903 lines · plain
1# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py2# RUN: llc -mtriple=amdgcn -mcpu=gfx1250 -run-pass post-RA-hazard-rec %s -o - | FileCheck -check-prefix=GFX1250 %s3 4# WMMA writes: D0, WMMA reads: A0/B0/Index05# VALU writes: D1, VALU reads: Use16# Hards could be:7# RAW: D0 overlaps Use18# WAW: D0 overlaps D19# WAR: A0/B0/Index0 overlaps D110 11---12name: test_wmma_f32_16x16x32_bf16_D0_overlaps_Use113body: |14 bb.0:15 ; GFX1250-LABEL: name: test_wmma_f32_16x16x32_bf16_D0_overlaps_Use116 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec17 ; GFX1250-NEXT: V_NOP_e32 implicit $exec18 ; GFX1250-NEXT: V_NOP_e32 implicit $exec19 ; GFX1250-NEXT: V_NOP_e32 implicit $exec20 ; GFX1250-NEXT: V_NOP_e32 implicit $exec21 ; GFX1250-NEXT: $vgpr25 = V_ADD_F32_e32 $vgpr24, $vgpr16, implicit $mode, implicit $exec22 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec23 $vgpr25 = V_ADD_F32_e32 $vgpr24, $vgpr16, implicit $mode, implicit $exec24...25 26---27name: test_wmma_f32_16x16x32_bf16_D0_overlaps_Use1_with_4_valus_in_between28body: |29 bb.0:30 ; GFX1250-LABEL: name: test_wmma_f32_16x16x32_bf16_D0_overlaps_Use1_with_4_valus_in_between31 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec32 ; GFX1250-NEXT: $vgpr26 = V_MOV_B32_e32 26, implicit $exec33 ; GFX1250-NEXT: $vgpr27 = V_MOV_B32_e32 27, implicit $exec34 ; GFX1250-NEXT: $vgpr28 = V_MOV_B32_e32 28, implicit $exec35 ; GFX1250-NEXT: $vgpr29 = V_MOV_B32_e32 29, implicit $exec36 ; GFX1250-NEXT: $vgpr25 = V_ADD_F32_e32 $vgpr24, $vgpr16, implicit $mode, implicit $exec37 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec38 $vgpr26 = V_MOV_B32_e32 26, implicit $exec39 $vgpr27 = V_MOV_B32_e32 27, implicit $exec40 $vgpr28 = V_MOV_B32_e32 28, implicit $exec41 $vgpr29 = V_MOV_B32_e32 29, implicit $exec42 $vgpr25 = V_ADD_F32_e32 $vgpr24, $vgpr16, implicit $mode, implicit $exec43...44 45---46name: test_wmma_f32_16x16x32_bf16_D0_overlaps_Use1_with_4_salus_in_between47body: |48 bb.0:49 ; GFX1250-LABEL: name: test_wmma_f32_16x16x32_bf16_D0_overlaps_Use1_with_4_salus_in_between50 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec51 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 052 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 153 ; GFX1250-NEXT: $sgpr2 = S_MOV_B32 254 ; GFX1250-NEXT: $sgpr3 = S_MOV_B32 355 ; GFX1250-NEXT: V_NOP_e32 implicit $exec56 ; GFX1250-NEXT: V_NOP_e32 implicit $exec57 ; GFX1250-NEXT: V_NOP_e32 implicit $exec58 ; GFX1250-NEXT: V_NOP_e32 implicit $exec59 ; GFX1250-NEXT: $vgpr25 = V_ADD_F32_e32 $vgpr24, $vgpr16, implicit $mode, implicit $exec60 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec61 $sgpr0 = S_MOV_B32 062 $sgpr1 = S_MOV_B32 163 $sgpr2 = S_MOV_B32 264 $sgpr3 = S_MOV_B32 365 $vgpr25 = V_ADD_F32_e32 $vgpr24, $vgpr16, implicit $mode, implicit $exec66...67 68---69name: test_wmma_f32_16x16x32_bf16_D0_overlaps_D170body: |71 bb.0:72 ; GFX1250-LABEL: name: test_wmma_f32_16x16x32_bf16_D0_overlaps_D173 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec74 ; GFX1250-NEXT: V_NOP_e32 implicit $exec75 ; GFX1250-NEXT: V_NOP_e32 implicit $exec76 ; GFX1250-NEXT: V_NOP_e32 implicit $exec77 ; GFX1250-NEXT: V_NOP_e32 implicit $exec78 ; GFX1250-NEXT: $vgpr16 = V_ADD_F32_e32 $vgpr24, $vgpr25, implicit $mode, implicit $exec79 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec80 $vgpr16 = V_ADD_F32_e32 $vgpr24, $vgpr25, implicit $mode, implicit $exec81...82 83---84name: test_wmma_f32_16x16x32_bf16_A0_overlaps_D185body: |86 bb.0:87 ; GFX1250-LABEL: name: test_wmma_f32_16x16x32_bf16_A0_overlaps_D188 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec89 ; GFX1250-NEXT: V_NOP_e32 implicit $exec90 ; GFX1250-NEXT: V_NOP_e32 implicit $exec91 ; GFX1250-NEXT: V_NOP_e32 implicit $exec92 ; GFX1250-NEXT: V_NOP_e32 implicit $exec93 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr24, $vgpr25, implicit $mode, implicit $exec94 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec95 $vgpr0 = V_ADD_F32_e32 $vgpr24, $vgpr25, implicit $mode, implicit $exec96...97 98---99name: test_wmma_f32_16x16x32_bf16_B0_overlaps_D1100body: |101 bb.0:102 ; GFX1250-LABEL: name: test_wmma_f32_16x16x32_bf16_B0_overlaps_D1103 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec104 ; GFX1250-NEXT: V_NOP_e32 implicit $exec105 ; GFX1250-NEXT: V_NOP_e32 implicit $exec106 ; GFX1250-NEXT: V_NOP_e32 implicit $exec107 ; GFX1250-NEXT: V_NOP_e32 implicit $exec108 ; GFX1250-NEXT: $vgpr8 = V_ADD_F32_e32 $vgpr24, $vgpr25, implicit $mode, implicit $exec109 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_F32_16X16X32_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, 0, implicit $exec110 $vgpr8 = V_ADD_F32_e32 $vgpr24, $vgpr25, implicit $mode, implicit $exec111...112 113---114name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_Use1115body: |116 bb.0:117 ; GFX1250-LABEL: name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_Use1118 ; GFX1250: early-clobber $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec119 ; GFX1250-NEXT: V_NOP_e32 implicit $exec120 ; GFX1250-NEXT: V_NOP_e32 implicit $exec121 ; GFX1250-NEXT: V_NOP_e32 implicit $exec122 ; GFX1250-NEXT: V_NOP_e32 implicit $exec123 ; GFX1250-NEXT: $vgpr27 = V_ADD_F32_e32 $vgpr22, $vgpr26, implicit $mode, implicit $exec124 $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec125 $vgpr27 = V_ADD_F32_e32 $vgpr22, $vgpr26, implicit $mode, implicit $exec126...127 128---129name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_Use1_with_4_valus_in_between130body: |131 bb.0:132 ; GFX1250-LABEL: name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_Use1_with_4_valus_in_between133 ; GFX1250: early-clobber $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec134 ; GFX1250-NEXT: $vgpr26 = V_MOV_B32_e32 26, implicit $exec135 ; GFX1250-NEXT: $vgpr27 = V_MOV_B32_e32 27, implicit $exec136 ; GFX1250-NEXT: $vgpr28 = V_MOV_B32_e32 28, implicit $exec137 ; GFX1250-NEXT: $vgpr29 = V_MOV_B32_e32 29, implicit $exec138 ; GFX1250-NEXT: $vgpr31 = V_ADD_F32_e32 $vgpr22, $vgpr30, implicit $mode, implicit $exec139 $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec140 $vgpr26 = V_MOV_B32_e32 26, implicit $exec141 $vgpr27 = V_MOV_B32_e32 27, implicit $exec142 $vgpr28 = V_MOV_B32_e32 28, implicit $exec143 $vgpr29 = V_MOV_B32_e32 29, implicit $exec144 $vgpr31 = V_ADD_F32_e32 $vgpr22, $vgpr30, implicit $mode, implicit $exec145...146 147---148name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_Use1_with_4_salus_in_between149body: |150 bb.0:151 ; GFX1250-LABEL: name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_Use1_with_4_salus_in_between152 ; GFX1250: early-clobber $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec153 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 0154 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 1155 ; GFX1250-NEXT: $sgpr2 = S_MOV_B32 2156 ; GFX1250-NEXT: $sgpr3 = S_MOV_B32 3157 ; GFX1250-NEXT: V_NOP_e32 implicit $exec158 ; GFX1250-NEXT: V_NOP_e32 implicit $exec159 ; GFX1250-NEXT: V_NOP_e32 implicit $exec160 ; GFX1250-NEXT: V_NOP_e32 implicit $exec161 ; GFX1250-NEXT: $vgpr27 = V_ADD_F32_e32 $vgpr22, $vgpr26, implicit $mode, implicit $exec162 $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec163 $sgpr0 = S_MOV_B32 0164 $sgpr1 = S_MOV_B32 1165 $sgpr2 = S_MOV_B32 2166 $sgpr3 = S_MOV_B32 3167 $vgpr27 = V_ADD_F32_e32 $vgpr22, $vgpr26, implicit $mode, implicit $exec168...169 170---171name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_D1172body: |173 bb.0:174 ; GFX1250-LABEL: name: test_wmma_f16_16x16x64_fp8_fp8_D0_overlaps_D1175 ; GFX1250: early-clobber $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec176 ; GFX1250-NEXT: V_NOP_e32 implicit $exec177 ; GFX1250-NEXT: V_NOP_e32 implicit $exec178 ; GFX1250-NEXT: V_NOP_e32 implicit $exec179 ; GFX1250-NEXT: V_NOP_e32 implicit $exec180 ; GFX1250-NEXT: $vgpr22 = V_ADD_F32_e32 $vgpr26, $vgpr27, implicit $mode, implicit $exec181 $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec182 $vgpr22 = V_ADD_F32_e32 $vgpr26, $vgpr27, implicit $mode, implicit $exec183...184 185---186name: test_wmma_f16_16x16x64_fp8_fp8_A0_overlaps_D1187body: |188 bb.0:189 ; GFX1250-LABEL: name: test_wmma_f16_16x16x64_fp8_fp8_A0_overlaps_D1190 ; GFX1250: early-clobber $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec191 ; GFX1250-NEXT: V_NOP_e32 implicit $exec192 ; GFX1250-NEXT: V_NOP_e32 implicit $exec193 ; GFX1250-NEXT: V_NOP_e32 implicit $exec194 ; GFX1250-NEXT: V_NOP_e32 implicit $exec195 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr26, $vgpr27, implicit $mode, implicit $exec196 $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec197 $vgpr0 = V_ADD_F32_e32 $vgpr26, $vgpr27, implicit $mode, implicit $exec198...199 200---201name: test_wmma_f16_16x16x64_fp8_fp8_B0_overlaps_D1202body: |203 bb.0:204 ; GFX1250-LABEL: name: test_wmma_f16_16x16x64_fp8_fp8_B0_overlaps_D1205 ; GFX1250: early-clobber $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec206 ; GFX1250-NEXT: V_NOP_e32 implicit $exec207 ; GFX1250-NEXT: V_NOP_e32 implicit $exec208 ; GFX1250-NEXT: V_NOP_e32 implicit $exec209 ; GFX1250-NEXT: V_NOP_e32 implicit $exec210 ; GFX1250-NEXT: $vgpr8 = V_ADD_F32_e32 $vgpr26, $vgpr27, implicit $mode, implicit $exec211 $vgpr22_vgpr23_vgpr24_vgpr25 = V_WMMA_F16_16X16X64_FP8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, 8, killed $vgpr22_vgpr23_vgpr24_vgpr25, 0, 0, 0, 0, implicit $exec212 $vgpr8 = V_ADD_F32_e32 $vgpr26, $vgpr27, implicit $mode, implicit $exec213...214 215---216name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_Use1217body: |218 bb.0:219 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_Use1220 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec221 ; GFX1250-NEXT: V_NOP_e32 implicit $exec222 ; GFX1250-NEXT: V_NOP_e32 implicit $exec223 ; GFX1250-NEXT: V_NOP_e32 implicit $exec224 ; GFX1250-NEXT: V_NOP_e32 implicit $exec225 ; GFX1250-NEXT: $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec226 