diff options
| author | Dimitry Andric <dim@FreeBSD.org> | 2018-07-28 10:51:19 +0000 |
|---|---|---|
| committer | Dimitry Andric <dim@FreeBSD.org> | 2018-07-28 10:51:19 +0000 |
| commit | eb11fae6d08f479c0799db45860a98af528fa6e7 (patch) | |
| tree | 44d492a50c8c1a7eb8e2d17ea3360ec4d066f042 /test/CodeGen/X86/bitcast-and-setcc-512.ll | |
| parent | b8a2042aa938069e862750553db0e4d82d25822c (diff) | |
| download | src-eb11fae6d08f479c0799db45860a98af528fa6e7.tar.gz src-eb11fae6d08f479c0799db45860a98af528fa6e7.zip | |
Vendor import of llvm trunk r338150:vendor/llvm/llvm-trunk-r338150
Notes
Notes:
svn path=/vendor/llvm/dist/; revision=336809
svn path=/vendor/llvm/llvm-trunk-r338150/; revision=336814; tag=vendor/llvm/llvm-trunk-r338150
Diffstat (limited to 'test/CodeGen/X86/bitcast-and-setcc-512.ll')
| -rw-r--r-- | test/CodeGen/X86/bitcast-and-setcc-512.ll | 976 |
1 files changed, 203 insertions, 773 deletions
diff --git a/test/CodeGen/X86/bitcast-and-setcc-512.ll b/test/CodeGen/X86/bitcast-and-setcc-512.ll index dfda374aa52f..183e32a518b8 100644 --- a/test/CodeGen/X86/bitcast-and-setcc-512.ll +++ b/test/CodeGen/X86/bitcast-and-setcc-512.ll @@ -13,23 +13,39 @@ define i8 @v8i64(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) { ; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm10 ; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm11 ; SSE-NEXT: pcmpgtq %xmm7, %xmm3 +; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,2,4,5,6,7] ; SSE-NEXT: pcmpgtq %xmm6, %xmm2 -; SSE-NEXT: shufps {{.*#+}} xmm2 = xmm2[0,2],xmm3[0,2] +; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1] ; SSE-NEXT: pcmpgtq %xmm5, %xmm1 +; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] ; SSE-NEXT: pcmpgtq %xmm4, %xmm0 -; SSE-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2] -; SSE-NEXT: packssdw %xmm2, %xmm0 +; SSE-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7] ; SSE-NEXT: pcmpgtq {{[0-9]+}}(%rsp), %xmm11 +; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm11[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,1,0,2,4,5,6,7] ; SSE-NEXT: pcmpgtq {{[0-9]+}}(%rsp), %xmm10 -; SSE-NEXT: shufps {{.*#+}} xmm10 = xmm10[0,2],xmm11[0,2] +; SSE-NEXT: pshufd {{.*#+}} xmm2 = xmm10[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE-NEXT: pcmpgtq {{[0-9]+}}(%rsp), %xmm9 +; SSE-NEXT: pshufd {{.*#+}} xmm1 = xmm9[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7] ; SSE-NEXT: pcmpgtq {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: shufps {{.*#+}} xmm8 = xmm8[0,2],xmm9[0,2] -; SSE-NEXT: packssdw %xmm10, %xmm8 -; SSE-NEXT: pand %xmm0, %xmm8 -; SSE-NEXT: packsswb %xmm0, %xmm8 -; SSE-NEXT: pmovmskb %xmm8, %eax -; SSE-NEXT: # kill: def %al killed %al killed %eax +; SSE-NEXT: pshufd {{.*#+}} xmm3 = xmm8[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm3 = xmm3[0,2,2,3,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1] +; SSE-NEXT: pblendw {{.*#+}} xmm3 = xmm3[0,1,2,3],xmm2[4,5,6,7] +; SSE-NEXT: pand %xmm0, %xmm3 +; SSE-NEXT: packsswb %xmm0, %xmm3 +; SSE-NEXT: pmovmskb %xmm3, %eax +; SSE-NEXT: # kill: def $al killed $al killed $eax ; SSE-NEXT: retq ; ; AVX1-LABEL: v8i64: @@ -38,33 +54,30 @@ define i8 @v8i64(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) { ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm9 ; AVX1-NEXT: vpcmpgtq %xmm8, %xmm9, %xmm8 ; AVX1-NEXT: vpcmpgtq %xmm3, %xmm1, %xmm1 -; AVX1-NEXT: vpackssdw %xmm8, %xmm1, %xmm1 -; AVX1-NEXT: vmovdqa {{.*#+}} xmm8 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] -; AVX1-NEXT: vpshufb %xmm8, %xmm1, %xmm9 +; AVX1-NEXT: vpackssdw %xmm8, %xmm1, %xmm8 ; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm3 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vpcmpgtq %xmm3, %xmm1, %xmm1 ; AVX1-NEXT: vpcmpgtq %xmm2, %xmm0, %xmm0 ; AVX1-NEXT: vpackssdw %xmm1, %xmm0, %xmm0 -; AVX1-NEXT: vpshufb %xmm8, %xmm0, %xmm0 -; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm9[0] +; AVX1-NEXT: vpackssdw %xmm8, %xmm0, %xmm0 ; AVX1-NEXT: vextractf128 $1, %ymm7, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm5, %xmm2 ; AVX1-NEXT: vpcmpgtq %xmm1, %xmm2, %xmm1 ; AVX1-NEXT: vpcmpgtq %xmm7, %xmm5, %xmm2 ; AVX1-NEXT: vpackssdw %xmm1, %xmm2, %xmm1 -; AVX1-NEXT: vpshufb %xmm8, %xmm1, %xmm1 ; AVX1-NEXT: vextractf128 $1, %ymm6, %xmm2 ; AVX1-NEXT: vextractf128 $1, %ymm4, %xmm3 ; AVX1-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2 ; AVX1-NEXT: vpcmpgtq %xmm6, %xmm4, %xmm3 ; AVX1-NEXT: vpackssdw %xmm2, %xmm3, %xmm2 -; AVX1-NEXT: vpshufb %xmm8, %xmm2, %xmm2 -; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0] +; AVX1-NEXT: vpackssdw %xmm1, %xmm2, %xmm1 ; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0 +; AVX1-NEXT: vpsllw $15, %xmm0, %xmm0 +; AVX1-NEXT: vpsraw $15, %xmm0, %xmm0 ; AVX1-NEXT: vpacksswb %xmm0, %xmm0, %xmm0 ; AVX1-NEXT: vpmovmskb %xmm0, %eax -; AVX1-NEXT: # kill: def %al killed %al killed %eax +; AVX1-NEXT: # kill: def $al killed $al killed $eax ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; @@ -73,26 +86,23 @@ define i8 @v8i64(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) { ; AVX2-NEXT: