| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6 | |
| ; RUN: llc -mtriple=x86_64-- -mcpu=x86-64 < %s | FileCheck %s --check-prefixes=SSE,SSE2 | |
| ; RUN: llc -mtriple=x86_64-- -mcpu=x86-64-v2 < %s | FileCheck %s --check-prefixes=SSE,SSE42 | |
| ; RUN: llc -mtriple=x86_64-- -mcpu=x86-64-v3 < %s | FileCheck %s --check-prefixes=AVX,AVX2 | |
| ; RUN: llc -mtriple=x86_64-- -mcpu=x86-64-v4 < %s | FileCheck %s --check-prefixes=AVX,AVX512 | |
| define <4 x i8> @test_bitextract_v4i8_const(<4 x b32> %src) { | |
| ; SSE2-LABEL: test_bitextract_v4i8_const: | |
| ; SSE2: # %bb.0: | |
| ; SSE2-NEXT: psrld $8, %xmm0 | |
| ; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0 | |
| ; SSE2-NEXT: packuswb %xmm0, %xmm0 | |
| ; SSE2-NEXT: packuswb %xmm0, %xmm0 | |
| ; SSE2-NEXT: retq | |
| ; | |
| ; SSE42-LABEL: test_bitextract_v4i8_const: | |
| ; SSE42: # %bb.0: | |
| ; SSE42-NEXT: pshufb {{.*#+}} xmm0 = xmm0[1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u] | |
| ; SSE42-NEXT: retq | |
| ; | |
| ; AVX2-LABEL: test_bitextract_v4i8_const: | |
| ; AVX2: # %bb.0: | |
| ; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u] | |
| ; AVX2-NEXT: retq | |
| ; | |
| ; AVX512-LABEL: test_bitextract_v4i8_const: | |
| ; AVX512: # %bb.0: | |
| ; AVX512-NEXT: vpsrld $8, %xmm0, %xmm0 | |
| ; AVX512-NEXT: vpmovdb %xmm0, %xmm0 | |
| ; AVX512-NEXT: retq | |
| %result = bitextract <4 x i8>, <4 x b32> %src, <4 x i32> splat (i32 8) | |
| ret <4 x i8> %result | |
| } | |
| define <4 x i8> @test_bitextract_v4i8_var(<4 x b32> %src, <4 x i32> %off) { | |
| ; SSE2-LABEL: test_bitextract_v4i8_var: | |
| ; SSE2: # %bb.0: | |
| ; SSE2-NEXT: pshuflw {{.*#+}} xmm2 = xmm1[2,3,3,3,4,5,6,7] | |
| ; SSE2-NEXT: movdqa %xmm0, %xmm3 | |
| ; SSE2-NEXT: psrld %xmm2, %xmm3 | |
| ; SSE2-NEXT: pshuflw {{.*#+}} xmm4 = xmm1[0,1,1,1,4,5,6,7] | |
| ; SSE2-NEXT: movdqa %xmm0, %xmm2 | |
| ; SSE2-NEXT: psrld %xmm4, %xmm2 | |
| ; SSE2-NEXT: pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3] | |
| ; SSE2-NEXT: pshuflw {{.*#+}} xmm4 = xmm1[2,3,3,3,4,5,6,7] | |
| ; SSE2-NEXT: movdqa %xmm0, %xmm5 | |
| ; SSE2-NEXT: psrld %xmm4, %xmm5 | |
| ; SSE2-NEXT: punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm3[0] | |
| ; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,1,1,1,4,5,6,7] | |
| ; SSE2-NEXT: psrld %xmm1, %xmm0 | |
| ; SSE2-NEXT: punpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm5[1] | |
| ; SSE2-NEXT: shufps {{.*#+}} xmm2 = xmm2[0,3],xmm0[0,3] | |
| ; SSE2-NEXT: andps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2 | |
| ; SSE2-NEXT: packuswb %xmm2, %xmm2 | |
| ; SSE2-NEXT: packuswb %xmm2, %xmm2 | |
| ; SSE2-NEXT: movdqa %xmm2, %xmm0 | |
| ; SSE2-NEXT: retq | |
| ; | |
| ; SSE42-LABEL: test_bitextract_v4i8_var: | |
| ; SSE42: # %bb.0: | |
| ; SSE42-NEXT: pshuflw {{.*#+}} xmm2 = xmm1[2,3,3,3,4,5,6,7] | |
| ; SSE42-NEXT: movdqa %xmm0, %xmm3 | |
| ; SSE42-NEXT: psrld %xmm2, %xmm3 | |
| ; SSE42-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3] | |
| ; SSE42-NEXT: pshuflw {{.*#+}} xmm4 = xmm2[2,3,3,3,4,5,6,7] | |
| ; SSE42-NEXT: movdqa %xmm0, %xmm5 | |
| ; SSE42-NEXT: psrld %xmm4, %xmm5 | |
| ; SSE42-NEXT: pblendw {{.*#+}} xmm5 = xmm3[0,1,2,3],xmm5[4,5,6,7] | |
| ; SSE42-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,1,1,1,4,5,6,7] | |
| ; SSE42-NEXT: movdqa %xmm0, %xmm3 | |
| ; SSE42-NEXT: psrld %xmm1, %xmm3 | |
| ; SSE42-NEXT: pshuflw {{.*#+}} xmm1 = xmm2[0,1,1,1,4,5,6,7] | |
| ; SSE42-NEXT: psrld %xmm1, %xmm0 | |
| ; SSE42-NEXT: pblendw {{.*#+}} xmm0 = xmm3[0,1,2,3],xmm0[4,5,6,7] | |
| ; SSE42-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm5[2,3],xmm0[4,5],xmm5[6,7] | |
| ; SSE42-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u] | |
| ; SSE42-NEXT: retq | |
| ; | |
| ; AVX2-LABEL: test_bitextract_v4i8_var: | |
| ; AVX2: # %bb.0: | |
| ; AVX2-NEXT: vpsrlvd %xmm1, %xmm0, %xmm0 | |
| ; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u] | |
| ; AVX2-NEXT: retq | |
| ; | |
| ; AVX512-LABEL: test_bitextract_v4i8_var: | |
| ; AVX512: # %bb.0: | |
| ; AVX512-NEXT: vpsrlvd %xmm1, %xmm0, %xmm0 | |
| ; AVX512-NEXT: vpmovdb %xmm0, %xmm0 | |
| ; AVX512-NEXT: retq | |
| %result = bitextract <4 x i8>, <4 x b32> %src, <4 x i32> %off | |
| ret <4 x i8> %result | |
| } | |
| define <2 x float> @test_bitextract_v2f32(<2 x b64> %src) { | |
| ; SSE2-LABEL: test_bitextract_v2f32: | |
| ; SSE2: # %bb.0: | |
| ; SSE2-NEXT: movdqa %xmm0, %xmm1 | |
| ; SSE2-NEXT: psrlq $32, %xmm1 | |
| ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3] | |
| ; SSE2-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2,2,3] | |
| ; SSE2-NEXT: retq | |
| ; | |
| ; SSE42-LABEL: test_bitextract_v2f32: | |
| ; SSE42: # %bb.0: | |
| ; SSE42-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,12,13,14,15,12,13,14,15],zero,zero,zero,zero | |
| ; SSE42-NEXT: retq | |
| ; | |
| ; AVX-LABEL: test_bitextract_v2f32: | |
| ; AVX: # %bb.0: | |
| ; AVX-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,12,13,14,15,12,13,14,15],zero,zero,zero,zero | |
| ; AVX-NEXT: retq | |
| %result = bitextract <2 x float>, <2 x b64> %src, <2 x i32> <i32 0, i32 32> | |
| ret <2 x float> %result | |
| } | |
| define <2 x ptr> @test_bitextract_v2ptr(<2 x b64> %src) { | |
| ; SSE-LABEL: test_bitextract_v2ptr: | |
| ; SSE: # %bb.0: | |
| ; SSE-NEXT: retq | |
| ; | |
| ; AVX-LABEL: test_bitextract_v2ptr: | |
| ; AVX: # %bb.0: | |
| ; AVX-NEXT: retq | |
| %result = bitextract <2 x ptr>, <2 x b64> %src, <2 x i32> zeroinitializer | |
| ret <2 x ptr> %result | |
| } | |
| define <16 x i4> @test_bitextract_v16i4_var(<16 x b8> %src, <16 x i32> %off) { | |
| ; SSE2-LABEL: test_bitextract_v16i4_var: | |
| ; SSE2: # %bb.0: | |
| ; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0] | |
| ; SSE2-NEXT: pand %xmm5, %xmm4 | |
| ; SSE2-NEXT: pand %xmm5, %xmm3 | |
| ; SSE2-NEXT: packuswb %xmm4, %xmm3 | |
| ; SSE2-NEXT: pand %xmm5, %xmm2 | |
| ; SSE2-NEXT: pand %xmm5, %xmm1 | |
| ; SSE2-NEXT: packuswb %xmm2, %xmm1 | |
| ; SSE2-NEXT: packuswb %xmm3, %xmm1 | |
| ; SSE2-NEXT: psllw $5, %xmm1 | |
| ; SSE2-NEXT: pxor %xmm2, %xmm2 | |
| ; SSE2-NEXT: pcmpgtb %xmm1, %xmm2 | |
| ; SSE2-NEXT: movdqa %xmm2, %xmm3 | |
| ; SSE2-NEXT: pandn %xmm0, %xmm3 | |
| ; SSE2-NEXT: psrlw $4, %xmm0 | |
| ; SSE2-NEXT: pand %xmm2, %xmm0 | |
| ; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0 | |
| ; SSE2-NEXT: por %xmm3, %xmm0 | |
| ; SSE2-NEXT: paddb %xmm1, %xmm1 | |
| ; SSE2-NEXT: pxor %xmm2, %xmm2 | |
| ; SSE2-NEXT: pcmpgtb %xmm1, %xmm2 | |
| ; SSE2-NEXT: movdqa %xmm2, %xmm3 | |
| ; SSE2-NEXT: pandn %xmm0, %xmm3 | |
| ; SSE2-NEXT: psrlw $2, %xmm0 | |
| ; SSE2-NEXT: pand %xmm2, %xmm0 | |
| ; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0 | |
| ; SSE2-NEXT: pxor %xmm2, %xmm2 | |
| ; SSE2-NEXT: por %xmm3, %xmm0 | |
| ; SSE2-NEXT: paddb %xmm1, %xmm1 | |
| ; SSE2-NEXT: pcmpgtb %xmm1, %xmm2 | |
| ; SSE2-NEXT: movdqa %xmm2, %xmm1 | |
| ; SSE2-NEXT: pandn %xmm0, %xmm1 | |
| ; SSE2-NEXT: psrlw $1, %xmm0 | |
| ; SSE2-NEXT: pand %xmm2, %xmm0 | |
| ; SSE2-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0 | |
| ; SSE2-NEXT: por %xmm1, %xmm0 | |
| ; SSE2-NEXT: retq | |
| ; | |
| ; SSE42-LABEL: test_bitextract_v16i4_var: | |
| ; SSE42: # %bb.0: | |
| ; SSE42-NEXT: movdqa %xmm0, %xmm5 | |
| ; SSE42-NEXT: movdqa {{.*#+}} xmm0 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0] | |
| ; SSE42-NEXT: pand %xmm0, %xmm4 | |
| ; SSE42-NEXT: pand %xmm0, %xmm3 | |
| ; SSE42-NEXT: packusdw %xmm4, %xmm3 | |
| ; SSE42-NEXT: pand %xmm0, %xmm2 | |
| ; SSE42-NEXT: pand %xmm1, %xmm0 | |
| ; SSE42-NEXT: packusdw %xmm2, %xmm0 | |
| ; SSE42-NEXT: packuswb %xmm3, %xmm0 | |
| ; SSE42-NEXT: psllw $5, %xmm0 | |
| ; SSE42-NEXT: movdqa %xmm0, %xmm1 | |