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec227 $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec228...229 230---231name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_Use1_with_4_valus_in_between232body: |233 bb.0:234 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_Use1_with_4_valus_in_between235 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec236 ; GFX1250-NEXT: $vgpr40 = V_MOV_B32_e32 40, implicit $exec237 ; GFX1250-NEXT: $vgpr41 = V_MOV_B32_e32 41, implicit $exec238 ; GFX1250-NEXT: $vgpr42 = V_MOV_B32_e32 42, implicit $exec239 ; GFX1250-NEXT: $vgpr43 = V_MOV_B32_e32 43, implicit $exec240 ; GFX1250-NEXT: $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec241 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec242 $vgpr40 = V_MOV_B32_e32 40, implicit $exec243 $vgpr41 = V_MOV_B32_e32 41, implicit $exec244 $vgpr42 = V_MOV_B32_e32 42, implicit $exec245 $vgpr43 = V_MOV_B32_e32 43, implicit $exec246 $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec247...248 249---250name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_Use1_with_4_salus_in_between251body: |252 bb.0:253 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_Use1_with_4_salus_in_between254 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec255 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 0256 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 1257 ; GFX1250-NEXT: $sgpr2 = S_MOV_B32 2258 ; GFX1250-NEXT: $sgpr3 = S_MOV_B32 3259 ; GFX1250-NEXT: V_NOP_e32 implicit $exec260 ; GFX1250-NEXT: V_NOP_e32 implicit $exec261 ; GFX1250-NEXT: V_NOP_e32 implicit $exec262 ; GFX1250-NEXT: V_NOP_e32 implicit $exec263 ; GFX1250-NEXT: $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec264 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec265 $sgpr0 = S_MOV_B32 0266 $sgpr1 = S_MOV_B32 1267 $sgpr2 = S_MOV_B32 2268 $sgpr3 = S_MOV_B32 3269 $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec270...271 272---273name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_D1274body: |275 bb.0:276 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_NoF8_D0_overlaps_D1277 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec278 ; GFX1250-NEXT: V_NOP_e32 implicit $exec279 ; GFX1250-NEXT: V_NOP_e32 implicit $exec280 ; GFX1250-NEXT: V_NOP_e32 implicit $exec281 ; GFX1250-NEXT: V_NOP_e32 implicit $exec282 ; GFX1250-NEXT: $vgpr32 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec283 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec284 $vgpr32 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec285...286 287---288name: test_wmma_F32_16x16x128_F8F6F4_NoF8_A0_overlaps_D1289body: |290 bb.0:291 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_NoF8_A0_overlaps_D1292 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec293 ; GFX1250-NEXT: V_NOP_e32 implicit $exec294 ; GFX1250-NEXT: V_NOP_e32 implicit $exec295 ; GFX1250-NEXT: V_NOP_e32 implicit $exec296 ; GFX1250-NEXT: V_NOP_e32 implicit $exec297 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec298 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec299 $vgpr0 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec300...301 302---303name: test_wmma_F32_16x16x128_F8F6F4_NoF8_B0_overlaps_D1304body: |305 bb.0:306 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_NoF8_B0_overlaps_D1307 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec308 ; GFX1250-NEXT: V_NOP_e32 implicit $exec309 ; GFX1250-NEXT: V_NOP_e32 implicit $exec310 ; GFX1250-NEXT: V_NOP_e32 implicit $exec311 ; GFX1250-NEXT: V_NOP_e32 implicit $exec312 ; GFX1250-NEXT: $vgpr16 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec313 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 2, 2, 0, 0, implicit $exec314 $vgpr16 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec315...316 317---318name: test_wmma_I32_16x16x64_IU8_D0_overlaps_Use1319body: |320 bb.0:321 ; GFX1250-LABEL: name: test_wmma_I32_16x16x64_IU8_D0_overlaps_Use1322 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec323 ; GFX1250-NEXT: V_NOP_e32 implicit $exec324 ; GFX1250-NEXT: V_NOP_e32 implicit $exec325 ; GFX1250-NEXT: V_NOP_e32 implicit $exec326 ; GFX1250-NEXT: V_NOP_e32 implicit $exec327 ; GFX1250-NEXT: V_NOP_e32 implicit $exec328 ; GFX1250-NEXT: V_NOP_e32 implicit $exec329 ; GFX1250-NEXT: V_NOP_e32 implicit $exec330 ; GFX1250-NEXT: V_NOP_e32 implicit $exec331 ; GFX1250-NEXT: $vgpr32 = V_ADD_F32_e32 $vgpr16, $vgpr33, implicit $mode, implicit $exec332 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec333 $vgpr32 = V_ADD_F32_e32 $vgpr16, $vgpr33, implicit $mode, implicit $exec334...335 336---337name: test_wmma_I32_16x16x64_IU8_D0_overlaps_Use1_with_8_valus_in_between338body: |339 bb.0:340 ; GFX1250-LABEL: name: test_wmma_I32_16x16x64_IU8_D0_overlaps_Use1_with_8_valus_in_between341 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec342 ; GFX1250-NEXT: $vgpr40 = V_MOV_B32_e32 40, implicit $exec343 ; GFX1250-NEXT: $vgpr41 = V_MOV_B32_e32 41, implicit $exec344 ; GFX1250-NEXT: $vgpr42 = V_MOV_B32_e32 42, implicit $exec345 ; GFX1250-NEXT: $vgpr43 = V_MOV_B32_e32 43, implicit $exec346 ; GFX1250-NEXT: $vgpr44 = V_MOV_B32_e32 44, implicit $exec347 ; GFX1250-NEXT: $vgpr45 = V_MOV_B32_e32 45, implicit $exec348 ; GFX1250-NEXT: $vgpr46 = V_MOV_B32_e32 46, implicit $exec349 ; GFX1250-NEXT: $vgpr47 = V_MOV_B32_e32 47, implicit $exec350 ; GFX1250-NEXT: $vgpr32 = V_ADD_F32_e32 $vgpr16, $vgpr33, implicit $mode, implicit $exec351 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec352 $vgpr40 = V_MOV_B32_e32 40, implicit $exec353 $vgpr41 = V_MOV_B32_e32 41, implicit $exec354 $vgpr42 = V_MOV_B32_e32 42, implicit $exec355 $vgpr43 = V_MOV_B32_e32 43, implicit $exec356 $vgpr44 = V_MOV_B32_e32 44, implicit $exec357 $vgpr45 = V_MOV_B32_e32 45, implicit $exec358 $vgpr46 = V_MOV_B32_e32 46, implicit $exec359 $vgpr47 = V_MOV_B32_e32 47, implicit $exec360 $vgpr32 = V_ADD_F32_e32 $vgpr16, $vgpr33, implicit $mode, implicit $exec361...362 363---364name: test_wmma_I32_16x16x64_IU8_D0_overlaps_Use1_with_8_salus_in_between365body: |366 bb.0:367 ; GFX1250-LABEL: name: test_wmma_I32_16x16x64_IU8_D0_overlaps_Use1_with_8_salus_in_between368 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec369 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 0370 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 1371 ; GFX1250-NEXT: $sgpr2 = S_MOV_B32 2372 ; GFX1250-NEXT: $sgpr3 = S_MOV_B32 3373 ; GFX1250-NEXT: $sgpr4 = S_MOV_B32 4374 ; GFX1250-NEXT: $sgpr5 = S_MOV_B32 5375 ; GFX1250-NEXT: $sgpr6 = S_MOV_B32 6376 ; GFX1250-NEXT: $sgpr7 = S_MOV_B32 7377 ; GFX1250-NEXT: V_NOP_e32 implicit $exec378 ; GFX1250-NEXT: V_NOP_e32 implicit $exec379 ; GFX1250-NEXT: V_NOP_e32 implicit $exec380 ; GFX1250-NEXT: V_NOP_e32 implicit $exec381 ; GFX1250-NEXT: V_NOP_e32 implicit $exec382 ; GFX1250-NEXT: V_NOP_e32 implicit $exec383 ; GFX1250-NEXT: V_NOP_e32 implicit $exec384 ; GFX1250-NEXT: V_NOP_e32 implicit $exec385 ; GFX1250-NEXT: $vgpr32 = V_ADD_F32_e32 $vgpr16, $vgpr33, implicit $mode, implicit $exec386 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec387 $sgpr0 = S_MOV_B32 0388 $sgpr1 = S_MOV_B32 1389 $sgpr2 = S_MOV_B32 2390 $sgpr3 = S_MOV_B32 3391 $sgpr4 = S_MOV_B32 4392 $sgpr5 = S_MOV_B32 5393 $sgpr6 = S_MOV_B32 6394 $sgpr7 = S_MOV_B32 7395 $vgpr32 = V_ADD_F32_e32 $vgpr16, $vgpr33, implicit $mode, implicit $exec396...397 398---399name: test_wmma_I32_16x16x64_IU8_D0_overlaps_D1400body: |401 bb.0:402 ; GFX1250-LABEL: name: test_wmma_I32_16x16x64_IU8_D0_overlaps_D1403 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec404 ; GFX1250-NEXT: V_NOP_e32 implicit $exec405 ; GFX1250-NEXT: V_NOP_e32 implicit $exec406 ; GFX1250-NEXT: V_NOP_e32 implicit $exec407 ; GFX1250-NEXT: V_NOP_e32 implicit $exec408 ; GFX1250-NEXT: V_NOP_e32 implicit $exec409 ; GFX1250-NEXT: V_NOP_e32 implicit $exec410 ; GFX1250-NEXT: V_NOP_e32 implicit $exec411 ; GFX1250-NEXT: V_NOP_e32 implicit $exec412 ; GFX1250-NEXT: $vgpr16 = V_ADD_F32_e32 $vgpr32, $vgpr33, implicit $mode, implicit $exec413 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec414 $vgpr16 = V_ADD_F32_e32 $vgpr32, $vgpr33, implicit $mode, implicit $exec415...416 417---418name: test_wmma_I32_16x16x64_IU8_A0_overlaps_D1419body: |420 bb.0:421 ; GFX1250-LABEL: name: test_wmma_I32_16x16x64_IU8_A0_overlaps_D1422 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec423 ; GFX1250-NEXT: V_NOP_e32 implicit $exec424 ; GFX1250-NEXT: V_NOP_e32 implicit $exec425 ; GFX1250-NEXT: V_NOP_e32 implicit $exec426 ; GFX1250-NEXT: V_NOP_e32 implicit $exec427 ; GFX1250-NEXT: V_NOP_e32 implicit $exec428 ; GFX1250-NEXT: V_NOP_e32 implicit $exec429 ; GFX1250-NEXT: V_NOP_e32 implicit $exec430 ; GFX1250-NEXT: V_NOP_e32 implicit $exec431 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr32, $vgpr33, implicit $mode, implicit $exec432 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec433 $vgpr0 = V_ADD_F32_e32 $vgpr32, $vgpr33, implicit $mode, implicit $exec434...435 436---437name: test_wmma_I32_16x16x64_IU8_B0_overlaps_D1438body: |439 bb.0:440 ; GFX1250-LABEL: name: test_wmma_I32_16x16x64_IU8_B0_overlaps_D1441 ; GFX1250: early-clobber $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec442 ; GFX1250-NEXT: V_NOP_e32 implicit $exec443 ; GFX1250-NEXT: V_NOP_e32 implicit $exec444 ; GFX1250-NEXT: V_NOP_e32 implicit $exec445 ; GFX1250-NEXT: V_NOP_e32 implicit $exec446 ; GFX1250-NEXT: V_NOP_e32 implicit $exec447 ; GFX1250-NEXT: V_NOP_e32 implicit $exec448 ; GFX1250-NEXT: V_NOP_e32 implicit $exec449 ; GFX1250-NEXT: V_NOP_e32 implicit $exec450 ; GFX1250-NEXT: $vgpr8 = V_ADD_F32_e32 $vgpr32, $vgpr33, implicit $mode, implicit $exec451 $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23 = V_WMMA_I32_16X16X64_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, 0, 0, 0, implicit $exec452 $vgpr8 = V_ADD_F32_e32 $vgpr32, $vgpr33, implicit $mode, implicit $exec453...454 455---456name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_Use1457body: |458 bb.0:459 