vpcmpgtq %ymm3, %ymm1, %ymm1 ; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm3 ; AVX2-NEXT: vpackssdw %xmm3, %xmm1, %xmm1 -; AVX2-NEXT: vmovdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] -; AVX2-NEXT: vpshufb %xmm3, %xmm1, %xmm1 ; AVX2-NEXT: vpcmpgtq %ymm2, %ymm0, %ymm0 ; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm2 ; AVX2-NEXT: vpackssdw %xmm2, %xmm0, %xmm0 -; AVX2-NEXT: vpshufb %xmm3, %xmm0, %xmm0 -; AVX2-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; AVX2-NEXT: vpackssdw %xmm1, %xmm0, %xmm0 ; AVX2-NEXT: vpcmpgtq %ymm7, %ymm5, %ymm1 ; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 ; AVX2-NEXT: vpackssdw %xmm2, %xmm1, %xmm1 -; AVX2-NEXT: vpshufb %xmm3, %xmm1, %xmm1 ; AVX2-NEXT: vpcmpgtq %ymm6, %ymm4, %ymm2 -; AVX2-NEXT: vextracti128 $1, %ymm2, %xmm4 -; AVX2-NEXT: vpackssdw %xmm4, %xmm2, %xmm2 -; AVX2-NEXT: vpshufb %xmm3, %xmm2, %xmm2 -; AVX2-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0] +; AVX2-NEXT: vextracti128 $1, %ymm2, %xmm3 +; AVX2-NEXT: vpackssdw %xmm3, %xmm2, %xmm2 +; AVX2-NEXT: vpackssdw %xmm1, %xmm2, %xmm1 ; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0 +; AVX2-NEXT: vpsllw $15, %xmm0, %xmm0 +; AVX2-NEXT: vpsraw $15, %xmm0, %xmm0 ; AVX2-NEXT: vpacksswb %xmm0, %xmm0, %xmm0 ; AVX2-NEXT: vpmovmskb %xmm0, %eax -; AVX2-NEXT: # kill: def %al killed %al killed %eax +; AVX2-NEXT: # kill: def $al killed $al killed $eax ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; @@ -101,7 +111,7 @@ define i8 @v8i64(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) { ; AVX512F-NEXT: vpcmpgtq %zmm1, %zmm0, %k1 ; AVX512F-NEXT: vpcmpgtq %zmm3, %zmm2, %k0 {%k1} ; AVX512F-NEXT: kmovw %k0, %eax -; AVX512F-NEXT: # kill: def %al killed %al killed %eax +; AVX512F-NEXT: # kill: def $al killed $al killed $eax ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; @@ -110,7 +120,7 @@ define i8 @v8i64(<8 x i64> %a, <8 x i64> %b, <8 x i64> %c, <8 x i64> %d) { ; AVX512BW-NEXT: vpcmpgtq %zmm1, %zmm0, %k1 ; AVX512BW-NEXT: vpcmpgtq %zmm3, %zmm2, %k0 {%k1} ; AVX512BW-NEXT: kmovd %k0, %eax -; AVX512BW-NEXT: # kill: def %al killed %al killed %eax +; AVX512BW-NEXT: # kill: def $al killed $al killed $eax ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq %x0 = icmp sgt <8 x i64> %a, %b @@ -128,23 +138,39 @@ define i8 @v8f64(<8 x double> %a, <8 x double> %b, <8 x double> %c, <8 x double> ; SSE-NEXT: movapd {{[0-9]+}}(%rsp), %xmm10 ; SSE-NEXT: movapd {{[0-9]+}}(%rsp), %xmm11 ; SSE-NEXT: cmpltpd %xmm3, %xmm7 +; SSE-NEXT: shufps {{.*#+}} xmm7 = xmm7[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm3 = xmm7[0,1,0,2,4,5,6,7] ; SSE-NEXT: cmpltpd %xmm2, %xmm6 -; SSE-NEXT: shufps {{.*#+}} xmm6 = xmm6[0,2],xmm7[0,2] +; SSE-NEXT: shufps {{.*#+}} xmm6 = xmm6[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm2 = xmm6[0,1,0,2,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1] ; SSE-NEXT: cmpltpd %xmm1, %xmm5 +; SSE-NEXT: shufps {{.*#+}} xmm5 = xmm5[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm5[0,2,2,3,4,5,6,7] ; SSE-NEXT: cmpltpd %xmm0, %xmm4 -; SSE-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,2],xmm5[0,2] -; SSE-NEXT: packssdw %xmm6, %xmm4 +; SSE-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm0 = xmm4[0,2,2,3,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7] ; SSE-NEXT: cmpltpd {{[0-9]+}}(%rsp), %xmm11 +; SSE-NEXT: shufps {{.*#+}} xmm11 = xmm11[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm11[0,1,0,2,4,5,6,7] ; SSE-NEXT: cmpltpd {{[0-9]+}}(%rsp), %xmm10 -; SSE-NEXT: shufps {{.*#+}} xmm10 = xmm10[0,2],xmm11[0,2] +; SSE-NEXT: shufps {{.*#+}} xmm10 = xmm10[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm2 = xmm10[0,1,0,2,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1] ; SSE-NEXT: cmpltpd {{[0-9]+}}(%rsp), %xmm9 +; SSE-NEXT: shufps {{.*#+}} xmm9 = xmm9[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm1 = xmm9[0,2,2,3,4,5,6,7] ; SSE-NEXT: cmpltpd {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: shufps {{.*#+}} xmm8 = xmm8[0,2],xmm9[0,2] -; SSE-NEXT: packssdw %xmm10, %xmm8 -; SSE-NEXT: pand %xmm4, %xmm8 -; SSE-NEXT: packsswb %xmm0, %xmm8 -; SSE-NEXT: pmovmskb %xmm8, %eax -; SSE-NEXT: # kill: def %al killed %al killed %eax +; SSE-NEXT: shufps {{.*#+}} xmm8 = xmm8[0,2,2,3] +; SSE-NEXT: pshuflw {{.*#+}} xmm3 = xmm8[0,2,2,3,4,5,6,7] +; SSE-NEXT: punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1] +; SSE-NEXT: pblendw {{.