llvm.amdgcn.raw.tbuffer.store.f16.ll 39 KB
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492
; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=UNPACKED %s
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx810 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=PACKED %s

define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; UNPACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; UNPACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; PACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; PACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
  ret void
}

define amdgpu_ps void @raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; UNPACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; UNPACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
  ; UNPACKED:   [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
  ; UNPACKED:   [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY]], implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 4 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; PACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; PACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 4 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
  ret void
}

; FIXME: Crashes
; define amdgpu_ps void @raw_tbuffer_store_v3f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
;   call void @llvm.amdgcn.raw.tbuffer.store.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
;   ret void
; }

define amdgpu_ps void @raw_tbuffer_store_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; UNPACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; UNPACKED:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; UNPACKED:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; UNPACKED:   [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
  ; UNPACKED:   [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
  ; UNPACKED:   [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
  ; UNPACKED:   [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY]], implicit $exec
  ; UNPACKED:   [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
  ; UNPACKED:   [[V_LSHRREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY9]], [[COPY1]], implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1, [[COPY1]], %subreg.sub2, [[V_LSHRREV_B32_e64_1]], %subreg.sub3
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 8 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; PACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; PACKED:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; PACKED:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; PACKED:   [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
  ; PACKED:   [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 8 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
  ret void
}

; Waterfall for rsrc
define amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   successors: %bb.2(0x80000000)
  ; UNPACKED:   liveins: $sgpr2, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; UNPACKED:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; UNPACKED:   [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr4
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr5
  ; UNPACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   [[COPY7:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
  ; UNPACKED:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
  ; UNPACKED:   [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
  ; UNPACKED: bb.2:
  ; UNPACKED:   successors: %bb.3(0x40000000), %bb.2(0x40000000)
  ; UNPACKED:   [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub0, implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub1, implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE1:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1
  ; UNPACKED:   [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE1]], [[COPY7]], implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub0, implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub1, implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE2:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_2]], %subreg.sub0, [[V_READFIRSTLANE_B32_3]], %subreg.sub1
  ; UNPACKED:   [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE2]], [[COPY8]], implicit $exec
  ; UNPACKED:   [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_1]], [[V_CMP_EQ_U64_e64_]], implicit-def $scc
  ; UNPACKED:   [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE3]], [[COPY6]], 0, 94, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
  ; UNPACKED:   $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
  ; UNPACKED:   S_CBRANCH_EXECNZ %bb.2, implicit $exec
  ; UNPACKED: bb.3:
  ; UNPACKED:   successors: %bb.4(0x80000000)
  ; UNPACKED:   $exec = S_MOV_B64_term [[S_MOV_B64_term]]
  ; UNPACKED: bb.4:
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   successors: %bb.2(0x80000000)
  ; PACKED:   liveins: $sgpr2, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; PACKED:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; PACKED:   [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr4
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr5
  ; PACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   [[COPY7:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
  ; PACKED:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
  ; PACKED:   [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
  ; PACKED: bb.2:
  ; PACKED:   successors: %bb.3(0x40000000), %bb.2(0x40000000)
  ; PACKED:   [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub0, implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub1, implicit $exec
  ; PACKED:   [[REG_SEQUENCE1:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1
  ; PACKED:   [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE1]], [[COPY7]], implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub0, implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub1, implicit $exec
  ; PACKED:   [[REG_SEQUENCE2:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_2]], %subreg.sub0, [[V_READFIRSTLANE_B32_3]], %subreg.sub1
  ; PACKED:   [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE2]], [[COPY8]], implicit $exec
  ; PACKED:   [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_1]], [[V_CMP_EQ_U64_e64_]], implicit-def $scc
  ; PACKED:   [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE3]], [[COPY6]], 0, 94, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
  ; PACKED:   $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
  ; PACKED:   S_CBRANCH_EXECNZ %bb.2, implicit $exec
  ; PACKED: bb.3:
  ; PACKED:   successors: %bb.4(0x80000000)
  ; PACKED:   $exec = S_MOV_B64_term [[S_MOV_B64_term]]
  ; PACKED: bb.4:
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
  ret void
}