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_Use1460 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 2, 0, 0, implicit $exec461 ; GFX1250-NEXT: V_NOP_e32 implicit $exec462 ; GFX1250-NEXT: V_NOP_e32 implicit $exec463 ; GFX1250-NEXT: V_NOP_e32 implicit $exec464 ; GFX1250-NEXT: V_NOP_e32 implicit $exec465 ; GFX1250-NEXT: V_NOP_e32 implicit $exec466 ; GFX1250-NEXT: V_NOP_e32 implicit $exec467 ; GFX1250-NEXT: V_NOP_e32 implicit $exec468 ; GFX1250-NEXT: V_NOP_e32 implicit $exec469 ; GFX1250-NEXT: $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec470 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 2, 0, 0, implicit $exec471 $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec472...473 474---475name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_Use1_with_8_valus_in_between476body: |477 bb.0:478 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_Use1_with_8_valus_in_between479 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 2, 0, 0, implicit $exec480 ; GFX1250-NEXT: $vgpr40 = V_MOV_B32_e32 40, implicit $exec481 ; GFX1250-NEXT: $vgpr41 = V_MOV_B32_e32 41, implicit $exec482 ; GFX1250-NEXT: $vgpr42 = V_MOV_B32_e32 42, implicit $exec483 ; GFX1250-NEXT: $vgpr43 = V_MOV_B32_e32 43, implicit $exec484 ; GFX1250-NEXT: $vgpr44 = V_MOV_B32_e32 44, implicit $exec485 ; GFX1250-NEXT: $vgpr45 = V_MOV_B32_e32 45, implicit $exec486 ; GFX1250-NEXT: $vgpr46 = V_MOV_B32_e32 46, implicit $exec487 ; GFX1250-NEXT: $vgpr47 = V_MOV_B32_e32 47, implicit $exec488 ; GFX1250-NEXT: $vgpr49 = V_ADD_F32_e32 $vgpr32, $vgpr48, implicit $mode, implicit $exec489 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 2, 0, 0, implicit $exec490 $vgpr40 = V_MOV_B32_e32 40, implicit $exec491 $vgpr41 = V_MOV_B32_e32 41, implicit $exec492 $vgpr42 = V_MOV_B32_e32 42, implicit $exec493 $vgpr43 = V_MOV_B32_e32 43, implicit $exec494 $vgpr44 = V_MOV_B32_e32 44, implicit $exec495 $vgpr45 = V_MOV_B32_e32 45, implicit $exec496 $vgpr46 = V_MOV_B32_e32 46, implicit $exec497 $vgpr47 = V_MOV_B32_e32 47, implicit $exec498 $vgpr49 = V_ADD_F32_e32 $vgpr32, $vgpr48, implicit $mode, implicit $exec499...500 501---502name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_Use1_with_8_salus_in_between503body: |504 bb.0:505 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_Use1_with_8_salus_in_between506 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec507 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 0508 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 1509 ; GFX1250-NEXT: $sgpr2 = S_MOV_B32 2510 ; GFX1250-NEXT: $sgpr3 = S_MOV_B32 3511 ; GFX1250-NEXT: $sgpr4 = S_MOV_B32 4512 ; GFX1250-NEXT: $sgpr5 = S_MOV_B32 5513 ; GFX1250-NEXT: $sgpr6 = S_MOV_B32 6514 ; GFX1250-NEXT: $sgpr7 = S_MOV_B32 7515 ; GFX1250-NEXT: V_NOP_e32 implicit $exec516 ; GFX1250-NEXT: V_NOP_e32 implicit $exec517 ; GFX1250-NEXT: V_NOP_e32 implicit $exec518 ; GFX1250-NEXT: V_NOP_e32 implicit $exec519 ; GFX1250-NEXT: V_NOP_e32 implicit $exec520 ; GFX1250-NEXT: V_NOP_e32 implicit $exec521 ; GFX1250-NEXT: V_NOP_e32 implicit $exec522 ; GFX1250-NEXT: V_NOP_e32 implicit $exec523 ; GFX1250-NEXT: $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec524 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec525 $sgpr0 = S_MOV_B32 0526 $sgpr1 = S_MOV_B32 1527 $sgpr2 = S_MOV_B32 2528 $sgpr3 = S_MOV_B32 3529 $sgpr4 = S_MOV_B32 4530 $sgpr5 = S_MOV_B32 5531 $sgpr6 = S_MOV_B32 6532 $sgpr7 = S_MOV_B32 7533 $vgpr48 = V_ADD_F32_e32 $vgpr32, $vgpr47, implicit $mode, implicit $exec534...535 536---537name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_D1538body: |539 bb.0:540 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_F8_D0_overlaps_D1541 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec542 ; GFX1250-NEXT: V_NOP_e32 implicit $exec543 ; GFX1250-NEXT: V_NOP_e32 implicit $exec544 ; GFX1250-NEXT: V_NOP_e32 implicit $exec545 ; GFX1250-NEXT: V_NOP_e32 implicit $exec546 ; GFX1250-NEXT: V_NOP_e32 implicit $exec547 ; GFX1250-NEXT: V_NOP_e32 implicit $exec548 ; GFX1250-NEXT: V_NOP_e32 implicit $exec549 ; GFX1250-NEXT: V_NOP_e32 implicit $exec550 ; GFX1250-NEXT: $vgpr32 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec551 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec552 $vgpr32 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec553...554 555---556name: test_wmma_F32_16x16x128_F8F6F4_F8_A0_overlaps_D1557body: |558 bb.0:559 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_F8_A0_overlaps_D1560 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec561 ; GFX1250-NEXT: V_NOP_e32 implicit $exec562 ; GFX1250-NEXT: V_NOP_e32 implicit $exec563 ; GFX1250-NEXT: V_NOP_e32 implicit $exec564 ; GFX1250-NEXT: V_NOP_e32 implicit $exec565 ; GFX1250-NEXT: V_NOP_e32 implicit $exec566 ; GFX1250-NEXT: V_NOP_e32 implicit $exec567 ; GFX1250-NEXT: V_NOP_e32 implicit $exec568 ; GFX1250-NEXT: V_NOP_e32 implicit $exec569 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec570 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec571 $vgpr0 