*#+}} xmm3 = xmm3[0,1,2,3],xmm2[4,5,6,7] +; SSE-NEXT: pand %xmm0, %xmm3 +; SSE-NEXT: packsswb %xmm0, %xmm3 +; SSE-NEXT: pmovmskb %xmm3, %eax +; SSE-NEXT: # kill: def $al killed $al killed $eax ; SSE-NEXT: retq ; ; AVX12-LABEL: v8f64: @@ -152,26 +178,23 @@ define i8 @v8f64(<8 x double> %a, <8 x double> %b, <8 x double> %c, <8 x double> ; AVX12-NEXT: vcmpltpd %ymm1, %ymm3, %ymm1 ; AVX12-NEXT: vextractf128 $1, %ymm1, %xmm3 ; AVX12-NEXT: vpackssdw %xmm3, %xmm1, %xmm1 -; AVX12-NEXT: vmovdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] -; AVX12-NEXT: vpshufb %xmm3, %xmm1, %xmm1 ; AVX12-NEXT: vcmpltpd %ymm0, %ymm2, %ymm0 ; AVX12-NEXT: vextractf128 $1, %ymm0, %xmm2 ; AVX12-NEXT: vpackssdw %xmm2, %xmm0, %xmm0 -; AVX12-NEXT: vpshufb %xmm3, %xmm0, %xmm0 -; AVX12-NEXT: vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; AVX12-NEXT: vpackssdw %xmm1, %xmm0, %xmm0 ; AVX12-NEXT: vcmpltpd %ymm5, %ymm7, %ymm1 ; AVX12-NEXT: vextractf128 $1, %ymm1, %xmm2 ; AVX12-NEXT: vpackssdw %xmm2, %xmm1, %xmm1 -; AVX12-NEXT: vpshufb %xmm3, %xmm1, %xmm1 ; AVX12-NEXT: vcmpltpd %ymm4, %ymm6, %ymm2 -; AVX12-NEXT: vextractf128 $1, %ymm2, %xmm4 -; AVX12-NEXT: vpackssdw %xmm4, %xmm2, %xmm2 -; AVX12-NEXT: vpshufb %xmm3, %xmm2, %xmm2 -; AVX12-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0] +; AVX12-NEXT: vextractf128 $1, %ymm2, %xmm3 +; AVX12-NEXT: vpackssdw %xmm3, %xmm2, %xmm2 +; AVX12-NEXT: vpackssdw %xmm1, %xmm2, %xmm1 ; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0 +; AVX12-NEXT: vpsllw $15, %xmm0, %xmm0 +; AVX12-NEXT: vpsraw $15, %xmm0, %xmm0 ; AVX12-NEXT: vpacksswb %xmm0, %xmm0, %xmm0 ; AVX12-NEXT: vpmovmskb %xmm0, %eax -; AVX12-NEXT: # kill: def %al killed %al killed %eax +; AVX12-NEXT: # kill: def $al killed $al killed $eax ; AVX12-NEXT: vzeroupper ; AVX12-NEXT: retq ; @@ -180,7 +203,7 @@ define i8 @v8f64(<8 x double> %a, <8 x double> %b, <8 x double> %c, <8 x double> ; AVX512F-NEXT: vcmpltpd %zmm0, %zmm1, %k1 ; AVX512F-NEXT: vcmpltpd %zmm2, %zmm3, %k0 {%k1} ; AVX512F-NEXT: kmovw %k0, %eax -; AVX512F-NEXT: # kill: def %al killed %al killed %eax +; AVX512F-NEXT: # kill: def $al killed $al killed $eax ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; @@ -189,7 +212,7 @@ define i8 @v8f64(<8 x double> %a, <8 x double> %b, <8 x double> %c, <8 x double> ; AVX512BW-NEXT: vcmpltpd %zmm0, %zmm1, %k1 ; AVX512BW-NEXT: vcmpltpd %zmm2, %zmm3, %k0 {%k1} ; AVX512BW-NEXT: kmovd %k0, %eax -; AVX512BW-NEXT: # kill: def %al killed %al killed %eax +; AVX512BW-NEXT: # kill: def $al killed $al killed $eax ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq %x0 = fcmp ogt <8 x double> %a, %b @@ -280,38 +303,22 @@ define i32 @v32i16(<32 x i16> %a, <32 x i16> %b, <32 x i16> %c, <32 x i16> %d) { ; ; AVX512F-LABEL: v32i16: ; AVX512F: # %bb.0: -; AVX512F-NEXT: pushq %rbp -; AVX512F-NEXT: .cfi_def_cfa_offset 16 -; AVX512F-NEXT: .cfi_offset %rbp, -16 -; AVX512F-NEXT: movq %rsp, %rbp -; AVX512F-NEXT: .cfi_def_cfa_register %rbp -; AVX512F-NEXT: andq $-32, %rsp -; AVX512F-NEXT: subq $32, %rsp +; AVX512F-NEXT: vpcmpgtw %ymm3, %ymm1, %ymm1 +; AVX512F-NEXT: vpmovsxwd %ymm1, %zmm1 +; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k1 ; AVX512F-NEXT: vpcmpgtw %ymm2, %ymm0, %ymm0 ; AVX512F-NEXT: vpmovsxwd %ymm0, %zmm0 -; AVX512F-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512F-NEXT: vpcmpgtw %ymm3, %ymm1, %ymm1 +; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k2 +; AVX512F-NEXT: vpcmpgtw %ymm7, %ymm5, %ymm0 +; AVX512F-NEXT: vpmovsxwd %ymm0, %zmm0 +; AVX512F-NEXT: vpcmpgtw %ymm6, %ymm4, %ymm1 ; AVX512F-NEXT: vpmovsxwd %ymm1, %zmm1 -; AVX512F-NEXT: vpmovdb %zmm1, %xmm1 -; AVX512F-NEXT: vpcmpgtw %ymm6, %ymm4, %ymm2 -; AVX512F-NEXT: vpmovsxwd %ymm2, %zmm2 -; AVX512F-NEXT: vpmovdb %zmm2, %xmm2 -; AVX512F-NEXT: vpand %xmm2, %xmm0, %xmm0 -; AVX512F-NEXT: vpcmpgtw %ymm7, %ymm5, %ymm2 -; AVX512F-NEXT: vpmovsxwd %ymm2, %zmm2 -; AVX512F-NEXT: vpmovdb %zmm2, %xmm2 -; AVX512F-NEXT: vpand %xmm2, %xmm1, %xmm1 -; AVX512F-NEXT: vpmovsxbd %xmm1, %zmm1 -; AVX512F-NEXT: vpslld $31, %zmm1, %zmm1 -; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k0 -; AVX512F-NEXT: kmovw %k0, {{[0-9]+}}(%rsp) -; AVX512F-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512F-NEXT: vpslld $31, %zmm0, %zmm0 -; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k0 -; AVX512F-NEXT: kmovw %k0, (%rsp) -; AVX512F-NEXT: movl (%rsp), %eax -; AVX512F-NEXT: movq %rbp, %rsp -; AVX512F-NEXT: popq %rbp +; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k0 {%k2} +; AVX512F-NEXT: kmovw %k0, %ecx +; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k0 {%k1} +; AVX512F-NEXT: kmovw %k0, %eax +; AVX512F-NEXT: shll $16, %eax +; AVX512F-NEXT: orl %ecx, %eax ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; @@ -333,26 +340,36 @@ define i16 @v16i32(<16 x i32> %a, <16 x i32> %b, <16 x i32> %c, <16 x i32> %d) { ; SSE-LABEL: v16i32: ; SSE: # %bb.0: ; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm9 ; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm10 +; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm9 ; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm11 ; SSE-NEXT: pcmpgtd %xmm7, %xmm3 +; SSE-NEXT: movdqa {{.*#+}} xmm7 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u> +; SSE-NEXT: pshufb %xmm7, %xmm3 ; SSE-NEXT: pcmpgtd %xmm6, %xmm2 -; SSE-NEXT: packssdw %xmm3, %xmm2 +; SSE-NEXT: pshufb %xmm7, %xmm2 +; SSE-NEXT: punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1] ; SSE-NEXT: pcmpgtd %xmm5, %xmm1 +; SSE-NEXT: movdqa {{.