; Waterfall for rsrc and soffset
define amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__vgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__vgpr_soffset
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   successors: %bb.2(0x80000000)
  ; UNPACKED:   liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; UNPACKED:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; UNPACKED:   [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr4
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr5
  ; UNPACKED:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   [[COPY7:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
  ; UNPACKED:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
  ; UNPACKED:   [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
  ; UNPACKED: bb.2:
  ; UNPACKED:   successors: %bb.3(0x40000000), %bb.2(0x40000000)
  ; UNPACKED:   [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub0, implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub1, implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE1:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1
  ; UNPACKED:   [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE1]], [[COPY7]], implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub0, implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub1, implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE2:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_2]], %subreg.sub0, [[V_READFIRSTLANE_B32_3]], %subreg.sub1
  ; UNPACKED:   [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE2]], [[COPY8]], implicit $exec
  ; UNPACKED:   [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_1]], [[V_CMP_EQ_U64_e64_]], implicit-def $scc
  ; UNPACKED:   [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
  ; UNPACKED:   [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
  ; UNPACKED:   [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
  ; UNPACKED:   [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U32_e64_]], [[S_AND_B64_]], implicit-def $scc
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE3]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
  ; UNPACKED:   $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
  ; UNPACKED:   S_CBRANCH_EXECNZ %bb.2, implicit $exec
  ; UNPACKED: bb.3:
  ; UNPACKED:   successors: %bb.4(0x80000000)
  ; UNPACKED:   $exec = S_MOV_B64_term [[S_MOV_B64_term]]
  ; UNPACKED: bb.4:
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__vgpr_soffset
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   successors: %bb.2(0x80000000)
  ; PACKED:   liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; PACKED:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; PACKED:   [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr4
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr5
  ; PACKED:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   [[COPY7:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
  ; PACKED:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
  ; PACKED:   [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
  ; PACKED: bb.2:
  ; PACKED:   successors: %bb.3(0x40000000), %bb.2(0x40000000)
  ; PACKED:   [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub0, implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY7]].sub1, implicit $exec
  ; PACKED:   [[REG_SEQUENCE1:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1
  ; PACKED:   [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE1]], [[COPY7]], implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub0, implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub1, implicit $exec
  ; PACKED:   [[REG_SEQUENCE2:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_2]], %subreg.sub0, [[V_READFIRSTLANE_B32_3]], %subreg.sub1
  ; PACKED:   [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE2]], [[COPY8]], implicit $exec
  ; PACKED:   [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_1]], [[V_CMP_EQ_U64_e64_]], implicit-def $scc
  ; PACKED:   [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
  ; PACKED:   [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
  ; PACKED:   [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
  ; PACKED:   [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U32_e64_]], [[S_AND_B64_]], implicit-def $scc
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE3]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
  ; PACKED:   $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
  ; PACKED:   S_CBRANCH_EXECNZ %bb.2, implicit $exec
  ; PACKED: bb.3:
  ; PACKED:   successors: %bb.4(0x80000000)
  ; PACKED:   $exec = S_MOV_B64_term [[S_MOV_B64_term]]
  ; PACKED: bb.4:
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
  ret void
}