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec572...573 574---575name: test_wmma_F32_16x16x128_F8F6F4_F8_B0_overlaps_D1576body: |577 bb.0:578 ; GFX1250-LABEL: name: test_wmma_F32_16x16x128_F8F6F4_F8_B0_overlaps_D1579 ; GFX1250: early-clobber $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec580 ; GFX1250-NEXT: V_NOP_e32 implicit $exec581 ; GFX1250-NEXT: V_NOP_e32 implicit $exec582 ; GFX1250-NEXT: V_NOP_e32 implicit $exec583 ; GFX1250-NEXT: V_NOP_e32 implicit $exec584 ; GFX1250-NEXT: V_NOP_e32 implicit $exec585 ; GFX1250-NEXT: V_NOP_e32 implicit $exec586 ; GFX1250-NEXT: V_NOP_e32 implicit $exec587 ; GFX1250-NEXT: V_NOP_e32 implicit $exec588 ; GFX1250-NEXT: $vgpr16 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec589 $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39 = V_WMMA_F32_16X16X128_F8F6F4_f8_f8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, killed $vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23_vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, 8, killed $vgpr32_vgpr33_vgpr34_vgpr35_vgpr36_vgpr37_vgpr38_vgpr39, 0, 0, 0, 0, implicit $exec590 $vgpr16 = V_ADD_F32_e32 $vgpr47, $vgpr48, implicit $mode, implicit $exec591...592 593---594name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_Use1595body: |596 bb.0:597 ; GFX1250-LABEL: name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_Use1598 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec599 ; GFX1250-NEXT: V_NOP_e32 implicit $exec600 ; GFX1250-NEXT: V_NOP_e32 implicit $exec601 ; GFX1250-NEXT: $vgpr34 = V_ADD_F32_e32 $vgpr24, $vgpr33, implicit $mode, implicit $exec602 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec603 $vgpr34 = V_ADD_F32_e32 $vgpr24, $vgpr33, implicit $mode, implicit $exec604...605 606---607name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_Use1_with_2_valus_in_between608body: |609 bb.0:610 ; GFX1250-LABEL: name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_Use1_with_2_valus_in_between611 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec612 ; GFX1250-NEXT: $vgpr40 = V_MOV_B32_e32 40, implicit $exec613 ; GFX1250-NEXT: $vgpr41 = V_MOV_B32_e32 41, implicit $exec614 ; GFX1250-NEXT: $vgpr34 = V_ADD_F32_e32 $vgpr24, $vgpr33, implicit $mode, implicit $exec615 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec616 $vgpr40 = V_MOV_B32_e32 40, implicit $exec617 $vgpr41 = V_MOV_B32_e32 41, implicit $exec618 $vgpr34 = V_ADD_F32_e32 $vgpr24, $vgpr33, implicit $mode, implicit $exec619...620 621---622name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_Use1_with_2_salus_in_between623body: |624 bb.0:625 ; GFX1250-LABEL: name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_Use1_with_2_salus_in_between626 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec627 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 0628 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 1629 ; GFX1250-NEXT: V_NOP_e32 implicit $exec630 ; GFX1250-NEXT: V_NOP_e32 implicit $exec631 ; GFX1250-NEXT: $vgpr34 = V_ADD_F32_e32 $vgpr24, $vgpr33, implicit $mode, implicit $exec632 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec633 $sgpr0 = S_MOV_B32 0634 $sgpr1 = S_MOV_B32 1635 $vgpr34 = V_ADD_F32_e32 $vgpr24, $vgpr33, implicit $mode, implicit $exec636...637 638---639name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_D1640body: |641 bb.0:642 ; GFX1250-LABEL: name: test_swmmac_f32_16x16x64_bf16_D0_overlaps_D1643 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec644 ; GFX1250-NEXT: V_NOP_e32 implicit $exec645 ; GFX1250-NEXT: V_NOP_e32 implicit $exec646 ; GFX1250-NEXT: $vgpr24 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec647 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec648 $vgpr24 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec649...650 651---652name: test_swmmac_f32_16x16x64_bf16_A0_overlaps_D1653body: |654 bb.0:655 ; GFX1250-LABEL: name: test_swmmac_f32_16x16x64_bf16_A0_overlaps_D1656 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec657 ; GFX1250-NEXT: V_NOP_e32 implicit $exec658 ; GFX1250-NEXT: V_NOP_e32 implicit $exec659 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec660 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec661 $vgpr0 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec662...663 664---665name: test_swmmac_f32_16x16x64_bf16_B0_overlaps_D1666body: |667 bb.0:668 ; GFX1250-LABEL: name: test_swmmac_f32_16x16x64_bf16_B0_overlaps_D1669 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec670 ; GFX1250-NEXT: V_NOP_e32 implicit $exec671 ; GFX1250-NEXT: V_NOP_e32 implicit $exec672 ; GFX1250-NEXT: $vgpr8 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec673 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec674 $vgpr8 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec675...676 677---678name: test_swmmac_f32_16x16x64_bf16_Index0_overlaps_D1679body: |680 bb.0:681 ; GFX1250-LABEL: name: test_swmmac_f32_16x16x64_bf16_Index0_overlaps_D1682 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec683 ; GFX1250-NEXT: V_NOP_e32 implicit $exec684 ; GFX1250-NEXT: V_NOP_e32 implicit $exec685 ; GFX1250-NEXT: $vgpr32 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec686 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32, 0, 0, 0, 0, 0, implicit $exec687 $vgpr32 = V_ADD_F32_e32 $vgpr33, $vgpr34, implicit $mode, implicit $exec688...689 690---691name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_Use1692body: |693 bb.0:694 ; GFX1250-LABEL: name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_Use1695 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec696 ; GFX1250-NEXT: V_NOP_e32 implicit $exec697 ; GFX1250-NEXT: V_NOP_e32 implicit $exec698 ; GFX1250-NEXT: $vgpr31 = V_ADD_F32_e32 $vgpr24, $vgpr30, implicit $mode, implicit $exec699 $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec700 $vgpr31 = V_ADD_F32_e32 $vgpr24, $vgpr30, implicit $mode, implicit $exec701...702 703---704name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_Use1_with_2_valus_in_between705body: |706 bb.0:707 ; GFX1250-LABEL: name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_Use1_with_2_valus_in_between708 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec709 ; GFX1250-NEXT: $vgpr40 = V_MOV_B32_e32 40, implicit $exec710 ; GFX1250-NEXT: $vgpr41 = V_MOV_B32_e32 41, implicit $exec711 ; GFX1250-NEXT: $vgpr31 = V_ADD_F32_e32 $vgpr24, $vgpr30, implicit $mode, implicit $exec712 $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec713 $vgpr40 = V_MOV_B32_e32 40, implicit $exec714 $vgpr41 = V_MOV_B32_e32 41, implicit $exec715 $vgpr31 = V_ADD_F32_e32 $vgpr24, $vgpr30, implicit $mode, implicit $exec716...717 718---719name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_Use1_with_2_salus_in_between720body: |721 bb.0:722 ; GFX1250-LABEL: name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_Use1_with_2_salus_in_between723 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec724 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 0725 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 1726 ; GFX1250-NEXT: V_NOP_e32 implicit $exec727 ; GFX1250-NEXT: V_NOP_e32 implicit $exec728 ; GFX1250-NEXT: $vgpr31 = V_ADD_F32_e32 $vgpr24, $vgpr30, implicit $mode, implicit $exec729 $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec730 $sgpr0 = S_MOV_B32 0731 $sgpr1 = S_MOV_B32 1732 $vgpr31 = V_ADD_F32_e32 $vgpr24, $vgpr30, implicit $mode, implicit $exec733...734 735---736name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_D1737body: |738 bb.0:739 ; GFX1250-LABEL: name: test_swmmac_f16_16x16x128_bf8_fp8_D0_overlaps_D1740 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec741 ; GFX1250-NEXT: V_NOP_e32 implicit $exec742 ; GFX1250-NEXT: V_NOP_e32 implicit $exec743 ; GFX1250-NEXT: $vgpr24 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec744 $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec745 $vgpr24 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec746...747 748---749name: test_swmmac_f16_16x16x128_bf8_fp8_A0_overlaps_D1750body: |751 bb.0:752 ; GFX1250-LABEL: name: test_swmmac_f16_16x16x128_bf8_fp8_A0_overlaps_D1753 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec754 ; GFX1250-NEXT: V_NOP_e32 implicit $exec755 ; GFX1250-NEXT: V_NOP_e32 implicit $exec756 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec757 $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec758 $vgpr0 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec759...760 761---762name: test_swmmac_f16_16x16x128_bf8_fp8_B0_overlaps_D1763body: |764 bb.0:765 ; GFX1250-LABEL: name: test_swmmac_f16_16x16x128_bf8_fp8_B0_overlaps_D1766 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec767 ; GFX1250-NEXT: V_NOP_e32 implicit $exec768 ; GFX1250-NEXT: V_NOP_e32 implicit $exec769 ; GFX1250-NEXT: $vgpr8 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec770 $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec771 $vgpr8 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec772...773 774---775name: test_swmmac_f16_16x16x128_bf8_fp8_Index0_overlaps_D1776body: |777 bb.0:778 ; GFX1250-LABEL: name: test_swmmac_f16_16x16x128_bf8_fp8_Index0_overlaps_D1779 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec780 ; GFX1250-NEXT: V_NOP_e32 implicit $exec781 ; GFX1250-NEXT: V_NOP_e32 implicit $exec782 ; GFX1250-NEXT: $vgpr28 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec783 $vgpr24_vgpr25_vgpr26_vgpr27 = V_SWMMAC_F16_16X16X128_BF8_FP8_w32_twoaddr killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27, killed $vgpr28_vgpr29, 0, 0, 0, implicit $exec784 $vgpr28 = V_ADD_F32_e32 $vgpr30, $vgpr31, implicit $mode, implicit $exec785...786 787---788name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_Use1789body: |790 bb.0:791 ; GFX1250-LABEL: name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_Use1792 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec793 ; GFX1250-NEXT: V_NOP_e32 implicit $exec794 ; GFX1250-NEXT: V_NOP_e32 implicit $exec795 ; GFX1250-NEXT: V_NOP_e32 implicit $exec796 ; GFX1250-NEXT: V_NOP_e32 implicit $exec797 ; GFX1250-NEXT: $vgpr34 = V_ADD_F32_e32 $vgpr33, $vgpr24, implicit $mode, implicit $exec798 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec799 $vgpr34 = V_ADD_F32_e32 $vgpr33, $vgpr24, implicit $mode, implicit $exec800...801 802---803name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_Use1_with_4_valus_in_between804body: |805 bb.0:806 ; GFX1250-LABEL: name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_Use1_with_4_valus_in_between807 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec808 ; GFX1250-NEXT: $vgpr40 = V_MOV_B32_e32 40, implicit $exec809 ; GFX1250-NEXT: $vgpr41 = V_MOV_B32_e32 41, implicit $exec810 ; GFX1250-NEXT: $vgpr42 = V_MOV_B32_e32 42, implicit $exec811 ; GFX1250-NEXT: $vgpr43 = V_MOV_B32_e32 43, implicit $exec812 ; GFX1250-NEXT: $vgpr34 = V_ADD_F32_e32 $vgpr34, $vgpr24, implicit $mode, implicit $exec813 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec814 $vgpr40 = V_MOV_B32_e32 40, implicit $exec815 $vgpr41 = V_MOV_B32_e32 41, implicit $exec816 $vgpr42 = V_MOV_B32_e32 42, implicit $exec817 $vgpr43 = V_MOV_B32_e32 43, implicit $exec818 $vgpr34 = V_ADD_F32_e32 $vgpr34, $vgpr24, implicit $mode, implicit $exec819...820 821---822name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_Use1_with_4_salus_in_between823body: |824 bb.0:825 ; GFX1250-LABEL: name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_Use1_with_4_salus_in_between826 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec827 ; GFX1250-NEXT: $sgpr0 = S_MOV_B32 0828 ; GFX1250-NEXT: $sgpr1 = S_MOV_B32 1829 ; GFX1250-NEXT: $sgpr2 = S_MOV_B32 2830 ; GFX1250-NEXT: $sgpr3 = S_MOV_B32 4831 ; GFX1250-NEXT: V_NOP_e32 implicit $exec832 ; GFX1250-NEXT: V_NOP_e32 implicit $exec833 ; GFX1250-NEXT: V_NOP_e32 implicit $exec834 ; GFX1250-NEXT: V_NOP_e32 implicit $exec835 ; GFX1250-NEXT: $vgpr34 = V_ADD_F32_e32 $vgpr34, $vgpr24, implicit $mode, implicit $exec836 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec837 $sgpr0 = S_MOV_B32 0838 $sgpr1 = S_MOV_B32 1839 $sgpr2 = S_MOV_B32 2840 $sgpr3 = S_MOV_B32 4841 $vgpr34 = V_ADD_F32_e32 $vgpr34, $vgpr24, implicit $mode, implicit $exec842...843 844---845name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_D1846body: |847 bb.0:848 ; GFX1250-LABEL: name: test_swmmac_i32_16x16x128_iu8_D0_overlaps_D1849 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec850 ; GFX1250-NEXT: V_NOP_e32 implicit $exec851 ; GFX1250-NEXT: V_NOP_e32 implicit $exec852 ; GFX1250-NEXT: V_NOP_e32 implicit $exec853 ; GFX1250-NEXT: V_NOP_e32 implicit $exec854 ; GFX1250-NEXT: $vgpr24 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec855 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec856 $vgpr24 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec857...858 859---860name: test_swmmac_i32_16x16x128_iu8_A0_overlaps_D1861body: |862 bb.0:863 ; GFX1250-LABEL: name: test_swmmac_i32_16x16x128_iu8_A0_overlaps_D1864 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec865 ; GFX1250-NEXT: V_NOP_e32 implicit $exec866 ; GFX1250-NEXT: V_NOP_e32 implicit $exec867 ; GFX1250-NEXT: V_NOP_e32 implicit $exec868 ; GFX1250-NEXT: V_NOP_e32 implicit $exec869 ; GFX1250-NEXT: $vgpr0 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec870 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec871 $vgpr0 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec872...873 874---875name: test_swmmac_i32_16x16x128_iu8_B0_overlaps_D1876body: |877 bb.0:878 ; GFX1250-LABEL: name: test_swmmac_i32_16x16x128_iu8_B0_overlaps_D1879 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec880 ; GFX1250-NEXT: V_NOP_e32 implicit $exec881 ; GFX1250-NEXT: V_NOP_e32 implicit $exec882 ; GFX1250-NEXT: V_NOP_e32 implicit $exec883 ; GFX1250-NEXT: V_NOP_e32 implicit $exec884 ; GFX1250-NEXT: $vgpr8 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec885 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec886 $vgpr8 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec887...888 889---890name: test_swmmac_i32_16x16x128_iu8_Index0_overlaps_D1891body: |892 bb.0:893 ; GFX1250-LABEL: name: test_swmmac_i32_16x16x128_iu8_Index0_overlaps_D1894 ; GFX1250: early-clobber $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec895 ; GFX1250-NEXT: V_NOP_e32 implicit $exec896 ; GFX1250-NEXT: V_NOP_e32 implicit $exec897 ; GFX1250-NEXT: V_NOP_e32 implicit $exec898 ; GFX1250-NEXT: V_NOP_e32 implicit $exec899 ; GFX1250-NEXT: $vgpr32 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec900 $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31 = V_SWMMAC_I32_16X16X128_IU8_w32_twoaddr 8, killed $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7, 8, killed $vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19_vgpr20_vgpr21_vgpr22_vgpr23, killed $vgpr24_vgpr25_vgpr26_vgpr27_vgpr28_vgpr29_vgpr30_vgpr31, killed $vgpr32_vgpr33, 0, 0, 0, 0, implicit $exec901 $vgpr32 = V_ADD_F32_e32 $vgpr34, $vgpr35, implicit $mode, implicit $exec902...903