*#+}} xmm3 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> +; SSE-NEXT: pshufb %xmm3, %xmm1 ; SSE-NEXT: pcmpgtd %xmm4, %xmm0 -; SSE-NEXT: packssdw %xmm1, %xmm0 -; SSE-NEXT: packsswb %xmm2, %xmm0 +; SSE-NEXT: pshufb %xmm3, %xmm0 +; SSE-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; SSE-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7] ; SSE-NEXT: pcmpgtd {{[0-9]+}}(%rsp), %xmm11 -; SSE-NEXT: pcmpgtd {{[0-9]+}}(%rsp), %xmm10 -; SSE-NEXT: packssdw %xmm11, %xmm10 +; SSE-NEXT: pshufb %xmm7, %xmm11 ; SSE-NEXT: pcmpgtd {{[0-9]+}}(%rsp), %xmm9 +; SSE-NEXT: pshufb %xmm7, %xmm9 +; SSE-NEXT: punpckldq {{.*#+}} xmm9 = xmm9[0],xmm11[0],xmm9[1],xmm11[1] +; SSE-NEXT: pcmpgtd {{[0-9]+}}(%rsp), %xmm10 +; SSE-NEXT: pshufb %xmm3, %xmm10 ; SSE-NEXT: pcmpgtd {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: packssdw %xmm9, %xmm8 -; SSE-NEXT: packsswb %xmm10, %xmm8 +; SSE-NEXT: pshufb %xmm3, %xmm8 +; SSE-NEXT: punpckldq {{.*#+}} xmm8 = xmm8[0],xmm10[0],xmm8[1],xmm10[1] +; SSE-NEXT: pblendw {{.*#+}} xmm8 = xmm8[0,1,2,3],xmm9[4,5,6,7] ; SSE-NEXT: pand %xmm0, %xmm8 ; SSE-NEXT: pmovmskb %xmm8, %eax -; SSE-NEXT: # kill: def %ax killed %ax killed %eax +; SSE-NEXT: # kill: def $ax killed $ax killed $eax ; SSE-NEXT: retq ; ; AVX1-LABEL: v16i32: @@ -381,7 +398,7 @@ define i16 @v16i32(<16 x i32> %a, <16 x i32> %b, <16 x i32> %c, <16 x i32> %d) { ; AVX1-NEXT: vpacksswb %xmm1, %xmm2, %xmm1 ; AVX1-NEXT: vpand %xmm1, %xmm0, %xmm0 ; AVX1-NEXT: vpmovmskb %xmm0, %eax -; AVX1-NEXT: # kill: def %ax killed %ax killed %eax +; AVX1-NEXT: # kill: def $ax killed $ax killed $eax ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; @@ -403,7 +420,7 @@ define i16 @v16i32(<16 x i32> %a, <16 x i32> %b, <16 x i32> %c, <16 x i32> %d) { ; AVX2-NEXT: vpacksswb %xmm1, %xmm2, %xmm1 ; AVX2-NEXT: vpand %xmm1, %xmm0, %xmm0 ; AVX2-NEXT: vpmovmskb %xmm0, %eax -; AVX2-NEXT: # kill: def %ax killed %ax killed %eax +; AVX2-NEXT: # kill: def $ax killed $ax killed $eax ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; @@ -412,7 +429,7 @@ define i16 @v16i32(<16 x i32> %a, <16 x i32> %b, <16 x i32> %c, <16 x i32> %d) { ; AVX512F-NEXT: vpcmpgtd %zmm1, %zmm0, %k1 ; AVX512F-NEXT: vpcmpgtd %zmm3, %zmm2, %k0 {%k1} ; AVX512F-NEXT: kmovw %k0, %eax -; AVX512F-NEXT: # kill: def %ax killed %ax killed %eax +; AVX512F-NEXT: # kill: def $ax killed $ax killed $eax ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; @@ -421,7 +438,7 @@ define i16 @v16i32(<16 x i32> %a, <16 x i32> %b, <16 x i32> %c, <16 x i32> %d) { ; AVX512BW-NEXT: vpcmpgtd %zmm1, %zmm0, %k1 ; AVX512BW-NEXT: vpcmpgtd %zmm3, %zmm2, %k0 {%k1} ; AVX512BW-NEXT: kmovd %k0, %eax -; AVX512BW-NEXT: # kill: def %ax killed %ax killed %eax +; AVX512BW-NEXT: # kill: def $ax killed $ax killed $eax ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq %x0 = icmp sgt <16 x i32> %a, %b @@ -435,26 +452,36 @@ define i16 @v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %c, <16 x floa ; SSE-LABEL: v16f32: ; SSE: # %bb.0: ; SSE-NEXT: movaps {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: movaps {{[0-9]+}}(%rsp), %xmm9 ; SSE-NEXT: movaps {{[0-9]+}}(%rsp), %xmm10 +; SSE-NEXT: movaps {{[0-9]+}}(%rsp), %xmm9 ; SSE-NEXT: movaps {{[0-9]+}}(%rsp), %xmm11 ; SSE-NEXT: cmpltps %xmm3, %xmm7 +; SSE-NEXT: movdqa {{.*#+}} xmm3 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u> +; SSE-NEXT: pshufb %xmm3, %xmm7 ; SSE-NEXT: cmpltps %xmm2, %xmm6 -; SSE-NEXT: packssdw %xmm7, %xmm6 +; SSE-NEXT: pshufb %xmm3, %xmm6 +; SSE-NEXT: punpckldq {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1] ; SSE-NEXT: cmpltps %xmm1, %xmm5 +; SSE-NEXT: movdqa {{.*#+}} xmm1 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u> +; SSE-NEXT: pshufb %xmm1, %xmm5 ; SSE-NEXT: cmpltps %xmm0, %xmm4 -; SSE-NEXT: packssdw %xmm5, %xmm4 -; SSE-NEXT: packsswb %xmm6, %xmm4 +; SSE-NEXT: pshufb %xmm1, %xmm4 +; SSE-NEXT: punpckldq {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1] +; SSE-NEXT: pblendw {{.*#+}} xmm4 = xmm4[0,1,2,3],xmm6[4,5,6,7] ; SSE-NEXT: cmpltps {{[0-9]+}}(%rsp), %xmm11 -; SSE-NEXT: cmpltps {{[0-9]+}}(%rsp), %xmm10 -; SSE-NEXT: packssdw %xmm11, %xmm10 +; SSE-NEXT: pshufb %xmm3, %xmm11 ; SSE-NEXT: cmpltps {{[0-9]+}}(%rsp), %xmm9 +; SSE-NEXT: pshufb %xmm3, %xmm9 +; SSE-NEXT: punpckldq {{.*#+}} xmm9 = xmm9[0],xmm11[0],xmm9[1],xmm11[1] +; SSE-NEXT: cmpltps {{[0-9]+}}(%rsp), %xmm10 +; SSE-NEXT: pshufb %xmm1, %xmm10 ; SSE-NEXT: cmpltps {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: packssdw %xmm9, %xmm8 -; SSE-NEXT: packsswb %xmm10, %xmm8 +; SSE-NEXT: pshufb %xmm1, %xmm8 +; SSE-NEXT: punpckldq {{.*#+}} xmm8 = xmm8[0],xmm10[0],xmm8[1],xmm10[1] +; SSE-NEXT: pblendw {{.