; Waterfall for rsrc and soffset, copy for voffset
define amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset(half %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   successors: %bb.2(0x80000000)
  ; UNPACKED:   liveins: $sgpr2, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; UNPACKED:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; UNPACKED:   [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr4
  ; UNPACKED:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr5
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY5]]
  ; UNPACKED:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
  ; UNPACKED:   [[COPY9:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
  ; UNPACKED:   [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
  ; UNPACKED: bb.2:
  ; UNPACKED:   successors: %bb.3(0x40000000), %bb.2(0x40000000)
  ; UNPACKED:   [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub0, implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub1, implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE1:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1
  ; UNPACKED:   [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE1]], [[COPY8]], implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY9]].sub0, implicit $exec
  ; UNPACKED:   [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY9]].sub1, implicit $exec
  ; UNPACKED:   [[REG_SEQUENCE2:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_2]], %subreg.sub0, [[V_READFIRSTLANE_B32_3]], %subreg.sub1
  ; UNPACKED:   [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE2]], [[COPY9]], implicit $exec
  ; UNPACKED:   [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_1]], [[V_CMP_EQ_U64_e64_]], implicit-def $scc
  ; UNPACKED:   [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
  ; UNPACKED:   [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
  ; UNPACKED:   [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
  ; UNPACKED:   [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U32_e64_]], [[S_AND_B64_]], implicit-def $scc
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE3]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
  ; UNPACKED:   $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
  ; UNPACKED:   S_CBRANCH_EXECNZ %bb.2, implicit $exec
  ; UNPACKED: bb.3:
  ; UNPACKED:   successors: %bb.4(0x80000000)
  ; UNPACKED:   $exec = S_MOV_B64_term [[S_MOV_B64_term]]
  ; UNPACKED: bb.4:
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   successors: %bb.2(0x80000000)
  ; PACKED:   liveins: $sgpr2, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; PACKED:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; PACKED:   [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr4
  ; PACKED:   [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr5
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY5]]
  ; PACKED:   [[COPY8:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub0_sub1
  ; PACKED:   [[COPY9:%[0-9]+]]:vreg_64 = COPY [[REG_SEQUENCE]].sub2_sub3
  ; PACKED:   [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
  ; PACKED: bb.2:
  ; PACKED:   successors: %bb.3(0x40000000), %bb.2(0x40000000)
  ; PACKED:   [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub0, implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY8]].sub1, implicit $exec
  ; PACKED:   [[REG_SEQUENCE1:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1
  ; PACKED:   [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE1]], [[COPY8]], implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_2:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY9]].sub0, implicit $exec
  ; PACKED:   [[V_READFIRSTLANE_B32_3:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY9]].sub1, implicit $exec
  ; PACKED:   [[REG_SEQUENCE2:%[0-9]+]]:sreg_64_xexec = REG_SEQUENCE [[V_READFIRSTLANE_B32_2]], %subreg.sub0, [[V_READFIRSTLANE_B32_3]], %subreg.sub1
  ; PACKED:   [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[REG_SEQUENCE2]], [[COPY9]], implicit $exec
  ; PACKED:   [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_1]], [[V_CMP_EQ_U64_e64_]], implicit-def $scc
  ; PACKED:   [[REG_SEQUENCE3:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[V_READFIRSTLANE_B32_]], %subreg.sub0, [[V_READFIRSTLANE_B32_1]], %subreg.sub1, [[V_READFIRSTLANE_B32_2]], %subreg.sub2, [[V_READFIRSTLANE_B32_3]], %subreg.sub3
  ; PACKED:   [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32_xm0 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
  ; PACKED:   [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
  ; PACKED:   [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U32_e64_]], [[S_AND_B64_]], implicit-def $scc
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE3]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
  ; PACKED:   $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
  ; PACKED:   S_CBRANCH_EXECNZ %bb.2, implicit $exec
  ; PACKED: bb.3:
  ; PACKED:   successors: %bb.4(0x80000000)
  ; PACKED:   $exec = S_MOV_B64_term [[S_MOV_B64_term]]
  ; PACKED: bb.4:
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
  ret void
}

define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; UNPACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; UNPACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; PACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; PACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1)
  ret void
}

define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; UNPACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; UNPACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 1, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; PACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; PACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 1, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2)
  ret void
}

define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; UNPACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; UNPACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 1, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; PACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; PACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 1, 0, 0, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3)
  ret void
}

define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
  ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
  ; UNPACKED: bb.1 (%ir-block.0):
  ; UNPACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; UNPACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; UNPACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; UNPACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; UNPACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; UNPACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; UNPACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; UNPACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; UNPACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; UNPACKED:   TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, 0, 1, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; UNPACKED:   S_ENDPGM 0
  ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
  ; PACKED: bb.1 (%ir-block.0):
  ; PACKED:   liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
  ; PACKED:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; PACKED:   [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
  ; PACKED:   [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
  ; PACKED:   [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
  ; PACKED:   [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
  ; PACKED:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; PACKED:   [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
  ; PACKED:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
  ; PACKED:   TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, 0, 1, 0, implicit $exec :: (dereferenceable store 2 into custom "TargetCustom7", align 1, addrspace 4)
  ; PACKED:   S_ENDPGM 0
  call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4)
  ret void
}

declare void @llvm.amdgcn.raw.tbuffer.store.f16(half, <4 x i32>, i32, i32, i32 immarg, i32 immarg)
declare void @llvm.amdgcn.raw.tbuffer.store.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg)
declare void @llvm.amdgcn.raw.tbuffer.store.v3f16(<3 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg)
declare void @llvm.amdgcn.raw.tbuffer.store.v4f16(<4 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg)