brintos

brintos / llvm-project-archived public Read only

0
0
Text · 7.4 KiB · c9e48f8 Raw
87 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py2; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512bw,+prefer-256-bit | FileCheck %s --check-prefix=AVX256BW3; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512bw,-prefer-256-bit | FileCheck %s --check-prefix=AVX512BWVL4; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,+prefer-256-bit | FileCheck %s --check-prefix=AVX512BW5; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512bw,-prefer-256-bit | FileCheck %s --check-prefix=AVX512BW6 7define <32 x i8> @test_div7_32i8(<32 x i8> %a) {8; AVX256BW-LABEL: test_div7_32i8:9; AVX256BW:       # %bb.0:10; AVX256BW-NEXT:    vpxor %xmm1, %xmm1, %xmm111; AVX256BW-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]12; AVX256BW-NEXT:    vpbroadcastw {{.*#+}} ymm3 = [37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0,37,0]13; AVX256BW-NEXT:    vpmullw %ymm3, %ymm2, %ymm214; AVX256BW-NEXT:    vpsrlw $8, %ymm2, %ymm215; AVX256BW-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]16; AVX256BW-NEXT:    vpmullw %ymm3, %ymm1, %ymm117; AVX256BW-NEXT:    vpsrlw $8, %ymm1, %ymm118; AVX256BW-NEXT:    vpackuswb %ymm2, %ymm1, %ymm119; AVX256BW-NEXT:    vpsubb %ymm1, %ymm0, %ymm020; AVX256BW-NEXT:    vpsrlw $1, %ymm0, %ymm021; AVX256BW-NEXT:    vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm022; AVX256BW-NEXT:    vpaddb %ymm1, %ymm0, %ymm023; AVX256BW-NEXT:    vpsrlw $2, %ymm0, %ymm024; AVX256BW-NEXT:    vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm025; AVX256BW-NEXT:    retq26;27; AVX512BWVL-LABEL: test_div7_32i8:28; AVX512BWVL:       # %bb.0:29; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero30; AVX512BWVL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1 # [37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37]31; AVX512BWVL-NEXT:    vpsrlw $8, %zmm1, %zmm132; AVX512BWVL-NEXT:    vpmovwb %zmm1, %ymm133; AVX512BWVL-NEXT:    vpsubb %ymm1, %ymm0, %ymm034; AVX512BWVL-NEXT:    vpsrlw $1, %ymm0, %ymm035; AVX512BWVL-NEXT:    vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm036; AVX512BWVL-NEXT:    vpaddb %ymm1, %ymm0, %ymm037; AVX512BWVL-NEXT:    vpsrlw $2, %ymm0, %ymm038; AVX512BWVL-NEXT:    vpandd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm039; AVX512BWVL-NEXT:    retq40;41; AVX512BW-LABEL: test_div7_32i8:42; AVX512BW:       # %bb.0:43; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero44; AVX512BW-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1 # [37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37]45; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm146; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm147; AVX512BW-NEXT:    vpsubb %ymm1, %ymm0, %ymm048; AVX512BW-NEXT:    vpsrlw $1, %ymm0, %ymm049; AVX512BW-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm050; AVX512BW-NEXT:    vpaddb %ymm1, %ymm0, %ymm051; AVX512BW-NEXT:    vpsrlw $2, %ymm0, %ymm052; AVX512BW-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm053; AVX512BW-NEXT:    retq54  %res = udiv <32 x i8> %a, <i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7>55  ret <32 x i8> %res56}57 58define <32 x i8> @test_mul_32i8(<32 x i8> %a, <32 x i8> %b) {59; AVX256BW-LABEL: test_mul_32i8:60; AVX256BW:       # %bb.0:61; AVX256BW-NEXT:    vpmullw %ymm1, %ymm0, %ymm262; AVX256BW-NEXT:    vpbroadcastw {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]63; AVX256BW-NEXT:    vpandn %ymm1, %ymm3, %ymm164; AVX256BW-NEXT:    vpmaddubsw %ymm1, %ymm0, %ymm065; AVX256BW-NEXT:    vpsllw $8, %ymm0, %ymm066; AVX256BW-NEXT:    vpternlogq {{.*#+}} ymm0 = ymm0 | (ymm2 & ymm3)67; AVX256BW-NEXT:    retq68;69; AVX512BWVL-LABEL: test_mul_32i8:70; AVX512BWVL:       # %bb.0:71; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero72; AVX512BWVL-NEXT:    vpmovzxbw {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero73; AVX512BWVL-NEXT:    vpmullw %zmm1, %zmm0, %zmm074; AVX512BWVL-NEXT:    vpmovwb %zmm0, %ymm075; AVX512BWVL-NEXT:    retq76;77; AVX512BW-LABEL: test_mul_32i8:78; AVX512BW:       # %bb.0:79; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero,ymm1[16],zero,ymm1[17],zero,ymm1[18],zero,ymm1[19],zero,ymm1[20],zero,ymm1[21],zero,ymm1[22],zero,ymm1[23],zero,ymm1[24],zero,ymm1[25],zero,ymm1[26],zero,ymm1[27],zero,ymm1[28],zero,ymm1[29],zero,ymm1[30],zero,ymm1[31],zero80; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero81; AVX512BW-NEXT:    vpmullw %zmm1, %zmm0, %zmm082; AVX512BW-NEXT:    vpmovwb %zmm0, %ymm083; AVX512BW-NEXT:    retq84  %res = mul <32 x i8> %a, %b85  ret <32 x i8> %res86}87