16 x i8> @combine_pshufb_as_vpbroadcastq128(<16 x i8> %a) { -; X32-LABEL: combine_pshufb_as_vpbroadcastq128: -; X32: # %bb.0: -; X32-NEXT: vpbroadcastq %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vpbroadcastq128: -; X64: # %bb.0: -; X64-NEXT: vpbroadcastq %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vpbroadcastq128: +; CHECK: # %bb.0: +; CHECK-NEXT: vpbroadcastq %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %a, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7>) ret <16 x i8> %1 } define <8 x i32> @combine_permd_as_vpbroadcastq256(<4 x i32> %a) { -; X32-LABEL: combine_permd_as_vpbroadcastq256: -; X32: # %bb.0: -; X32-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X32-NEXT: vpbroadcastq %xmm0, %ymm0 -; X32-NEXT: vpaddd {{\.LCPI.*}}, %ymm0, %ymm0 -; X32-NEXT: retl +; X86-LABEL: combine_permd_as_vpbroadcastq256: +; X86: # %bb.0: +; X86-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 +; X86-NEXT: vpbroadcastq %xmm0, %ymm0 +; X86-NEXT: vpaddd {{\.LCPI.*}}, %ymm0, %ymm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_permd_as_vpbroadcastq256: ; X64: # %bb.0: @@ -321,15 +244,10 @@ define <8 x i32> @combine_permd_as_vpbroadcastq256(<4 x i32> %a) { } define <4 x float> @combine_pshufb_as_vpbroadcastss128(<4 x float> %a) { -; X32-LABEL: combine_pshufb_as_vpbroadcastss128: -; X32: # %bb.0: -; X32-NEXT: vbroadcastss %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vpbroadcastss128: -; X64: # %bb.0: -; X64-NEXT: vbroadcastss %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vpbroadcastss128: +; CHECK: # %bb.0: +; CHECK-NEXT: vbroadcastss %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = bitcast <4 x float> %a to <16 x i8> %2 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %1, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 0, i8 1, i8 2, i8 3, i8 0, i8 1, i8 2, i8 3, i8 0, i8 1, i8 2, i8 3>) %3 = bitcast <16 x i8> %2 to <4 x float> @@ -337,34 +255,22 @@ define <4 x float> @combine_pshufb_as_vpbroadcastss128(<4 x float> %a) { } define <8 x float> @combine_permps_as_vpbroadcastss256(<4 x float> %a) { -; X32-LABEL: combine_permps_as_vpbroadcastss256: -; X32: # %bb.0: -; X32-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X32-NEXT: vbroadcastss %xmm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_permps_as_vpbroadcastss256: -; X64: # %bb.0: -; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X64-NEXT: vbroadcastss %xmm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_permps_as_vpbroadcastss256: +; CHECK: # %bb.0: +; CHECK-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 +; CHECK-NEXT: vbroadcastss %xmm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <4 x float> %a, <4 x float> undef, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef> %2 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %1, <8 x i32> zeroinitializer) ret <8 x float> %2 } define <4 x double> @combine_permps_as_vpbroadcastsd256(<2 x double> %a) { -; X32-LABEL: combine_permps_as_vpbroadcastsd256: -; X32: # %bb.0: -; X32-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X32-NEXT: vbroadcastsd %xmm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_permps_as_vpbroadcastsd256: -; X64: # %bb.0: -; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 -; X64-NEXT: vbroadcastsd %xmm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_permps_as_vpbroadcastsd256: +; CHECK: # %bb.0: +; CHECK-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 +; CHECK-NEXT: vbroadcastsd %xmm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <2 x double> %a, <2 x double> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef> %2 = bitcast <4 x double> %1 to <8 x float> %3 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %2, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>) @@ -373,45 +279,30 @@ define <4 x double> @combine_permps_as_vpbroadcastsd256(<2 x double> %a) { } define <16 x i8> @combine_vpbroadcast_pshufb_as_vpbroadcastb128(<16 x i8> %a) { -; X32-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastb128: -; X32: # %bb.0: -; X32-NEXT: vpbroadcastb %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastb128: -; X64: # %bb.0: -; X64-NEXT: vpbroadcastb %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastb128: +; CHECK: # %bb.0: +; CHECK-NEXT: vpbroadcastb %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <16 x i8> %a, <16 x i8> undef, <16 x i32> zeroinitializer %2 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %1, <16 x i8> zeroinitializer) ret <16 x i8> %2 } define <32 x i8> @combine_vpbroadcast_pshufb_as_vpbroadcastb256(<32 x i8> %a) { -; X32-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastb256: -; X32: # %bb.0: -; X32-NEXT: vpbroadcastb %xmm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastb256: -; X64: # %bb.0: -; X64-NEXT: vpbroadcastb %xmm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastb256: +; CHECK: # %bb.0: +; CHECK-NEXT: vpbroadcastb %xmm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <32 x i8> %a, <32 x i8> undef, <32 x i32> zeroinitializer %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> zeroinitializer) ret <32 x i8> %2 } define <4 x float> @combine_vpbroadcast_pshufb_as_vpbroadcastss128(<4 x float> %a) { -; X32-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastss128: -; X32: # %bb.0: -; X32-NEXT: vbroadcastss %xmm0, %xmm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastss128: -; X64: # %bb.0: -; X64-NEXT: vbroadcastss %xmm0, %xmm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_vpbroadcast_pshufb_as_vpbroadcastss128: +; CHECK: # %bb.0: +; CHECK-NEXT: vbroadcastss %xmm0, %xmm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> zeroinitializer %2 = bitcast <4 x float> %1 to <16 x i8> %3 = call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %2, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 0, i8 1, i8 2, i8 3, i8 0, i8 1, i8 2, i8 3, i8 0, i8 1, i8 2, i8 3>) @@ -420,34 +311,22 @@ define <4 x float> @combine_vpbroadcast_pshufb_as_vpbroadcastss128(<4 x float> % } define <8 x float> @combine_vpbroadcast_permd_as_vpbroadcastss256(<4 x float> %a) { -; X32-LABEL: combine_vpbroadcast_permd_as_vpbroadcastss256: -; X32: # %bb.0: -; X32-NEXT: vbroadcastss %xmm0, %ymm0 -; X32-NEXT: vbroadcastss %xmm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_vpbroadcast_permd_as_vpbroadcastss256: -; X64: # %bb.0: -; X64-NEXT: vbroadcastss %xmm0, %ymm0 -; X64-NEXT: vbroadcastss %xmm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_vpbroadcast_permd_as_vpbroadcastss256: +; CHECK: # %bb.0: +; CHECK-NEXT: vbroadcastss %xmm0, %ymm0 +; CHECK-NEXT: vbroadcastss %xmm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <4 x float> %a, <4 x float> undef, <8 x i32> zeroinitializer %2 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %1, <8 x i32> zeroinitializer) ret <8 x float> %2 } define <4 x double> @combine_vpbroadcast_permd_as_vpbroadcastsd256(<2 x double> %a) { -; X32-LABEL: combine_vpbroadcast_permd_as_vpbroadcastsd256: -; X32: # %bb.0: -; X32-NEXT: vbroadcastsd %xmm0, %ymm0 -; X32-NEXT: vbroadcastsd %xmm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_vpbroadcast_permd_as_vpbroadcastsd256: -; X64: # %bb.0: -; X64-NEXT: vbroadcastsd %xmm0, %ymm0 -; X64-NEXT: vbroadcastsd %xmm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_vpbroadcast_permd_as_vpbroadcastsd256: +; CHECK: # %bb.0: +; CHECK-NEXT: vbroadcastsd %xmm0, %ymm0 +; CHECK-NEXT: vbroadcastsd %xmm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <2 x double> %a, <2 x double> undef, <4 x i32> zeroinitializer %2 = bitcast <4 x double> %1 to <8 x float> %3 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %2, <8 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>) @@ -456,43 +335,28 @@ define <4 x double> @combine_vpbroadcast_permd_as_vpbroadcastsd256(<2 x double> } define <8 x i32> @combine_permd_as_permq(<8 x i32> %a) { -; X32-LABEL: combine_permd_as_permq: -; X32: # %bb.0: -; X32-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1] -; X32-NEXT: retl -; -; X64-LABEL: combine_permd_as_permq: -; X64: # %bb.0: -; X64-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1] -; X64-NEXT: retq +; CHECK-LABEL: combine_permd_as_permq: +; CHECK: # %bb.0: +; CHECK-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,2,2,1] +; CHECK-NEXT: ret{{[l|q]}} %1 = call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a, <8 x i32> <i32 0, i32 1, i32 4, i32 5, i32 4, i32 5, i32 2, i32 3>) ret <8 x i32> %1 } define <8 x float> @combine_permps_as_permpd(<8 x float> %a) { -; X32-LABEL: combine_permps_as_permpd: -; X32: # %bb.0: -; X32-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,0,1] -; X32-NEXT: retl -; -; X64-LABEL: combine_permps_as_permpd: -; X64: # %bb.0: -; X64-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,0,1] -; X64-NEXT: retq +; CHECK-LABEL: combine_permps_as_permpd: +; CHECK: # %bb.0: +; CHECK-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[3,2,0,1] +; CHECK-NEXT: ret{{[l|q]}} %1 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a, <8 x i32> <i32 6, i32 7, i32 4, i32 5, i32 0, i32 1, i32 2, i32 3>) ret <8 x float> %1 } define <4 x i64> @combine_pshufb_as_zext(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_zext: -; X32: # %bb.0: -; X32-NEXT: vpmovzxwq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_zext: -; X64: # %bb.0: -; X64-NEXT: vpmovzxwq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_zext: +; CHECK: # %bb.0: +; CHECK-NEXT: vpmovzxwq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15> %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> <i8 8, i8 9, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 10, i8 11, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 4, i8 5, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 6, i8 7, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>) %3 = bitcast <32 x i8> %2 to <4 x i64> @@ -500,19 +364,12 @@ define <4 x i64> @combine_pshufb_as_zext(<32 x i8> %a0) { } define <4 x i64> @combine_pshufb_as_zext128(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_zext128: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0] -; X32-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1] -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[15,14],zero,zero,zero,zero,zero,zero,ymm0[13,12],zero,zero,zero,zero,zero,zero,ymm0[31,30],zero,zero,zero,zero,zero,zero,ymm0[29,28],zero,zero,zero,zero,zero,zero -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_zext128: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0] -; X64-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1] -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[15,14],zero,zero,zero,zero,zero,zero,ymm0[13,12],zero,zero,zero,zero,zero,zero,ymm0[31,30],zero,zero,zero,zero,zero,zero,ymm0[29,28],zero,zero,zero,zero,zero,zero -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_zext128: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0] +; CHECK-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1] +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[15,14],zero,zero,zero,zero,zero,zero,ymm0[13,12],zero,zero,zero,zero,zero,zero,ymm0[31,30],zero,zero,zero,zero,zero,zero,ymm0[29,28],zero,zero,zero,zero,zero,zero +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0> %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> <i8 15, i8 14, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 13, i8 12, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 15, i8 14, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 13, i8 12, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>) %3 = bitcast <32 x i8> %2 to <4 x i64> @@ -520,17 +377,11 @@ define <4 x i64> @combine_pshufb_as_zext128(<32 x i8> %a0) { } define <4 x double> @combine_pshufb_as_vzmovl_64(<4 x double> %a0) { -; X32-LABEL: combine_pshufb_as_vzmovl_64: -; X32: # %bb.0: -; X32-NEXT: vxorps %xmm1, %xmm1, %xmm1 -; X32-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vzmovl_64: -; X64: # %bb.0: -; X64-NEXT: vxorps %xmm1, %xmm1, %xmm1 -; X64-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vzmovl_64: +; CHECK: # %bb.0: +; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; CHECK-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3] +; CHECK-NEXT: ret{{[l|q]}} %1 = bitcast <4 x double> %a0 to <32 x i8> %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>) %3 = bitcast <32 x i8> %2 to <4 x double> @@ -538,17 +389,11 @@ define <4 x double> @combine_pshufb_as_vzmovl_64(<4 x double> %a0) { } define <8 x float> @combine_pshufb_as_vzmovl_32(<8 x float> %a0) { -; X32-LABEL: combine_pshufb_as_vzmovl_32: -; X32: # %bb.0: -; X32-NEXT: vxorps %xmm1, %xmm1, %xmm1 -; X32-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_vzmovl_32: -; X64: # %bb.0: -; X64-NEXT: vxorps %xmm1, %xmm1, %xmm1 -; X64-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_vzmovl_32: +; CHECK: # %bb.0: +; CHECK-NEXT: vxorps %xmm1, %xmm1, %xmm1 +; CHECK-NEXT: vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3] +; CHECK-NEXT: ret{{[l|q]}} %1 = bitcast <8 x float> %a0 to <32 x i8> %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>) %3 = bitcast <32 x i8> %2 to <8 x float> @@ -556,168 +401,112 @@ define <8 x float> @combine_pshufb_as_vzmovl_32(<8 x float> %a0) { } define <32 x i8> @combine_pshufb_as_pslldq(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_pslldq: -; X32: # %bb.0: -; X32-NEXT: vpslldq {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,18,19,20,21] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_pslldq: -; X64: # %bb.0: -; X64-NEXT: vpslldq {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,18,19,20,21] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_pslldq: +; CHECK: # %bb.0: +; CHECK-NEXT: vpslldq {{.*#+}} ymm0 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[0,1,2,3,4,5],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[16,17,18,19,20,21] +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5>) ret <32 x i8> %res0 } define <32 x i8> @combine_pshufb_as_psrldq(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_psrldq: -; X32: # %bb.0: -; X32-NEXT: vpsrldq {{.*#+}} ymm0 = ymm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_psrldq: -; X64: # %bb.0: -; X64-NEXT: vpsrldq {{.*#+}} ymm0 = ymm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_psrldq: +; CHECK: # %bb.0: +; CHECK-NEXT: vpsrldq {{.*#+}} ymm0 = ymm0[15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,ymm0[31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128, i8 128>) ret <32 x i8> %res0 } define <32 x i8> @combine_pshufb_as_psrlw(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_psrlw: -; X32: # %bb.0: -; X32-NEXT: vpsrlw $8, %ymm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_psrlw: -; X64: # %bb.0: -; X64-NEXT: vpsrlw $8, %ymm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_psrlw: +; CHECK: # %bb.0: +; CHECK-NEXT: vpsrlw $8, %ymm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 1, i8 128, i8 3, i8 128, i8 5, i8 128, i8 7, i8 128, i8 9, i8 128, i8 11, i8 128, i8 13, i8 128, i8 15, i8 128, i8 17, i8 128, i8 19, i8 128, i8 21, i8 128, i8 23, i8 128, i8 25, i8 128, i8 27, i8 128, i8 29, i8 128, i8 31, i8 128>) ret <32 x i8> %res0 } define <32 x i8> @combine_pshufb_as_pslld(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_pslld: -; X32: # %bb.0: -; X32-NEXT: vpslld $24, %ymm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_pslld: -; X64: # %bb.0: -; X64-NEXT: vpslld $24, %ymm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_pslld: +; CHECK: # %bb.0: +; CHECK-NEXT: vpslld $24, %ymm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 128, i8 128, i8 128, i8 0, i8 128, i8 128, i8 128, i8 4, i8 128, i8 128, i8 128, i8 8, i8 128, i8 128, i8 128, i8 12, i8 128, i8 128, i8 128, i8 16, i8 128, i8 128, i8 128, i8 20, i8 128, i8 128, i8 128, i8 24, i8 128, i8 128, i8 128, i8 28>) ret <32 x i8> %res0 } define <32 x i8> @combine_pshufb_as_psrlq(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_psrlq: -; X32: # %bb.0: -; X32-NEXT: vpsrlq $40, %ymm0, %ymm0 -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_psrlq: -; X64: # %bb.0: -; X64-NEXT: vpsrlq $40, %ymm0, %ymm0 -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_psrlq: +; CHECK: # %bb.0: +; CHECK-NEXT: vpsrlq $40, %ymm0, %ymm0 +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 5, i8 6, i8 7, i8 128, i8 128, i8 128, i8 128, i8 128, i8 13, i8 14, i8 15, i8 128, i8 128, i8 128, i8 128, i8 128, i8 21, i8 22, i8 23, i8 128, i8 128, i8 128, i8 128, i8 128, i8 29, i8 30, i8 31, i8 128, i8 128, i8 128, i8 128, i8 128>) ret <32 x i8> %res0 } define <32 x i8> @combine_pshufb_as_pshuflw(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_pshuflw: -; X32: # %bb.0: -; X32-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_pshuflw: -; X64: # %bb.0: -; X64-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_pshuflw: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshuflw {{.*#+}} ymm0 = ymm0[1,0,3,2,4,5,6,7,9,8,11,10,12,13,14,15] +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 2, i8 3, i8 0, i8 1, i8 6, i8 7, i8 4, i8 5, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15, i8 2, i8 3, i8 0, i8 1, i8 6, i8 7, i8 4, i8 5, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>) ret <32 x i8> %res0 } define <32 x i8> @combine_pshufb_as_pshufhw(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_pshufhw: -; X32: # %bb.0: -; X32-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_pshufhw: -; X64: # %bb.0: -; X64-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_pshufhw: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,5,4,7,6,8,9,10,11,13,12,15,14] +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 10, i8 11, i8 8, i8 9, i8 14, i8 15, i8 12, i8 13, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 10, i8 11, i8 8, i8 9, i8 14, i8 15, i8 12, i8 13>) ret <32 x i8> %res0 } define <32 x i8> @combine_pshufb_not_as_pshufw(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_not_as_pshufw: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5,10,11,8,9,14,15,12,13,18,19,16,17,22,23,20,21,26,27,24,25,30,31,28,29] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_not_as_pshufw: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5,10,11,8,9,14,15,12,13,18,19,16,17,22,23,20,21,26,27,24,25,30,31,28,29] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_not_as_pshufw: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5,10,11,8,9,14,15,12,13,18,19,16,17,22,23,20,21,26,27,24,25,30,31,28,29] +; CHECK-NEXT: ret{{[l|q]}} %res0 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 2, i8 3, i8 0, i8 1, i8 6, i8 7, i8 4, i8 5, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15, i8 2, i8 3, i8 0, i8 1, i8 6, i8 7, i8 4, i8 5, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>) %res1 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %res0, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 10, i8 11, i8 8, i8 9, i8 14, i8 15, i8 12, i8 13, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 10, i8 11, i8 8, i8 9, i8 14, i8 15, i8 12, i8 13>) ret <32 x i8> %res1 } define <32 x i8> @combine_pshufb_as_unpacklo_undef(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_unpacklo_undef: -; X32: # %bb.0: -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_unpacklo_undef: -; X64: # %bb.0: -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_unpacklo_undef: +; CHECK: # %bb.0: +; CHECK-NEXT: ret{{[l|q]}} %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 undef, i8 0, i8 undef, i8 1, i8 undef, i8 2, i8 undef, i8 3, i8 undef, i8 4, i8 undef, i8 5, i8 undef, i8 6, i8 undef, i8 7, i8 undef, i8 16, i8 undef, i8 17, i8 undef, i8 18, i8 undef, i8 19, i8 undef, i8 20, i8 undef, i8 21, i8 undef, i8 22, i8 undef, i8 23>) %2 = shufflevector <32 x i8> %1, <32 x i8> undef, <32 x i32> <i32 0, i32 0, i32 2, i32 2, i32 4, i32 4, i32 6, i32 6, i32 8, i32 8, i32 10, i32 10, i32 12, i32 12, i32 14, i32 14, i32 16, i32 16, i32 18, i32 18, i32 20, i32 20, i32 22, i32 22, i32 24, i32 24, i32 26, i32 26, i32 28, i32 28, i32 30, i32 30> ret <32 x i8> %2 } define <32 x i8> @combine_pshufb_as_unpacklo_zero(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_unpacklo_zero: -; X32: # %bb.0: -; X32-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X32-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_unpacklo_zero: -; X64: # %bb.0: -; X64-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X64-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_unpacklo_zero: +; CHECK: # %bb.0: +; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1 +; CHECK-NEXT: vpunpcklwd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11] +; CHECK-NEXT: ret{{[l|q]}} %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 0, i8 1, i8 -1, i8 -1, i8 2, i8 3, i8 -1, i8 -1, i8 4, i8 5, i8 -1, i8 -1, i8 6, i8 7, i8 -1, i8 -1, i8 16, i8 17, i8 -1, i8 -1, i8 18, i8 19, i8 -1, i8 -1, i8 20, i8 21, i8 -1, i8 -1, i8 22, i8 23, i8 -1, i8 -1>) ret <32 x i8> %1 } define <32 x i8> @combine_pshufb_as_unpackhi_zero(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_as_unpackhi_zero: -; X32: # %bb.0: -; X32-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X32-NEXT: vpunpckhbw {{.*#+}} ymm0 = ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15],ymm1[24],ymm0[24],ymm1[25],ymm0[25],ymm1[26],ymm0[26],ymm1[27],ymm0[27],ymm1[28],ymm0[28],ymm1[29],ymm0[29],ymm1[30],ymm0[30],ymm1[31],ymm0[31] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_as_unpackhi_zero: -; X64: # %bb.0: -; X64-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; X64-NEXT: vpunpckhbw {{.*#+}} ymm0 = ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15],ymm1[24],ymm0[24],ymm1[25],ymm0[25],ymm1[26],ymm0[26],ymm1[27],ymm0[27],ymm1[28],ymm0[28],ymm1[29],ymm0[29],ymm1[30],ymm0[30],ymm1[31],ymm0[31] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_as_unpackhi_zero: +; CHECK: # %bb.0: +; CHECK-NEXT: vpxor %xmm1, %xmm1, %xmm1 +; CHECK-NEXT: vpunpckhbw {{.*#+}} ymm0 = ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15],ymm1[24],ymm0[24],ymm1[25],ymm0[25],ymm1[26],ymm0[26],ymm1[27],ymm0[27],ymm1[28],ymm0[28],ymm1[29],ymm0[29],ymm1[30],ymm0[30],ymm1[31],ymm0[31] +; CHECK-NEXT: ret{{[l|q]}} %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 -1, i8 8, i8 -1, i8 9, i8 -1, i8 10, i8 -1, i8 11, i8 -1, i8 12, i8 -1, i8 13, i8 -1, i8 14, i8 -1, i8 15, i8 -1, i8 24, i8 -1, i8 25, i8 -1, i8 26, i8 -1, i8 27, i8 -1, i8 28, i8 -1, i8 29, i8 -1, i8 30, i8 -1, i8 31>) ret <32 x i8> %1 } define <32 x i8> @combine_psrlw_pshufb(<16 x i16> %a0) { -; X32-LABEL: combine_psrlw_pshufb: -; X32: # %bb.0: -; X32-NEXT: vandps {{\.LCPI.*}}, %ymm0, %ymm0 -; X32-NEXT: retl +; X86-LABEL: combine_psrlw_pshufb: +; X86: # %bb.0: +; X86-NEXT: vandps {{\.LCPI.*}}, %ymm0, %ymm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_psrlw_pshufb: ; X64: # %bb.0: @@ -730,10 +519,10 @@ define <32 x i8> @combine_psrlw_pshufb(<16 x i16> %a0) { } define <32 x i8> @combine_pslld_pshufb(<8 x i32> %a0) { -; X32-LABEL: combine_pslld_pshufb: -; X32: # %bb.0: -; X32-NEXT: vandps {{\.LCPI.*}}, %ymm0, %ymm0 -; X32-NEXT: retl +; X86-LABEL: combine_pslld_pshufb: +; X86: # %bb.0: +; X86-NEXT: vandps {{\.LCPI.*}}, %ymm0, %ymm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_pslld_pshufb: ; X64: # %bb.0: @@ -746,15 +535,10 @@ define <32 x i8> @combine_pslld_pshufb(<8 x i32> %a0) { } define <32 x i8> @combine_psrlq_pshufb(<4 x i64> %a0) { -; X32-LABEL: combine_psrlq_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[7,6,5,4],zero,zero,zero,zero,ymm0[15,14,13,12],zero,zero,zero,zero,ymm0[23,22,21],zero,zero,zero,zero,ymm0[31,30,29,28],zero -; X32-NEXT: retl -; -; X64-LABEL: combine_psrlq_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[7,6,5,4],zero,zero,zero,zero,ymm0[15,14,13,12],zero,zero,zero,zero,ymm0[23,22,21],zero,zero,zero,zero,ymm0[31,30,29,28],zero -; X64-NEXT: retq +; CHECK-LABEL: combine_psrlq_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[7,6,5,4],zero,zero,zero,zero,ymm0[15,14,13,12],zero,zero,zero,zero,ymm0[23,22,21],zero,zero,zero,zero,ymm0[31,30,29,28],zero +; CHECK-NEXT: ret{{[l|q]}} %1 = lshr <4 x i64> %a0, <i64 32, i64 32, i64 32, i64 32> %2 = bitcast <4 x i64> %1 to <32 x i8> %3 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %2, <32 x i8> <i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 15, i8 14, i8 13, i8 12, i8 11, i8 10, i8 9, i8 8, i8 23, i8 22, i8 21, i8 20, i8 19, i8 18, i8 17, i8 31, i8 30, i8 29, i8 28, i8 27, i8 26, i8 25, i8 24, i8 23>) @@ -762,15 +546,10 @@ define <32 x i8> @combine_psrlq_pshufb(<4 x i64> %a0) { } define <32 x i8> @combine_unpack_unpack_pshufb(<32 x i8> %a0) { -; X32-LABEL: combine_unpack_unpack_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,0,4,8,1,1,5,9,2,2,6,10,3,3,7,11,16,16,20,24,17,17,21,25,18,18,22,26,19,19,23,27] -; X32-NEXT: retl -; -; X64-LABEL: combine_unpack_unpack_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,0,4,8,1,1,5,9,2,2,6,10,3,3,7,11,16,16,20,24,17,17,21,25,18,18,22,26,19,19,23,27] -; X64-NEXT: retq +; CHECK-LABEL: combine_unpack_unpack_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,0,4,8,1,1,5,9,2,2,6,10,3,3,7,11,16,16,20,24,17,17,21,25,18,18,22,26,19,19,23,27] +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 16, i32 17, i32 18, i32 19, i32 16, i32 17, i32 18, i32 19, i32 16, i32 17, i32 18, i32 19> %2 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 20, i32 21, i32 22, i32 23, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef> %3 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 24, i32 25, i32 26, i32 27, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef> @@ -781,17 +560,11 @@ define <32 x i8> @combine_unpack_unpack_pshufb(<32 x i8> %a0) { } define <16 x i16> @shuffle_combine_packssdw_pshufb(<8 x i32> %a0) { -; X32-LABEL: shuffle_combine_packssdw_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpsrad $31, %ymm0, %ymm0 -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[12,13,8,9,4,5,0,1,12,13,8,9,4,5,0,1,16,17,20,21,24,25,28,29,28,29,24,25,20,21,16,17] -; X32-NEXT: retl -; -; X64-LABEL: shuffle_combine_packssdw_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpsrad $31, %ymm0, %ymm0 -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[12,13,8,9,4,5,0,1,12,13,8,9,4,5,0,1,16,17,20,21,24,25,28,29,28,29,24,25,20,21,16,17] -; X64-NEXT: retq +; CHECK-LABEL: shuffle_combine_packssdw_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpsrad $31, %ymm0, %ymm0 +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[12,13,8,9,4,5,0,1,12,13,8,9,4,5,0,1,16,17,20,21,24,25,28,29,28,29,24,25,20,21,16,17] +; CHECK-NEXT: ret{{[l|q]}} %1 = ashr <8 x i32> %a0, <i32 31, i32 31, i32 31, i32 31, i32 31, i32 31, i32 31, i32 31> %2 = tail call <16 x i16> @llvm.x86.avx2.packssdw(<8 x i32> %1, <8 x i32> %1) %3 = shufflevector <16 x i16> %2, <16 x i16> undef, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0, i32 8, i32 9, i32 10, i32 11, i32 11, i32 10, i32 9, i32 8> @@ -800,17 +573,11 @@ define <16 x i16> @shuffle_combine_packssdw_pshufb(<8 x i32> %a0) { declare <16 x i16> @llvm.x86.avx2.packssdw(<8 x i32>, <8 x i32>) nounwind readnone define <32 x i8> @shuffle_combine_packsswb_pshufb(<16 x i16> %a0, <16 x i16> %a1) { -; X32-LABEL: shuffle_combine_packsswb_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpsraw $15, %ymm0, %ymm0 -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,12,10,8,6,4,2,0,14,12,10,8,6,4,2,0,30,28,26,24,22,20,18,16,30,28,26,24,22,20,18,16] -; X32-NEXT: retl -; -; X64-LABEL: shuffle_combine_packsswb_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpsraw $15, %ymm0, %ymm0 -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,12,10,8,6,4,2,0,14,12,10,8,6,4,2,0,30,28,26,24,22,20,18,16,30,28,26,24,22,20,18,16] -; X64-NEXT: retq +; CHECK-LABEL: shuffle_combine_packsswb_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpsraw $15, %ymm0, %ymm0 +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,12,10,8,6,4,2,0,14,12,10,8,6,4,2,0,30,28,26,24,22,20,18,16,30,28,26,24,22,20,18,16] +; CHECK-NEXT: ret{{[l|q]}} %1 = ashr <16 x i16> %a0, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15> %2 = ashr <16 x i16> %a1, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15> %3 = tail call <32 x i8> @llvm.x86.avx2.packsswb(<16 x i16> %1, <16 x i16> %2) @@ -820,15 +587,10 @@ define <32 x i8> @shuffle_combine_packsswb_pshufb(<16 x i16> %a0, <16 x i16> %a1 declare <32 x i8> @llvm.x86.avx2.packsswb(<16 x i16>, <16 x i16>) nounwind readnone define <16 x i16> @shuffle_combine_packusdw_pshufb(<8 x i32> %a0, <8 x i32> %a1) { -; X32-LABEL: shuffle_combine_packusdw_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,15,10,11,6,7,2,3,14,15,10,11,6,7,2,3,18,19,22,23,26,27,30,31,30,31,26,27,22,23,18,19] -; X32-NEXT: retl -; -; X64-LABEL: shuffle_combine_packusdw_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,15,10,11,6,7,2,3,14,15,10,11,6,7,2,3,18,19,22,23,26,27,30,31,30,31,26,27,22,23,18,19] -; X64-NEXT: retq +; CHECK-LABEL: shuffle_combine_packusdw_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[14,15,10,11,6,7,2,3,14,15,10,11,6,7,2,3,18,19,22,23,26,27,30,31,30,31,26,27,22,23,18,19] +; CHECK-NEXT: ret{{[l|q]}} %1 = lshr <8 x i32> %a0, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16> %2 = tail call <16 x i16> @llvm.x86.avx2.packusdw(<8 x i32> %1, <8 x i32> %1) %3 = shufflevector <16 x i16> %2, <16 x i16> undef, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0, i32 8, i32 9, i32 10, i32 11, i32 11, i32 10, i32 9, i32 8> @@ -837,15 +599,10 @@ define <16 x i16> @shuffle_combine_packusdw_pshufb(<8 x i32> %a0, <8 x i32> %a1) declare <16 x i16> @llvm.x86.avx2.packusdw(<8 x i32>, <8 x i32>) nounwind readnone define <32 x i8> @shuffle_combine_packuswb_pshufb(<16 x i16> %a0, <16 x i16> %a1) { -; X32-LABEL: shuffle_combine_packuswb_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[15,13,11,9,7,5,3,1,15,13,11,9,7,5,3,1,31,29,27,25,23,21,19,17,31,29,27,25,23,21,19,17] -; X32-NEXT: retl -; -; X64-LABEL: shuffle_combine_packuswb_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[15,13,11,9,7,5,3,1,15,13,11,9,7,5,3,1,31,29,27,25,23,21,19,17,31,29,27,25,23,21,19,17] -; X64-NEXT: retq +; CHECK-LABEL: shuffle_combine_packuswb_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[15,13,11,9,7,5,3,1,15,13,11,9,7,5,3,1,31,29,27,25,23,21,19,17,31,29,27,25,23,21,19,17] +; CHECK-NEXT: ret{{[l|q]}} %1 = lshr <16 x i16> %a0, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8> %2 = lshr <16 x i16> %a1, <i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8, i16 8> %3 = tail call <32 x i8> @llvm.x86.avx2.packuswb(<16 x i16> %1, <16 x i16> %2) @@ -855,10 +612,10 @@ define <32 x i8> @shuffle_combine_packuswb_pshufb(<16 x i16> %a0, <16 x i16> %a1 declare <32 x i8> @llvm.x86.avx2.packuswb(<16 x i16>, <16 x i16>) nounwind readnone define <16 x i8> @combine_pshufb_insertion_as_broadcast_v2i64(i64 %a0) { -; X32-LABEL: combine_pshufb_insertion_as_broadcast_v2i64: -; X32: # %bb.0: -; X32-NEXT: vpbroadcastq {{[0-9]+}}(%esp), %xmm0 -; X32-NEXT: retl +; X86-LABEL: combine_pshufb_insertion_as_broadcast_v2i64: +; X86: # %bb.0: +; X86-NEXT: vpbroadcastq {{[0-9]+}}(%esp), %xmm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_pshufb_insertion_as_broadcast_v2i64: ; X64: # %bb.0: @@ -872,10 +629,10 @@ define <16 x i8> @combine_pshufb_insertion_as_broadcast_v2i64(i64 %a0) { } define <8 x i32> @combine_permd_insertion_as_broadcast_v4i64(i64 %a0) { -; X32-LABEL: combine_permd_insertion_as_broadcast_v4i64: -; X32: # %bb.0: -; X32-NEXT: vbroadcastsd {{[0-9]+}}(%esp), %ymm0 -; X32-NEXT: retl +; X86-LABEL: combine_permd_insertion_as_broadcast_v4i64: +; X86: # %bb.0: +; X86-NEXT: vbroadcastsd {{[0-9]+}}(%esp), %ymm0 +; X86-NEXT: retl ; ; X64-LABEL: combine_permd_insertion_as_broadcast_v4i64: ; X64: # %bb.0: @@ -889,15 +646,10 @@ define <8 x i32> @combine_permd_insertion_as_broadcast_v4i64(i64 %a0) { } define <32 x i8> @combine_pshufb_pshufb_or_as_blend(<32 x i8> %a0, <32 x i8> %a1) { -; X32-LABEL: combine_pshufb_pshufb_or_as_blend: -; X32: # %bb.0: -; X32-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_pshufb_or_as_blend: -; X64: # %bb.0: -; X64-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_pshufb_or_as_blend: +; CHECK: # %bb.0: +; CHECK-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] +; CHECK-NEXT: ret{{[l|q]}} %1 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>) %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a1, <32 x i8> <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>) %3 = or <32 x i8> %1, %2 @@ -905,15 +657,10 @@ define <32 x i8> @combine_pshufb_pshufb_or_as_blend(<32 x i8> %a0, <32 x i8> %a1 } define <32 x i8> @combine_pshufb_pshufb_or_as_unpcklbw(<32 x i8> %a0, <32 x i8> %a1) { -; X32-LABEL: combine_pshufb_pshufb_or_as_unpcklbw: -; X32: # %bb.0: -; X32-NEXT: vpunpcklbw {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_pshufb_or_as_unpcklbw: -; X64: # %bb.0: -; X64-NEXT: vpunpcklbw {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_pshufb_or_as_unpcklbw: +; CHECK: # %bb.0: +; CHECK-NEXT: vpunpcklbw {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23] +; CHECK-NEXT: ret{{[l|q]}} %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 0, i8 -1, i8 1, i8 -1, i8 2, i8 -1, i8 3, i8 -1, i8 4, i8 -1, i8 5, i8 -1, i8 6, i8 -1, i8 7, i8 -1, i8 0, i8 -1, i8 1, i8 -1, i8 2, i8 -1, i8 3, i8 -1, i8 4, i8 -1, i8 5, i8 -1, i8 6, i8 -1, i8 7, i8 -1>) %2 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a1, <32 x i8> <i8 -1, i8 0, i8 -1, i8 1, i8 -1, i8 2, i8 -1, i8 3, i8 -1, i8 4, i8 -1, i8 5, i8 -1, i8 6, i8 -1, i8 7, i8 -1, i8 0, i8 -1, i8 1, i8 -1, i8 2, i8 -1, i8 3, i8 -1, i8 4, i8 -1, i8 5, i8 -1, i8 6, i8 -1, i8 7>) %3 = or <32 x i8> %1, %2 @@ -921,15 +668,10 @@ define <32 x i8> @combine_pshufb_pshufb_or_as_unpcklbw(<32 x i8> %a0, <32 x i8> } define <32 x i8> @combine_pshufb_pshufb_or_pshufb(<32 x i8> %a0) { -; X32-LABEL: combine_pshufb_pshufb_or_pshufb: -; X32: # %bb.0: -; X32-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,0,0,0,4,4,4,4] -; X32-NEXT: retl -; -; X64-LABEL: combine_pshufb_pshufb_or_pshufb: -; X64: # %bb.0: -; X64-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,0,0,0,4,4,4,4] -; X64-NEXT: retq +; CHECK-LABEL: combine_pshufb_pshufb_or_pshufb: +; CHECK: # %bb.0: +; CHECK-NEXT: vpermilps {{.*#+}} ymm0 = ymm0[0,0,0,0,4,4,4,4] +; CHECK-NEXT: ret{{[l|q]}} %1 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1>) %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %a0, <32 x i8> <i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3, i8 -1, i8 -1, i8 -1, i8 -1, i8 0, i8 1, i8 2, i8 3>) %3 = or <32 x i8> %1, %2 @@ -938,59 +680,38 @@ define <32 x i8> @combine_pshufb_pshufb_or_pshufb(<32 x i8> %a0) { } define <8 x i32> @constant_fold_permd() { -; X32-LABEL: constant_fold_permd: -; X32: # %bb.0: -; X32-NEXT: vmovaps {{.*#+}} ymm0 = [5,7,3,2,8,2,6,1] -; X32-NEXT: retl -; -; X64-LABEL: constant_fold_permd: -; X64: # %bb.0: -; X64-NEXT: vmovaps {{.*#+}} ymm0 = [5,7,3,2,8,2,6,1] -; X64-NEXT: retq +; CHECK-LABEL: constant_fold_permd: +; CHECK: # %bb.0: +; CHECK-NEXT: vmovaps {{.*#+}} ymm0 = [5,7,3,2,8,2,6,1] +; CHECK-NEXT: ret{{[l|q]}} %1 = call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>, <8 x i32> <i32 4, i32 6, i32 2, i32 1, i32 7, i32 1, i32 5, i32 0>) ret <8 x i32> %1 } define <8 x float> @constant_fold_permps() { -; X32-LABEL: constant_fold_permps: -; X32: # %bb.0: -; X32-NEXT: vmovaps {{.