diff options
Diffstat (limited to 'llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll | 352 |
1 files changed, 155 insertions, 197 deletions
diff --git a/llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll b/llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll index 01890e2adba..fd9787beb96 100644 --- a/llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll +++ b/llvm/test/CodeGen/X86/shuffle-vs-trunc-256.ll @@ -408,20 +408,17 @@ define void @shuffle_v32i8_to_v8i8(<32 x i8>* %L, <8 x i8>* %S) nounwind { ; AVX512BWVL-LABEL: shuffle_v32i8_to_v8i8: ; AVX512BWVL: # %bb.0: ; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512BWVL-NEXT: vmovdqa 16(%rdi), %xmm1 -; AVX512BWVL-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX512BWVL-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX512BWVL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX512BWVL-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX512BWVL-NEXT: vmovq %xmm0, (%rsi) +; AVX512BWVL-NEXT: vmovdqa {{.*#+}} xmm1 = [0,2,4,6,8,10,12,14] +; AVX512BWVL-NEXT: vpermi2w 16(%rdi), %xmm0, %xmm1 +; AVX512BWVL-NEXT: vpmovwb %xmm1, (%rsi) ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: shuffle_v32i8_to_v8i8: ; AVX512VBMIVL: # %bb.0: ; AVX512VBMIVL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512VBMIVL-NEXT: vpbroadcastq {{.*#+}} xmm1 = [2024390091656922112,2024390091656922112] -; AVX512VBMIVL-NEXT: vpermi2b 16(%rdi), %xmm0, %xmm1 -; AVX512VBMIVL-NEXT: vmovq %xmm1, (%rsi) +; AVX512VBMIVL-NEXT: vmovdqa {{.*#+}} xmm1 = [0,2,4,6,8,10,12,14] +; AVX512VBMIVL-NEXT: vpermi2w 16(%rdi), %xmm0, %xmm1 +; AVX512VBMIVL-NEXT: vpmovwb %xmm1, (%rsi) ; AVX512VBMIVL-NEXT: retq %vec = load <32 x i8>, <32 x i8>* %L %strided.vec = shufflevector <32 x i8> %vec, <32 x i8> undef, <8 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28> @@ -430,21 +427,32 @@ define void @shuffle_v32i8_to_v8i8(<32 x i8>* %L, <8 x i8>* %S) nounwind { } define void @trunc_v8i32_to_v8i8(<32 x i8>* %L, <8 x i8>* %S) nounwind { -; AVX-LABEL: trunc_v8i32_to_v8i8: -; AVX: # %bb.0: -; AVX-NEXT: vmovdqa (%rdi), %xmm0 -; AVX-NEXT: vmovdqa 16(%rdi), %xmm1 -; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX-NEXT: vmovq %xmm0, (%rsi) -; AVX-NEXT: retq +; AVX1-LABEL: trunc_v8i32_to_v8i8: +; AVX1: # %bb.0: +; AVX1-NEXT: vmovdqa (%rdi), %xmm0 +; AVX1-NEXT: vmovdqa 16(%rdi), %xmm1 +; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> +; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1 +; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 +; AVX1-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; AVX1-NEXT: vmovq %xmm0, (%rsi) +; AVX1-NEXT: retq +; +; AVX2-LABEL: trunc_v8i32_to_v8i8: +; AVX2: # %bb.0: +; AVX2-NEXT: vmovdqa (%rdi), %ymm0 +; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] +; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] +; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u] +; AVX2-NEXT: vmovq %xmm0, (%rsi) +; AVX2-NEXT: vzeroupper +; AVX2-NEXT: retq ; ; AVX512F-LABEL: trunc_v8i32_to_v8i8: ; AVX512F: # %bb.0: ; AVX512F-NEXT: vmovdqa (%rdi), %ymm0 -; AVX512F-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512F-NEXT: vpmovdw %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u] ; AVX512F-NEXT: vmovq %xmm0, (%rsi) ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq @@ -459,7 +467,8 @@ define void @trunc_v8i32_to_v8i8(<32 x i8>* %L, <8 x i8>* %S) nounwind { ; AVX512BW-LABEL: trunc_v8i32_to_v8i8: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0 -; AVX512BW-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512BW-NEXT: vpmovdw %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u] ; AVX512BW-NEXT: vmovq %xmm0, (%rsi) ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq @@ -490,59 +499,53 @@ define <2 x i64> @trunc_v8i32_to_v8i8_return_v2i64(<8 x i32> %vec) nounwind { ; AVX1-LABEL: trunc_v8i32_to_v8i8_return_v2i64: ; AVX1: # %bb.0: ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> +; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX1-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX1-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; ; AVX2-LABEL: trunc_v8i32_to_v8i8_return_v2i64: ; AVX2: # %bb.0: -; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX2-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX2-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX2-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX2-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] +; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] +; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; ; AVX512F-LABEL: trunc_v8i32_to_v8i8_return_v2i64: ; AVX512F: # %bb.0: ; AVX512F-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512F-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512F-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512F-NEXT: vpmovdw %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; ; AVX512VL-LABEL: trunc_v8i32_to_v8i8_return_v2i64: ; AVX512VL: # %bb.0: ; AVX512VL-NEXT: vpmovdb %ymm0, %xmm0 -; AVX512VL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VL-NEXT: vzeroupper ; AVX512VL-NEXT: retq ; ; AVX512BW-LABEL: trunc_v8i32_to_v8i8_return_v2i64: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512BW-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512BW-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512BW-NEXT: vpmovdw %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq ; ; AVX512BWVL-LABEL: trunc_v8i32_to_v8i8_return_v2i64: ; AVX512BWVL: # %bb.0: ; AVX512BWVL-NEXT: vpmovdb %ymm0, %xmm0 -; AVX512BWVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512BWVL-NEXT: vzeroupper ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: trunc_v8i32_to_v8i8_return_v2i64: ; AVX512VBMIVL: # %bb.0: ; AVX512VBMIVL-NEXT: vpmovdb %ymm0, %xmm0 -; AVX512VBMIVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VBMIVL-NEXT: vzeroupper ; AVX512VBMIVL-NEXT: retq %truncated.vec = trunc <8 x i32> %vec to <8 x i8> @@ -674,59 +677,53 @@ define <16 x i8> @trunc_v8i32_to_v8i8_return_v16i8(<8 x i32> %vec) nounwind { ; AVX1-LABEL: trunc_v8i32_to_v8i8_return_v16i8: ; AVX1: # %bb.0: ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> +; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] ; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1 ; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX1-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX1-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; ; AVX2-LABEL: trunc_v8i32_to_v8i8_return_v16i8: ; AVX2: # %bb.0: -; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX2-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX2-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX2-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX2-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31] +; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3] +; AVX2-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; ; AVX512F-LABEL: trunc_v8i32_to_v8i8_return_v16i8: ; AVX512F: # %bb.0: ; AVX512F-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512F-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512F-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512F-NEXT: vpmovdw %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; ; AVX512VL-LABEL: trunc_v8i32_to_v8i8_return_v16i8: ; AVX512VL: # %bb.0: ; AVX512VL-NEXT: vpmovdb %ymm0, %xmm0 -; AVX512VL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VL-NEXT: vzeroupper ; AVX512VL-NEXT: retq ; ; AVX512BW-LABEL: trunc_v8i32_to_v8i8_return_v16i8: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512BW-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512BW-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512BW-NEXT: vpmovdw %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq ; ; AVX512BWVL-LABEL: trunc_v8i32_to_v8i8_return_v16i8: ; AVX512BWVL: # %bb.0: ; AVX512BWVL-NEXT: vpmovdb %ymm0, %xmm0 -; AVX512BWVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512BWVL-NEXT: vzeroupper ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: trunc_v8i32_to_v8i8_return_v16i8: ; AVX512VBMIVL: # %bb.0: ; AVX512VBMIVL-NEXT: vpmovdb %ymm0, %xmm0 -; AVX512VBMIVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VBMIVL-NEXT: vzeroupper ; AVX512VBMIVL-NEXT: retq %truncated = trunc <8 x i32> %vec to <8 x i8> @@ -740,72 +737,58 @@ define <2 x i64> @trunc_v4i64_to_v4i16_return_v2i64(<4 x i64> %vec) nounwind { ; AVX1-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX1: # %bb.0: ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 -; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] -; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] -; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; AVX1-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX1-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] +; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; ; AVX2-SLOW-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX2-SLOW: # %bb.0: -; AVX2-SLOW-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] -; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] -; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX2-SLOW-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX2-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm1 +; AVX2-SLOW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] +; AVX2-SLOW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX2-SLOW-NEXT: vzeroupper ; AVX2-SLOW-NEXT: retq ; ; AVX2-FAST-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX2-FAST: # %bb.0: -; AVX2-FAST-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15] -; AVX2-FAST-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX2-FAST-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX2-FAST-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX2-FAST-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm1 = [0,2,4,6,4,6,6,7] +; AVX2-FAST-NEXT: vpermd %ymm0, %ymm1, %ymm0 +; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX2-FAST-NEXT: vzeroupper ; AVX2-FAST-NEXT: retq ; ; AVX512F-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX512F: # %bb.0: ; AVX512F-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512F-NEXT: vpmovqw %zmm0, %xmm0 -; AVX512F-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512F-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; ; AVX512VL-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX512VL: # %bb.0: ; AVX512VL-NEXT: vpmovqw %ymm0, %xmm0 -; AVX512VL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VL-NEXT: vzeroupper ; AVX512VL-NEXT: retq ; ; AVX512BW-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512BW-NEXT: vpmovqw %zmm0, %xmm0 -; AVX512BW-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq ; ; AVX512BWVL-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX512BWVL: # %bb.0: ; AVX512BWVL-NEXT: vpmovqw %ymm0, %xmm0 -; AVX512BWVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512BWVL-NEXT: vzeroupper ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: trunc_v4i64_to_v4i16_return_v2i64: ; AVX512VBMIVL: # %bb.0: ; AVX512VBMIVL-NEXT: vpmovqw %ymm0, %xmm0 -; AVX512VBMIVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VBMIVL-NEXT: vzeroupper ; AVX512VBMIVL-NEXT: retq %truncated = trunc <4 x i64> %vec to <4 x i16> @@ -947,72 +930,58 @@ define <8 x i16> @trunc_v4i64_to_v4i16_return_v8i16(<4 x i64> %vec) nounwind { ; AVX1-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX1: # %bb.0: ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 -; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] -; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] -; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; AVX1-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX1-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] +; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; ; AVX2-SLOW-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX2-SLOW: # %bb.0: -; AVX2-SLOW-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] -; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] -; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX2-SLOW-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX2-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm1 +; AVX2-SLOW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] +; AVX2-SLOW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX2-SLOW-NEXT: vzeroupper ; AVX2-SLOW-NEXT: retq ; ; AVX2-FAST-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX2-FAST: # %bb.0: -; AVX2-FAST-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15] -; AVX2-FAST-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX2-FAST-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX2-FAST-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX2-FAST-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm1 = [0,2,4,6,4,6,6,7] +; AVX2-FAST-NEXT: vpermd %ymm0, %ymm1, %ymm0 +; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX2-FAST-NEXT: vzeroupper ; AVX2-FAST-NEXT: retq ; ; AVX512F-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX512F: # %bb.0: ; AVX512F-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512F-NEXT: vpmovqw %zmm0, %xmm0 -; AVX512F-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512F-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; ; AVX512VL-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX512VL: # %bb.0: ; AVX512VL-NEXT: vpmovqw %ymm0, %xmm0 -; AVX512VL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VL-NEXT: vzeroupper ; AVX512VL-NEXT: retq ; ; AVX512BW-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512BW-NEXT: vpmovqw %zmm0, %xmm0 -; AVX512BW-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero +; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13],zero,zero,zero,zero,zero,zero,zero,zero ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq ; ; AVX512BWVL-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX512BWVL: # %bb.0: ; AVX512BWVL-NEXT: vpmovqw %ymm0, %xmm0 -; AVX512BWVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512BWVL-NEXT: vzeroupper ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: trunc_v4i64_to_v4i16_return_v8i16: ; AVX512VBMIVL: # %bb.0: ; AVX512VBMIVL-NEXT: vpmovqw %ymm0, %xmm0 -; AVX512VBMIVL-NEXT: vmovq {{.*#+}} xmm0 = xmm0[0],zero ; AVX512VBMIVL-NEXT: vzeroupper ; AVX512VBMIVL-NEXT: retq %truncated = trunc <4 x i64> %vec to <4 x i16> @@ -1024,66 +993,58 @@ define <16 x i8> @trunc_v4i64_to_v4i8_return_v16i8(<4 x i64> %vec) nounwind { ; AVX1-LABEL: trunc_v4i64_to_v4i8_return_v16i8: ; AVX1: # %bb.0: ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX1-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX1-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX1-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; AVX1-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; AVX1-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7] +; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] +; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[u],zero ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; -; AVX2-LABEL: trunc_v4i64_to_v4i8_return_v16i8: -; AVX2: # %bb.0: -; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX2-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX2-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX2-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; AVX2-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; AVX2-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7] -; AVX2-NEXT: vzeroupper -; AVX2-NEXT: retq +; AVX2-SLOW-LABEL: trunc_v4i64_to_v4i8_return_v16i8: +; AVX2-SLOW: # %bb.0: +; AVX2-SLOW-NEXT: vextractf128 $1, %ymm0, %xmm1 +; AVX2-SLOW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] +; AVX2-SLOW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[u],zero +; AVX2-SLOW-NEXT: vzeroupper +; AVX2-SLOW-NEXT: retq +; +; AVX2-FAST-LABEL: trunc_v4i64_to_v4i8_return_v16i8: +; AVX2-FAST: # %bb.0: +; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm1 = [0,2,4,6,4,6,6,7] +; AVX2-FAST-NEXT: vpermd %ymm0, %ymm1, %ymm0 +; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[u],zero +; AVX2-FAST-NEXT: vzeroupper +; AVX2-FAST-NEXT: retq ; ; AVX512F-LABEL: trunc_v4i64_to_v4i8_return_v16i8: ; AVX512F: # %bb.0: ; AVX512F-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512F-NEXT: vpmovqb %zmm0, %xmm0 -; AVX512F-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; AVX512F-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7] +; AVX512F-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[u],zero ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; ; AVX512VL-LABEL: trunc_v4i64_to_v4i8_return_v16i8: ; AVX512VL: # %bb.0: ; AVX512VL-NEXT: vpmovqb %ymm0, %xmm0 -; AVX512VL-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; AVX512VL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7] ; AVX512VL-NEXT: vzeroupper ; AVX512VL-NEXT: retq ; ; AVX512BW-LABEL: trunc_v4i64_to_v4i8_return_v16i8: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 -; AVX512BW-NEXT: vpmovqb %zmm0, %xmm0 -; AVX512BW-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; AVX512BW-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7] +; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm0[u],zero ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq ; ; AVX512BWVL-LABEL: trunc_v4i64_to_v4i8_return_v16i8: ; AVX512BWVL: # %bb.0: ; AVX512BWVL-NEXT: vpmovqb %ymm0, %xmm0 -; AVX512BWVL-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; AVX512BWVL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7] ; AVX512BWVL-NEXT: vzeroupper ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: trunc_v4i64_to_v4i8_return_v16i8: ; AVX512VBMIVL: # %bb.0: ; AVX512VBMIVL-NEXT: vpmovqb %ymm0, %xmm0 -; AVX512VBMIVL-NEXT: vpxor %xmm1, %xmm1, %xmm1 -; AVX512VBMIVL-NEXT: vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7] ; AVX512VBMIVL-NEXT: vzeroupper ; AVX512VBMIVL-NEXT: retq %truncated = trunc <4 x i64> %vec to <4 x i8> @@ -1135,13 +1096,9 @@ define void @shuffle_v16i16_to_v4i16(<16 x i16>* %L, <4 x i16>* %S) nounwind { ; ; AVX512VL-LABEL: shuffle_v16i16_to_v4i16: ; AVX512VL: # %bb.0: -; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512VL-NEXT: vmovdqa 16(%rdi), %xmm1 -; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15] -; AVX512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX512VL-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; AVX512VL-NEXT: vmovq %xmm0, (%rsi) +; AVX512VL-NEXT: vmovaps (%rdi), %xmm0 +; AVX512VL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX512VL-NEXT: vpmovdw %xmm0, (%rsi) ; AVX512VL-NEXT: retq ; ; AVX512BW-LABEL: shuffle_v16i16_to_v4i16: @@ -1157,18 +1114,16 @@ define void @shuffle_v16i16_to_v4i16(<16 x i16>* %L, <4 x i16>* %S) nounwind { ; ; AVX512BWVL-LABEL: shuffle_v16i16_to_v4i16: ; AVX512BWVL: # %bb.0: -; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512BWVL-NEXT: vmovdqa {{.*#+}} xmm1 = [0,4,8,12,4,5,12,13] -; AVX512BWVL-NEXT: vpermi2w 16(%rdi), %xmm0, %xmm1 -; AVX512BWVL-NEXT: vmovq %xmm1, (%rsi) +; AVX512BWVL-NEXT: vmovaps (%rdi), %xmm0 +; AVX512BWVL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX512BWVL-NEXT: vpmovdw %xmm0, (%rsi) ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: shuffle_v16i16_to_v4i16: ; AVX512VBMIVL: # %bb.0: -; AVX512VBMIVL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512VBMIVL-NEXT: vmovdqa {{.