*#+}} xmm8 = xmm8[0,1,2,3],xmm9[4,5,6,7] ; SSE-NEXT: pand %xmm4, %xmm8 ; SSE-NEXT: pmovmskb %xmm8, %eax -; SSE-NEXT: # kill: def %ax killed %ax killed %eax +; SSE-NEXT: # kill: def $ax killed $ax killed $eax ; SSE-NEXT: retq ; ; AVX12-LABEL: v16f32: @@ -475,7 +502,7 @@ define i16 @v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %c, <16 x floa ; AVX12-NEXT: vpacksswb %xmm1, %xmm2, %xmm1 ; AVX12-NEXT: vpand %xmm1, %xmm0, %xmm0 ; AVX12-NEXT: vpmovmskb %xmm0, %eax -; AVX12-NEXT: # kill: def %ax killed %ax killed %eax +; AVX12-NEXT: # kill: def $ax killed $ax killed $eax ; AVX12-NEXT: vzeroupper ; AVX12-NEXT: retq ; @@ -484,7 +511,7 @@ define i16 @v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %c, <16 x floa ; AVX512F-NEXT: vcmpltps %zmm0, %zmm1, %k1 ; AVX512F-NEXT: vcmpltps %zmm2, %zmm3, %k0 {%k1} ; AVX512F-NEXT: kmovw %k0, %eax -; AVX512F-NEXT: # kill: def %ax killed %ax killed %eax +; AVX512F-NEXT: # kill: def $ax killed $ax killed $eax ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; @@ -493,7 +520,7 @@ define i16 @v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %c, <16 x floa ; AVX512BW-NEXT: vcmpltps %zmm0, %zmm1, %k1 ; AVX512BW-NEXT: vcmpltps %zmm2, %zmm3, %k0 {%k1} ; AVX512BW-NEXT: kmovd %k0, %eax -; AVX512BW-NEXT: # kill: def %ax killed %ax killed %eax +; AVX512BW-NEXT: # kill: def $ax killed $ax killed $eax ; AVX512BW-NEXT: vzeroupper ; AVX512BW-NEXT: retq %x0 = fcmp ogt <16 x float> %a, %b @@ -506,221 +533,29 @@ define i16 @v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %c, <16 x floa define i64 @v64i8(<64 x i8> %a, <64 x i8> %b, <64 x i8> %c, <64 x i8> %d) { ; SSE-LABEL: v64i8: ; SSE: # %bb.0: -; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm11 ; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm10 -; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm9 +; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm11 ; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: pcmpgtb %xmm6, %xmm2 +; SSE-NEXT: movdqa {{[0-9]+}}(%rsp), %xmm9 ; SSE-NEXT: pcmpgtb %xmm7, %xmm3 -; SSE-NEXT: pcmpgtb %xmm4, %xmm0 +; SSE-NEXT: pcmpgtb %xmm6, %xmm2 ; SSE-NEXT: pcmpgtb %xmm5, %xmm1 -; SSE-NEXT: pcmpgtb {{[0-9]+}}(%rsp), %xmm8 -; SSE-NEXT: pand %xmm2, %xmm8 +; SSE-NEXT: pcmpgtb %xmm4, %xmm0 ; SSE-NEXT: pcmpgtb {{[0-9]+}}(%rsp), %xmm9 ; SSE-NEXT: pand %xmm3, %xmm9 -; SSE-NEXT: pcmpgtb {{[0-9]+}}(%rsp), %xmm10 -; SSE-NEXT: pand %xmm0, %xmm10 +; SSE-NEXT: pcmpgtb {{[0-9]+}}(%rsp), %xmm8 +; SSE-NEXT: pand %xmm2, %xmm8 ; SSE-NEXT: pcmpgtb {{[0-9]+}}(%rsp), %xmm11 ; SSE-NEXT: pand %xmm1, %xmm11 -; SSE-NEXT: pextrb $15, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $14, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $13, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $12, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $11, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $10, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $9, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $8, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $7, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $6, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $5, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $4, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $3, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $2, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $1, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $0, %xmm11, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $15, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $14, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $13, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $12, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $11, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $10, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $9, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $8, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $7, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $6, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $5, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $4, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $3, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $2, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $1, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $0, %xmm10, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $15, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $14, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $13, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $12, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $11, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $10, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $9, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $8, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $7, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $6, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $5, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $4, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $3, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $2, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $1, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $0, %xmm9, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $15, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $14, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $13, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $12, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $11, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $10, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $9, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $8, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $7, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $6, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $5, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $4, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $3, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $2, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $1, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: pextrb $0, %xmm8, %eax -; SSE-NEXT: andb $1, %al -; SSE-NEXT: movb %al, -{{[0-9]+}}(%rsp) -; SSE-NEXT: movl -{{[0-9]+}}(%rsp), %eax -; SSE-NEXT: shll $16, %eax -; SSE-NEXT: movzwl -{{[0-9]+}}(%rsp), %ecx +; SSE-NEXT: pcmpgtb {{[0-9]+}}(%rsp), %xmm10 +; SSE-NEXT: pand %xmm0, %xmm10 +; SSE-NEXT: pmovmskb %xmm10, %eax +; SSE-NEXT: pmovmskb %xmm11, %ecx +; SSE-NEXT: shll $16, %ecx ; SSE-NEXT: orl %eax, %ecx -; SSE-NEXT: movl -{{[0-9]+}}(%rsp), %edx -; SSE-NEXT: shll $16, %edx -; SSE-NEXT: movzwl -{{[0-9]+}}(%rsp), %eax +; SSE-NEXT: pmovmskb %xmm8, %edx +; SSE-NEXT: pmovmskb %xmm9, %eax +; SSE-NEXT: shll $16, %eax ; SSE-NEXT: orl %edx, %eax ; SSE-NEXT: shlq $32, %rax ; SSE-NEXT: orq %rcx, %rax @@ -728,495 +563,90 @@ define i64 @v64i8(<64 x i8> %a, <64 x i8> %b, <64 x i8> %c, <64 x i8> %d) { ; ; AVX1-LABEL: v64i8: ; AVX1: # %bb.0: -; AVX1-NEXT: pushq %rbp -; AVX1-NEXT: .cfi_def_cfa_offset 16 -; AVX1-NEXT: .cfi_offset %rbp, -16 -; AVX1-NEXT: movq %rsp, %rbp -; AVX1-NEXT: .cfi_def_cfa_register %rbp -; AVX1-NEXT: andq $-32, %rsp -; AVX1-NEXT: subq $64, %rsp ; AVX1-NEXT: vextractf128 $1, %ymm3, %xmm8 ; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm9 ; AVX1-NEXT: vpcmpgtb %xmm8, %xmm9, %xmm8 -; AVX1-NEXT: vpcmpgtb %xmm3, %xmm1, %xmm1 -; AVX1-NEXT: vinsertf128 $1, %xmm8, %ymm1, %ymm8 +; AVX1-NEXT: vpcmpgtb %xmm3, %xmm1, %xmm9 ; AVX1-NEXT: vextractf128 $1, %ymm2, %xmm3 ; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 ; AVX1-NEXT: vpcmpgtb %xmm3, %xmm1, %xmm1 ; AVX1-NEXT: vpcmpgtb %xmm2, %xmm0, %xmm0 -; AVX1-NEXT: vinsertf128 $1, %xmm1, %ymm0, %ymm1 -; AVX1-NEXT: vextractf128 $1, %ymm7, %xmm0 -; AVX1-NEXT: vextractf128 $1, %ymm5, %xmm2 -; AVX1-NEXT: vpcmpgtb %xmm0, %xmm2, %xmm0 -; AVX1-NEXT: vpcmpgtb %xmm7, %xmm5, %xmm2 -; AVX1-NEXT: vinsertf128 $1, %xmm0, %ymm2, %ymm0 -; AVX1-NEXT: vandps %ymm0, %ymm8, %ymm0 -; AVX1-NEXT: vextractf128 $1, %ymm6, %xmm2 -; AVX1-NEXT: vextractf128 $1, %ymm4, %xmm3 +; AVX1-NEXT: vextractf128 $1, %ymm7, %xmm2 +; AVX1-NEXT: vextractf128 $1, %ymm5, %xmm3 ; AVX1-NEXT: vpcmpgtb %xmm2, %xmm3, %xmm2 -; AVX1-NEXT: vpcmpgtb %xmm6, %xmm4, %xmm3 -; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm3, %ymm2 -; AVX1-NEXT: vandps %ymm2, %ymm1, %ymm1 -; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2 -; AVX1-NEXT: vpextrb $15, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $14, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $13, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $12, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $11, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $10, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $9, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $8, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $7, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $6, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $5, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $4, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $3, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $2, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $1, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $0, %xmm2, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $15, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $14, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $13, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $12, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $11, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $10, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $9, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $8, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $7, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $6, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $5, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $4, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $3, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $2, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $1, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vpextrb $0, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, (%rsp) -; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1 -; AVX1-NEXT: vpextrb $15, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $14, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $13, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $12, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $11, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $10, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $9, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $8, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $7, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $6, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $5, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $4, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $3, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $2, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $1, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $0, %xmm1, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $15, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $14, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $13, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $12, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $11, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $10, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $9, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $8, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $7, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $6, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $5, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $4, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $3, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $2, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $1, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: vpextrb $0, %xmm0, %eax -; AVX1-NEXT: andb $1, %al -; AVX1-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX1-NEXT: movl (%rsp), %ecx -; AVX1-NEXT: movl {{[0-9]+}}(%rsp), %eax +; AVX1-NEXT: vpand %xmm2, %xmm8, %xmm2 +; AVX1-NEXT: vpcmpgtb %xmm7, %xmm5, %xmm3 +; AVX1-NEXT: vpand %xmm3, %xmm9, %xmm3 +; AVX1-NEXT: vextractf128 $1, %ymm6, %xmm5 +; AVX1-NEXT: vextractf128 $1, %ymm4, %xmm7 +; AVX1-NEXT: vpcmpgtb %xmm5, %xmm7, %xmm5 +; AVX1-NEXT: vpand %xmm5, %xmm1, %xmm1 +; AVX1-NEXT: vpcmpgtb %xmm6, %xmm4, %xmm4 +; AVX1-NEXT: vpand %xmm4, %xmm0, %xmm0 +; AVX1-NEXT: vpmovmskb %xmm0, %eax +; AVX1-NEXT: vpmovmskb %xmm1, %ecx +; AVX1-NEXT: shll $16, %ecx +; AVX1-NEXT: orl %eax, %ecx +; AVX1-NEXT: vpmovmskb %xmm3, %edx +; AVX1-NEXT: vpmovmskb %xmm2, %eax +; AVX1-NEXT: shll $16, %eax +; AVX1-NEXT: orl %edx, %eax ; AVX1-NEXT: shlq $32, %rax ; AVX1-NEXT: orq %rcx, %rax -; AVX1-NEXT: movq %rbp, %rsp -; AVX1-NEXT: popq %rbp ; AVX1-NEXT: vzeroupper ; AVX1-NEXT: retq ; ; AVX2-LABEL: v64i8: ; AVX2: # %bb.0: -; AVX2-NEXT: pushq %rbp -; AVX2-NEXT: .cfi_def_cfa_offset 16 -; AVX2-NEXT: .cfi_offset %rbp, -16 -; AVX2-NEXT: movq %rsp, %rbp -; AVX2-NEXT: .cfi_def_cfa_register %rbp -; AVX2-NEXT: andq $-32, %rsp -; AVX2-NEXT: subq $64, %rsp ; AVX2-NEXT: vpcmpgtb %ymm3, %ymm1, %ymm1 -; AVX2-NEXT: vpcmpgtb %ymm2, %ymm0, %ymm2 -; AVX2-NEXT: vpcmpgtb %ymm7, %ymm5, %ymm0 -; AVX2-NEXT: vpand %ymm0, %ymm1, %ymm0 -; AVX2-NEXT: vpcmpgtb %ymm6, %ymm4, %ymm1 -; AVX2-NEXT: vpand %ymm1, %ymm2, %ymm1 -; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2 -; AVX2-NEXT: vpextrb $15, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $14, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $13, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $12, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $11, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $10, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $9, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $8, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $7, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $6, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $5, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $4, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $3, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $2, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $1, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $0, %xmm2, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $15, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $14, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $13, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $12, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $11, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $10, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $9, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $8, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $7, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $6, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $5, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $4, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $3, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $2, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $1, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vpextrb $0, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, (%rsp) -; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1 -; AVX2-NEXT: vpextrb $15, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $14, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $13, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $12, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $11, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $10, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $9, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $8, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $7, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $6, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $5, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $4, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $3, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $2, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $1, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $0, %xmm1, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $15, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $14, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $13, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $12, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $11, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $10, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $9, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $8, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $7, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $6, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $5, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $4, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $3, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $2, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $1, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: vpextrb $0, %xmm0, %eax -; AVX2-NEXT: andb $1, %al -; AVX2-NEXT: movb %al, {{[0-9]+}}(%rsp) -; AVX2-NEXT: movl (%rsp), %ecx -; AVX2-NEXT: movl {{[0-9]+}}(%rsp), %eax +; AVX2-NEXT: vpcmpgtb %ymm2, %ymm0, %ymm0 +; AVX2-NEXT: vpcmpgtb %ymm7, %ymm5, %ymm2 +; AVX2-NEXT: vpand %ymm2, %ymm1, %ymm1 +; AVX2-NEXT: vpcmpgtb %ymm6, %ymm4, %ymm2 +; AVX2-NEXT: vpand %ymm2, %ymm0, %ymm0 +; AVX2-NEXT: vpmovmskb %ymm0, %ecx +; AVX2-NEXT: vpmovmskb %ymm1, %eax ; AVX2-NEXT: shlq $32, %rax ; AVX2-NEXT: orq %rcx, %rax -; AVX2-NEXT: movq %rbp, %rsp -; AVX2-NEXT: popq %rbp ; AVX2-NEXT: vzeroupper ; AVX2-NEXT: retq ; ; AVX512F-LABEL: v64i8: ; AVX512F: # %bb.0: -; AVX512F-NEXT: pushq %rbp -; AVX512F-NEXT: .cfi_def_cfa_offset 16 -; AVX512F-NEXT: .cfi_offset %rbp, -16 -; AVX512F-NEXT: movq %rsp, %rbp -; AVX512F-NEXT: .cfi_def_cfa_register %rbp -; AVX512F-NEXT: andq $-32, %rsp -; AVX512F-NEXT: subq $64, %rsp ; AVX512F-NEXT: vpcmpgtb %ymm3, %ymm1, %ymm1 +; AVX512F-NEXT: vextracti128 $1, %ymm1, %xmm3 +; AVX512F-NEXT: vpmovsxbd %xmm3, %zmm3 +; AVX512F-NEXT: vptestmd %zmm3, %zmm3, %k1 +; AVX512F-NEXT: vpmovsxbd %xmm1, %zmm1 +; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k2 ; AVX512F-NEXT: vpcmpgtb %ymm2, %ymm0, %ymm0 -; AVX512F-NEXT: vpcmpgtb %ymm7, %ymm5, %ymm2 -; AVX512F-NEXT: vpand %ymm2, %ymm1, %ymm1 -; AVX512F-NEXT: vpcmpgtb %ymm6, %ymm4, %ymm2 -; AVX512F-NEXT: vpand %ymm2, %ymm0, %ymm0 -; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm2 -; AVX512F-NEXT: vpmovsxbd %xmm2, %zmm2 -; AVX512F-NEXT: vpslld $31, %zmm2, %zmm2 -; AVX512F-NEXT: vptestmd %zmm2, %zmm2, %k0 -; AVX512F-NEXT: kmovw %k0, {{[0-9]+}}(%rsp) +; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm1 +; AVX512F-NEXT: vpmovsxbd %xmm1, %zmm1 +; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k3 ; AVX512F-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512F-NEXT: vpslld $31, %zmm0, %zmm0 -; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k0 -; AVX512F-NEXT: kmovw %k0, (%rsp) -; AVX512F-NEXT: vextracti128 $1, %ymm1, %xmm0 +; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k4 +; AVX512F-NEXT: vpcmpgtb %ymm7, %ymm5, %ymm0 +; AVX512F-NEXT: vextracti128 $1, %ymm0, %xmm1 +; AVX512F-NEXT: vpmovsxbd %xmm1, %zmm1 ; AVX512F-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512F-NEXT: vpslld $31, %zmm0, %zmm0 -; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k0 -; AVX512F-NEXT: kmovw %k0, {{[0-9]+}}(%rsp) -; AVX512F-NEXT: vpmovsxbd %xmm1, %zmm0 -; AVX512F-NEXT: vpslld $31, %zmm0, %zmm0 -; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k0 -; AVX512F-NEXT: kmovw %k0, {{[0-9]+}}(%rsp) -; AVX512F-NEXT: movl (%rsp), %ecx -; AVX512F-NEXT: movl {{[0-9]+}}(%rsp), %eax +; AVX512F-NEXT: vpcmpgtb %ymm6, %ymm4, %ymm2 +; AVX512F-NEXT: vextracti128 $1, %ymm2, %xmm3 +; AVX512F-NEXT: vpmovsxbd %xmm3, %zmm3 +; AVX512F-NEXT: vpmovsxbd %xmm2, %zmm2 +; AVX512F-NEXT: vptestmd %zmm2, %zmm2, %k0 {%k4} +; AVX512F-NEXT: kmovw %k0, %eax +; AVX512F-NEXT: vptestmd %zmm3, %zmm3, %k0 {%k3} +; AVX512F-NEXT: kmovw %k0, %ecx +; AVX512F-NEXT: shll $16, %ecx +; AVX512F-NEXT: orl %eax, %ecx +; AVX512F-NEXT: vptestmd %zmm0, %zmm0, %k0 {%k2} +; AVX512F-NEXT: kmovw %k0, %edx +; AVX512F-NEXT: vptestmd %zmm1, %zmm1, %k0 {%k1} +; AVX512F-NEXT: kmovw %k0, %eax +; AVX512F-NEXT: shll $16, %eax +; AVX512F-NEXT: orl %edx, %eax ; AVX512F-NEXT: shlq $32, %rax ; AVX512F-NEXT: orq %rcx, %rax -; AVX512F-NEXT: movq %rbp, %rsp -; AVX512F-NEXT: popq %rbp ; AVX512F-NEXT: vzeroupper ; AVX512F-NEXT: retq ; |