*#+}} ymm0 = [5.000000e+00,7.000000e+00,3.000000e+00,2.000000e+00,8.000000e+00,2.000000e+00,6.000000e+00,1.000000e+00] -; X32-NEXT: retl -; -; X64-LABEL: constant_fold_permps: -; X64: # %bb.0: -; X64-NEXT: vmovaps {{.*#+}} ymm0 = [5.000000e+00,7.000000e+00,3.000000e+00,2.000000e+00,8.000000e+00,2.000000e+00,6.000000e+00,1.000000e+00] -; X64-NEXT: retq +; CHECK-LABEL: constant_fold_permps: +; CHECK: # %bb.0: +; CHECK-NEXT: vmovaps {{.*#+}} ymm0 = [5.000000e+00,7.000000e+00,3.000000e+00,2.000000e+00,8.000000e+00,2.000000e+00,6.000000e+00,1.000000e+00] +; CHECK-NEXT: ret{{[l|q]}} %1 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0>, <8 x i32> <i32 4, i32 6, i32 2, i32 1, i32 7, i32 1, i32 5, i32 0>) ret <8 x float> %1 } define <32 x i8> @constant_fold_pshufb_256() { -; X32-LABEL: constant_fold_pshufb_256: -; X32: # %bb.0: -; X32-NEXT: vmovaps {{.*#+}} ymm0 = <14,0,0,0,u,u,0,0,0,0,0,0,0,0,8,9,255,0,0,0,u,u,0,0,241,0,0,0,0,0,249,250> -; X32-NEXT: retl -; -; X64-LABEL: constant_fold_pshufb_256: -; X64: # %bb.0: -; X64-NEXT: vmovaps {{.*#+}} ymm0 = <14,0,0,0,u,u,0,0,0,0,0,0,0,0,8,9,255,0,0,0,u,u,0,0,241,0,0,0,0,0,249,250> -; X64-NEXT: retq +; CHECK-LABEL: constant_fold_pshufb_256: +; CHECK: # %bb.0: +; CHECK-NEXT: vmovaps {{.*#+}} ymm0 = <14,0,0,0,u,u,0,0,0,0,0,0,0,0,8,9,255,0,0,0,u,u,0,0,241,0,0,0,0,0,249,250> +; CHECK-NEXT: ret{{[l|q]}} %1 = tail call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> <i8 15, i8 14, i8 13, i8 12, i8 11, i8 10, i8 9, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 -1, i8 -2, i8 -3, i8 -4, i8 -5, i8 -6, i8 -7, i8 -8, i8 -9, i8 -10, i8 -11, i8 -12, i8 -13, i8 -14, i8 -15>, <32 x i8> <i8 1, i8 -1, i8 -1, i8 -1, i8 undef, i8 undef, i8 -1, i8 -1, i8 15, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 7, i8 6, i8 1, i8 -1, i8 -1, i8 -1, i8 undef, i8 undef, i8 -1, i8 -1, i8 15, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 7, i8 6>) ret <32 x i8> %1 } define <32 x i8> @PR27320(<8 x i32> %a0) { -; X32-LABEL: PR27320: -; X32: # %bb.0: -; X32-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1] -; X32-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,1,2,3,4,4,5,6,7,7,8,9,10,10,11,28,29,29,30,31,16,16,17,18,19,19,20,21,22,22,23] -; X32-NEXT: retl -; -; X64-LABEL: PR27320: -; X64: # %bb.0: -; X64-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1] -; X64-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,1,2,3,4,4,5,6,7,7,8,9,10,10,11,28,29,29,30,31,16,16,17,18,19,19,20,21,22,22,23] -; X64-NEXT: retq +; CHECK-LABEL: PR27320: +; CHECK: # %bb.0: +; CHECK-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,1,2,1] +; CHECK-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,1,2,3,4,4,5,6,7,7,8,9,10,10,11,28,29,29,30,31,16,16,17,18,19,19,20,21,22,22,23] +; CHECK-NEXT: ret{{[l|q]}} %1 = shufflevector <8 x i32> %a0, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 undef, i32 3, i32 4, i32 5, i32 undef> %2 = bitcast <8 x i32> %1 to <32 x i8> %3 = shufflevector <32 x i8> %2, <32 x i8> undef, <32 x i32> <i32 0, i32 1, i32 1, i32 2, i32 3, i32 4, i32 4, i32 5, i32 6, i32 7, i32 7, i32 8, i32 9, i32 10, i32 10, i32 11, i32 16, i32 17, i32 17, i32 18, i32 19, i32 20, i32 20, i32 21, i32 22, i32 23, i32 23, i32 24, i32 25, i32 26, i32 26, i32 27> @@ -998,25 +719,15 @@ define <32 x i8> @PR27320(<8 x i32> %a0) { } define internal fastcc <8 x float> @PR34577(<8 x float> %inp0, <8 x float> %inp1, <8 x float> %inp2) { -; X32-LABEL: PR34577: -; X32: # %bb.0: # %entry -; X32-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3] -; X32-NEXT: vxorps %xmm2, %xmm2, %xmm2 -; X32-NEXT: vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7] -; X32-NEXT: vmovaps {{.*#+}} ymm2 = <u,u,7,2,u,u,3,2> -; X32-NEXT: vpermps %ymm1, %ymm2, %ymm1 -; X32-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] -; X32-NEXT: retl -; -; X64-LABEL: PR34577: -; X64: # %bb.0: # %entry -; X64-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3] -; X64-NEXT: vxorps %xmm2, %xmm2, %xmm2 -; X64-NEXT: vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7] -; X64-NEXT: vmovaps {{.*#+}} ymm2 = <u,u,7,2,u,u,3,2> -; X64-NEXT: vpermps %ymm1, %ymm2, %ymm1 -; X64-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] -; X64-NEXT: retq +; CHECK-LABEL: PR34577: +; CHECK: # %bb.0: # %entry +; CHECK-NEXT: vpermpd {{.*#+}} ymm0 = ymm0[0,1,1,3] +; CHECK-NEXT: vxorps %xmm2, %xmm2, %xmm2 +; CHECK-NEXT: vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7] +; CHECK-NEXT: vmovaps {{.*#+}} ymm2 = <u,u,7,2,u,u,3,2> +; CHECK-NEXT: vpermps %ymm1, %ymm2, %ymm1 +; CHECK-NEXT: vblendps {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7] +; CHECK-NEXT: ret{{[l|q]}} entry: %shuf0 = shufflevector <8 x float> %inp0, <8 x float> %inp2, <8 x i32> <i32 1, i32 10, i32 11, i32 13, i32 2, i32 13, i32 5, i32 0> %sel = select <8 x i1> <i1 false, i1 true, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false>, <8 x float> %shuf0, <8 x float> zeroinitializer |