*#+}} xmm1 = [0,4,8,12,4,5,12,13] -; AVX512VBMIVL-NEXT: vpermi2w 16(%rdi), %xmm0, %xmm1 -; AVX512VBMIVL-NEXT: vmovq %xmm1, (%rsi) +; AVX512VBMIVL-NEXT: vmovaps (%rdi), %xmm0 +; AVX512VBMIVL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX512VBMIVL-NEXT: vpmovdw %xmm0, (%rsi) ; AVX512VBMIVL-NEXT: retq %vec = load <16 x i16>, <16 x i16>* %L %strided.vec = shufflevector <16 x i16> %vec, <16 x i16> undef, <4 x i32> <i32 0, i32 4, i32 8, i32 12> @@ -1179,39 +1134,34 @@ define void @shuffle_v16i16_to_v4i16(<16 x i16>* %L, <4 x i16>* %S) nounwind { define void @trunc_v4i64_to_v4i16(<16 x i16>* %L, <4 x i16>* %S) nounwind { ; AVX1-LABEL: trunc_v4i64_to_v4i16: ; AVX1: # %bb.0: -; AVX1-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3] -; AVX1-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; AVX1-NEXT: vpshufd {{.*#+}} xmm1 = mem[0,2,2,3] -; AVX1-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; AVX1-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] +; AVX1-NEXT: vmovaps (%rdi), %xmm0 +; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] ; AVX1-NEXT: vmovq %xmm0, (%rsi) ; AVX1-NEXT: retq ; ; AVX2-SLOW-LABEL: trunc_v4i64_to_v4i16: ; AVX2-SLOW: # %bb.0: -; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm0 = mem[0,2,2,3] -; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] -; AVX2-SLOW-NEXT: vpshufd {{.*#+}} xmm1 = mem[0,2,2,3] -; AVX2-SLOW-NEXT: vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] -; AVX2-SLOW-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1] +; AVX2-SLOW-NEXT: vmovaps (%rdi), %xmm0 +; AVX2-SLOW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX2-SLOW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] ; AVX2-SLOW-NEXT: vmovq %xmm0, (%rsi) ; AVX2-SLOW-NEXT: retq ; ; AVX2-FAST-LABEL: trunc_v4i64_to_v4i16: ; AVX2-FAST: # %bb.0: -; AVX2-FAST-NEXT: vmovdqa (%rdi), %xmm0 -; AVX2-FAST-NEXT: vmovdqa 16(%rdi), %xmm1 -; AVX2-FAST-NEXT: vmovdqa {{.*#+}} xmm2 = [0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15] -; AVX2-FAST-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX2-FAST-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX2-FAST-NEXT: vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm0 = [0,2,4,6,4,6,6,7] +; AVX2-FAST-NEXT: vpermd (%rdi), %ymm0, %ymm0 +; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] ; AVX2-FAST-NEXT: vmovq %xmm0, (%rsi) +; AVX2-FAST-NEXT: vzeroupper ; AVX2-FAST-NEXT: retq ; ; AVX512F-LABEL: trunc_v4i64_to_v4i16: ; AVX512F: # %bb.0: ; AVX512F-NEXT: vmovdqa (%rdi), %ymm0 -; AVX512F-NEXT: vpmovqw %zmm0, %xmm0 +; AVX512F-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] ; AVX512F-NEXT: vmovq %xmm0, (%rsi) ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq @@ -1226,7 +1176,8 @@ define void @trunc_v4i64_to_v4i16(<16 x i16>* %L, <4 x i16>* %S) nounwind { ; AVX512BW-LABEL: trunc_v4i64_to_v4i16: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0 -; AVX512BW-NEXT: vpmovqw %zmm0, %xmm0 +; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] ; AVX512BW-NEXT: vmovq %xmm0, (%rsi) ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq @@ -1276,13 +1227,9 @@ define void @shuffle_v32i8_to_v4i8(<32 x i8>* %L, <4 x i8>* %S) nounwind { ; ; AVX512VL-LABEL: shuffle_v32i8_to_v4i8: ; AVX512VL: # %bb.0: -; AVX512VL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512VL-NEXT: vmovdqa 16(%rdi), %xmm1 -; AVX512VL-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX512VL-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX512VL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX512VL-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; AVX512VL-NEXT: vmovd %xmm0, (%rsi) +; AVX512VL-NEXT: vmovaps (%rdi), %xmm0 +; AVX512VL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX512VL-NEXT: vpmovdb %xmm0, (%rsi) ; AVX512VL-NEXT: retq ; ; AVX512BW-LABEL: shuffle_v32i8_to_v4i8: @@ -1298,21 +1245,16 @@ define void @shuffle_v32i8_to_v4i8(<32 x i8>* %L, <4 x i8>* %S) nounwind { ; ; AVX512BWVL-LABEL: shuffle_v32i8_to_v4i8: ; AVX512BWVL: # %bb.0: -; AVX512BWVL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512BWVL-NEXT: vmovdqa 16(%rdi), %xmm1 -; AVX512BWVL-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX512BWVL-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX512BWVL-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX512BWVL-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; AVX512BWVL-NEXT: vmovd %xmm0, (%rsi) +; AVX512BWVL-NEXT: vmovaps (%rdi), %xmm0 +; AVX512BWVL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX512BWVL-NEXT: vpmovdb %xmm0, (%rsi) ; AVX512BWVL-NEXT: retq ; ; AVX512VBMIVL-LABEL: shuffle_v32i8_to_v4i8: ; AVX512VBMIVL: # %bb.0: -; AVX512VBMIVL-NEXT: vmovdqa (%rdi), %xmm0 -; AVX512VBMIVL-NEXT: vpbroadcastd {{.*#+}} xmm1 = [403703808,403703808,403703808,403703808] -; AVX512VBMIVL-NEXT: vpermi2b 16(%rdi), %xmm0, %xmm1 -; AVX512VBMIVL-NEXT: vmovd %xmm1, (%rsi) +; AVX512VBMIVL-NEXT: vmovaps (%rdi), %xmm0 +; AVX512VBMIVL-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX512VBMIVL-NEXT: vpmovdb %xmm0, (%rsi) ; AVX512VBMIVL-NEXT: retq %vec = load <32 x i8>, <32 x i8>* %L %strided.vec = shufflevector <32 x i8> %vec, <32 x i8> undef, <4 x i32> <i32 0, i32 8, i32 16, i32 24> @@ -1321,21 +1263,36 @@ define void @shuffle_v32i8_to_v4i8(<32 x i8>* %L, <4 x i8>* %S) nounwind { } define void @trunc_v4i64_to_v4i8(<32 x i8>* %L, <4 x i8>* %S) nounwind { -; AVX-LABEL: trunc_v4i64_to_v4i8: -; AVX: # %bb.0: -; AVX-NEXT: vmovdqa (%rdi), %xmm0 -; AVX-NEXT: vmovdqa 16(%rdi), %xmm1 -; AVX-NEXT: vmovdqa {{.*#+}} xmm2 = <0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u> -; AVX-NEXT: vpshufb %xmm2, %xmm1, %xmm1 -; AVX-NEXT: vpshufb %xmm2, %xmm0, %xmm0 -; AVX-NEXT: vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; AVX-NEXT: vmovd %xmm0, (%rsi) -; AVX-NEXT: retq +; AVX1-LABEL: trunc_v4i64_to_v4i8: +; AVX1: # %bb.0: +; AVX1-NEXT: vmovaps (%rdi), %xmm0 +; AVX1-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX1-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u] +; AVX1-NEXT: vmovd %xmm0, (%rsi) +; AVX1-NEXT: retq +; +; AVX2-SLOW-LABEL: trunc_v4i64_to_v4i8: +; AVX2-SLOW: # %bb.0: +; AVX2-SLOW-NEXT: vmovaps (%rdi), %xmm0 +; AVX2-SLOW-NEXT: vshufps {{.*#+}} xmm0 = xmm0[0,2],mem[0,2] +; AVX2-SLOW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u] +; AVX2-SLOW-NEXT: vmovd %xmm0, (%rsi) +; AVX2-SLOW-NEXT: retq +; +; AVX2-FAST-LABEL: trunc_v4i64_to_v4i8: +; AVX2-FAST: # %bb.0: +; AVX2-FAST-NEXT: vmovdqa {{.*#+}} ymm0 = [0,2,4,6,4,6,6,7] +; AVX2-FAST-NEXT: vpermd (%rdi), %ymm0, %ymm0 +; AVX2-FAST-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u] +; AVX2-FAST-NEXT: vmovd %xmm0, (%rsi) +; AVX2-FAST-NEXT: vzeroupper +; AVX2-FAST-NEXT: retq ; ; AVX512F-LABEL: trunc_v4i64_to_v4i8: ; AVX512F: # %bb.0: ; AVX512F-NEXT: vmovdqa (%rdi), %ymm0 -; AVX512F-NEXT: vpmovqb %zmm0, %xmm0 +; AVX512F-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512F-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u] ; AVX512F-NEXT: vmovd %xmm0, (%rsi) ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq @@ -1350,7 +1307,8 @@ define void @trunc_v4i64_to_v4i8(<32 x i8>* %L, <4 x i8>* %S) nounwind { ; AVX512BW-LABEL: trunc_v4i64_to_v4i8: ; AVX512BW: # %bb.0: ; AVX512BW-NEXT: vmovdqa (%rdi), %ymm0 -; AVX512BW-NEXT: vpmovqb %zmm0, %xmm0 +; AVX512BW-NEXT: vpmovqd %zmm0, %ymm0 +; AVX512BW-NEXT: vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u] ; AVX512BW-NEXT: vmovd %xmm0, (%rsi) ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq |
