diff options
Diffstat (limited to 'test/CodeGen/X86/vector-shift-ashr-128.ll')
| -rw-r--r-- | test/CodeGen/X86/vector-shift-ashr-128.ll | 154 |
1 files changed, 106 insertions, 48 deletions
diff --git a/test/CodeGen/X86/vector-shift-ashr-128.ll b/test/CodeGen/X86/vector-shift-ashr-128.ll index ea33f22cc07a0..ca670f40ab3fb 100644 --- a/test/CodeGen/X86/vector-shift-ashr-128.ll +++ b/test/CodeGen/X86/vector-shift-ashr-128.ll @@ -531,23 +531,42 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind { ; XOP-NEXT: vpshab %xmm1, %xmm0, %xmm0 ; XOP-NEXT: retq ; -; AVX512-LABEL: var_shift_v16i8: -; AVX512: # %bb.0: -; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero -; AVX512-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512-NEXT: vpsravd %zmm1, %zmm0, %zmm0 -; AVX512-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512-NEXT: vzeroupper -; AVX512-NEXT: retq +; AVX512DQ-LABEL: var_shift_v16i8: +; AVX512DQ: # %bb.0: +; AVX512DQ-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero +; AVX512DQ-NEXT: vpmovsxbd %xmm0, %zmm0 +; AVX512DQ-NEXT: vpsravd %zmm1, %zmm0, %zmm0 +; AVX512DQ-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512DQ-NEXT: vzeroupper +; AVX512DQ-NEXT: retq ; -; AVX512VL-LABEL: var_shift_v16i8: -; AVX512VL: # %bb.0: -; AVX512VL-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero -; AVX512VL-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512VL-NEXT: vpsravd %zmm1, %zmm0, %zmm0 -; AVX512VL-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512VL-NEXT: vzeroupper -; AVX512VL-NEXT: retq +; AVX512BW-LABEL: var_shift_v16i8: +; AVX512BW: # %bb.0: +; AVX512BW-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero +; AVX512BW-NEXT: vpmovsxbw %xmm0, %ymm0 +; AVX512BW-NEXT: vpsravw %zmm1, %zmm0, %zmm0 +; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0 +; AVX512BW-NEXT: # kill: def %xmm0 killed %xmm0 killed %ymm0 +; AVX512BW-NEXT: vzeroupper +; AVX512BW-NEXT: retq +; +; AVX512DQVL-LABEL: var_shift_v16i8: +; AVX512DQVL: # %bb.0: +; AVX512DQVL-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero +; AVX512DQVL-NEXT: vpmovsxbd %xmm0, %zmm0 +; AVX512DQVL-NEXT: vpsravd %zmm1, %zmm0, %zmm0 +; AVX512DQVL-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512DQVL-NEXT: vzeroupper +; AVX512DQVL-NEXT: retq +; +; AVX512BWVL-LABEL: var_shift_v16i8: +; AVX512BWVL: # %bb.0: +; AVX512BWVL-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero +; AVX512BWVL-NEXT: vpmovsxbw %xmm0, %ymm0 +; AVX512BWVL-NEXT: vpsravw %ymm1, %ymm0, %ymm0 +; AVX512BWVL-NEXT: vpmovwb %ymm0, %xmm0 +; AVX512BWVL-NEXT: vzeroupper +; AVX512BWVL-NEXT: retq ; ; X32-SSE-LABEL: var_shift_v16i8: ; X32-SSE: # %bb.0: @@ -948,25 +967,46 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, <16 x i8> %b) nounwind { ; XOPAVX2-NEXT: vpshab %xmm1, %xmm0, %xmm0 ; XOPAVX2-NEXT: retq ; -; AVX512-LABEL: splatvar_shift_v16i8: -; AVX512: # %bb.0: -; AVX512-NEXT: vpbroadcastb %xmm1, %xmm1 -; AVX512-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero -; AVX512-NEXT: vpsravd %zmm1, %zmm0, %zmm0 -; AVX512-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512-NEXT: vzeroupper -; AVX512-NEXT: retq +; AVX512DQ-LABEL: splatvar_shift_v16i8: +; AVX512DQ: # %bb.0: +; AVX512DQ-NEXT: vpbroadcastb %xmm1, %xmm1 +; AVX512DQ-NEXT: vpmovsxbd %xmm0, %zmm0 +; AVX512DQ-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero +; AVX512DQ-NEXT: vpsravd %zmm1, %zmm0, %zmm0 +; AVX512DQ-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512DQ-NEXT: vzeroupper +; AVX512DQ-NEXT: retq ; -; AVX512VL-LABEL: splatvar_shift_v16i8: -; AVX512VL: # %bb.0: -; AVX512VL-NEXT: vpbroadcastb %xmm1, %xmm1 -; AVX512VL-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512VL-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero -; AVX512VL-NEXT: vpsravd %zmm1, %zmm0, %zmm0 -; AVX512VL-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512VL-NEXT: vzeroupper -; AVX512VL-NEXT: retq +; AVX512BW-LABEL: splatvar_shift_v16i8: +; AVX512BW: # %bb.0: +; AVX512BW-NEXT: vpbroadcastb %xmm1, %xmm1 +; AVX512BW-NEXT: vpmovsxbw %xmm0, %ymm0 +; AVX512BW-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero +; AVX512BW-NEXT: vpsravw %zmm1, %zmm0, %zmm0 +; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0 +; AVX512BW-NEXT: # kill: def %xmm0 killed %xmm0 killed %ymm0 +; AVX512BW-NEXT: vzeroupper +; AVX512BW-NEXT: retq +; +; AVX512DQVL-LABEL: splatvar_shift_v16i8: +; AVX512DQVL: # %bb.0: +; AVX512DQVL-NEXT: vpbroadcastb %xmm1, %xmm1 +; AVX512DQVL-NEXT: vpmovsxbd %xmm0, %zmm0 +; AVX512DQVL-NEXT: vpmovzxbd {{.*#+}} zmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero,xmm1[2],zero,zero,zero,xmm1[3],zero,zero,zero,xmm1[4],zero,zero,zero,xmm1[5],zero,zero,zero,xmm1[6],zero,zero,zero,xmm1[7],zero,zero,zero,xmm1[8],zero,zero,zero,xmm1[9],zero,zero,zero,xmm1[10],zero,zero,zero,xmm1[11],zero,zero,zero,xmm1[12],zero,zero,zero,xmm1[13],zero,zero,zero,xmm1[14],zero,zero,zero,xmm1[15],zero,zero,zero +; AVX512DQVL-NEXT: vpsravd %zmm1, %zmm0, %zmm0 +; AVX512DQVL-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512DQVL-NEXT: vzeroupper +; AVX512DQVL-NEXT: retq +; +; AVX512BWVL-LABEL: splatvar_shift_v16i8: +; AVX512BWVL: # %bb.0: +; AVX512BWVL-NEXT: vpbroadcastb %xmm1, %xmm1 +; AVX512BWVL-NEXT: vpmovsxbw %xmm0, %ymm0 +; AVX512BWVL-NEXT: vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero +; AVX512BWVL-NEXT: vpsravw %ymm1, %ymm0, %ymm0 +; AVX512BWVL-NEXT: vpmovwb %ymm0, %xmm0 +; AVX512BWVL-NEXT: vzeroupper +; AVX512BWVL-NEXT: retq ; ; X32-SSE-LABEL: splatvar_shift_v16i8: ; X32-SSE: # %bb.0: @@ -1441,21 +1481,39 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) nounwind { ; XOP-NEXT: vpshab %xmm1, %xmm0, %xmm0 ; XOP-NEXT: retq ; -; AVX512-LABEL: constant_shift_v16i8: -; AVX512: # %bb.0: -; AVX512-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512-NEXT: vpsravd {{.*}}(%rip), %zmm0, %zmm0 -; AVX512-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512-NEXT: vzeroupper -; AVX512-NEXT: retq +; AVX512DQ-LABEL: constant_shift_v16i8: +; AVX512DQ: # %bb.0: +; AVX512DQ-NEXT: vpmovsxbd %xmm0, %zmm0 +; AVX512DQ-NEXT: vpsravd {{.*}}(%rip), %zmm0, %zmm0 +; AVX512DQ-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512DQ-NEXT: vzeroupper +; AVX512DQ-NEXT: retq ; -; AVX512VL-LABEL: constant_shift_v16i8: -; AVX512VL: # %bb.0: -; AVX512VL-NEXT: vpmovsxbd %xmm0, %zmm0 -; AVX512VL-NEXT: vpsravd {{.*}}(%rip), %zmm0, %zmm0 -; AVX512VL-NEXT: vpmovdb %zmm0, %xmm0 -; AVX512VL-NEXT: vzeroupper -; AVX512VL-NEXT: retq +; AVX512BW-LABEL: constant_shift_v16i8: +; AVX512BW: # %bb.0: +; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm1 = [0,1,2,3,4,5,6,7,7,6,5,4,3,2,1,0] +; AVX512BW-NEXT: vpmovsxbw %xmm0, %ymm0 +; AVX512BW-NEXT: vpsravw %zmm1, %zmm0, %zmm0 +; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0 +; AVX512BW-NEXT: # kill: def %xmm0 killed %xmm0 killed %ymm0 +; AVX512BW-NEXT: vzeroupper +; AVX512BW-NEXT: retq +; +; AVX512DQVL-LABEL: constant_shift_v16i8: +; AVX512DQVL: # %bb.0: +; AVX512DQVL-NEXT: vpmovsxbd %xmm0, %zmm0 +; AVX512DQVL-NEXT: vpsravd {{.*}}(%rip), %zmm0, %zmm0 +; AVX512DQVL-NEXT: vpmovdb %zmm0, %xmm0 +; AVX512DQVL-NEXT: vzeroupper +; AVX512DQVL-NEXT: retq +; +; AVX512BWVL-LABEL: constant_shift_v16i8: +; AVX512BWVL: # %bb.0: +; AVX512BWVL-NEXT: vpmovsxbw %xmm0, %ymm0 +; AVX512BWVL-NEXT: vpsravw {{.*}}(%rip), %ymm0, %ymm0 +; AVX512BWVL-NEXT: vpmovwb %ymm0, %xmm0 +; AVX512BWVL-NEXT: vzeroupper +; AVX512BWVL-NEXT: retq ; ; X32-SSE-LABEL: constant_shift_v16i8: ; X32-SSE: # %bb.0: |