| ; SSE42-NEXT: movdqa %xmm5, %xmm2 | |
| ; SSE42-NEXT: psrlw $4, %xmm2 | |
| ; SSE42-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2 | |
| ; SSE42-NEXT: paddb %xmm0, %xmm1 | |
| ; SSE42-NEXT: pblendvb %xmm0, %xmm2, %xmm5 | |
| ; SSE42-NEXT: movdqa %xmm5, %xmm2 | |
| ; SSE42-NEXT: psrlw $2, %xmm2 | |
| ; SSE42-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2 | |
| ; SSE42-NEXT: movdqa %xmm1, %xmm0 | |
| ; SSE42-NEXT: pblendvb %xmm0, %xmm2, %xmm5 | |
| ; SSE42-NEXT: movdqa %xmm5, %xmm2 | |
| ; SSE42-NEXT: psrlw $1, %xmm2 | |
| ; SSE42-NEXT: pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2 | |
| ; SSE42-NEXT: paddb %xmm1, %xmm1 | |
| ; SSE42-NEXT: movdqa %xmm1, %xmm0 | |
| ; SSE42-NEXT: pblendvb %xmm0, %xmm2, %xmm5 | |
| ; SSE42-NEXT: movdqa %xmm5, %xmm0 | |
| ; SSE42-NEXT: retq | |
| ; | |
| ; AVX2-LABEL: test_bitextract_v16i4_var: | |
| ; AVX2: # %bb.0: | |
| ; AVX2-NEXT: vpbroadcastd {{.*#+}} ymm3 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0] | |
| ; AVX2-NEXT: vpand %ymm3, %ymm2, %ymm2 | |
| ; AVX2-NEXT: vpand %ymm3, %ymm1, %ymm1 | |
| ; AVX2-NEXT: vpackusdw %ymm2, %ymm1, %ymm1 | |
| ; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 | |
| ; AVX2-NEXT: vpackuswb %xmm2, %xmm1, %xmm1 | |
| ; AVX2-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,1,3] | |
| ; AVX2-NEXT: vpsllw $5, %xmm1, %xmm1 | |
| ; AVX2-NEXT: vpsrlw $4, %xmm0, %xmm2 | |
| ; AVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2, %xmm2 | |
| ; AVX2-NEXT: vpaddb %xmm1, %xmm1, %xmm3 | |
| ; AVX2-NEXT: vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 | |
| ; AVX2-NEXT: vpsrlw $2, %xmm0, %xmm1 | |
| ; AVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 | |
| ; AVX2-NEXT: vpblendvb %xmm3, %xmm1, %xmm0, %xmm0 | |
| ; AVX2-NEXT: vpsrlw $1, %xmm0, %xmm1 | |
| ; AVX2-NEXT: vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1 | |
| ; AVX2-NEXT: vpaddb %xmm3, %xmm3, %xmm2 | |
| ; AVX2-NEXT: vpblendvb %xmm2, %xmm1, %xmm0, %xmm0 | |
| ; AVX2-NEXT: vzeroupper | |
| ; AVX2-NEXT: retq | |
| ; | |
| ; AVX512-LABEL: test_bitextract_v16i4_var: | |
| ; AVX512: # %bb.0: | |
| ; AVX512-NEXT: vpmovdw %zmm1, %ymm1 | |
| ; AVX512-NEXT: vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero | |
| ; AVX512-NEXT: vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm1, %ymm1 | |
| ; AVX512-NEXT: vpsrlvw %ymm1, %ymm0, %ymm0 | |
| ; AVX512-NEXT: vpmovwb %ymm0, %xmm0 | |
| ; AVX512-NEXT: vzeroupper | |
| ; AVX512-NEXT: retq | |
| %result = bitextract <16 x i4>, <16 x b8> %src, <16 x i32> %off | |
| ret <16 x i4> %result | |
| } |