# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -mtriple=amdgcn -mcpu=tahiti -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -disable-gisel-legality-check -o - %s | FileCheck -check-prefix=GFX6 %s
# RUN: llc -mtriple=amdgcn -mcpu=hawaii -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX7 %s
# RUN: llc -mtriple=amdgcn -mcpu=fiji -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX7 %s
# RUN: llc -mtriple=amdgcn -mcpu=gfx900 -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX9 %s
# RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -mattr=+cumode -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX10 %s
# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX9 %s
---
name: load_local_s32_from_4
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s32_from_4
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s32), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX7-LABEL: name: load_local_s32_from_4
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s32), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX9-LABEL: name: load_local_s32_from_4
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s32), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
; GFX10-LABEL: name: load_local_s32_from_4
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s32), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_LOAD %0 :: (load (s32), align 4, addrspace 3)
$vgpr0 = COPY %1
...
---
name: load_local_s32_from_2
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s32_from_2
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_U16_:%[0-9]+]]:vgpr_32 = DS_READ_U16 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s16), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_U16_]]
; GFX7-LABEL: name: load_local_s32_from_2
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_U16_:%[0-9]+]]:vgpr_32 = DS_READ_U16 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s16), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_U16_]]
; GFX9-LABEL: name: load_local_s32_from_2
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_U16_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U16_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s16), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_U16_gfx9_]]
; GFX10-LABEL: name: load_local_s32_from_2
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_U16_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U16_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s16), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_U16_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_LOAD %0 :: (load (s16), align 2, addrspace 3)
$vgpr0 = COPY %1
...
---
name: load_local_s32_from_1
legalized: true
regBankSelected: true
tracksRegLiveness: true
machineFunctionInfo:
scratchRSrcReg: $sgpr0_sgpr1_sgpr2_sgpr3
stackPtrOffsetReg: $sgpr32
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s32_from_1
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX7-LABEL: name: load_local_s32_from_1
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX9-LABEL: name: load_local_s32_from_1
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
; GFX10-LABEL: name: load_local_s32_from_1
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_LOAD %0 :: (load (s8), align 1, addrspace 3)
$vgpr0 = COPY %1
...
---
name: load_local_v2s32
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_v2s32
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (<2 x s32>), addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX7-LABEL: name: load_local_v2s32
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (<2 x s32>), addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX9-LABEL: name: load_local_v2s32
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (<2 x s32>), addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
; GFX10-LABEL: name: load_local_v2s32
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (<2 x s32>), addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(<2 x s32>) = G_LOAD %0 :: (load (<2 x s32>), align 8, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_v2s32_align4
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_v2s32_align4
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[LOAD:%[0-9]+]]:vreg_64(<2 x s32>) = G_LOAD [[COPY]](p3) :: (load (<2 x s32>), align 4, addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x s32>)
; GFX7-LABEL: name: load_local_v2s32_align4
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ2_B32_:%[0-9]+]]:vreg_64 = DS_READ2_B32 [[COPY]], 0, 1, 0, implicit $m0, implicit $exec :: (load (<2 x s32>), align 4, addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_]]
; GFX9-LABEL: name: load_local_v2s32_align4
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 0, 1, 0, implicit $exec :: (load (<2 x s32>), align 4, addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
; GFX10-LABEL: name: load_local_v2s32_align4
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 0, 1, 0, implicit $exec :: (load (<2 x s32>), align 4, addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(<2 x s32>) = G_LOAD %0 :: (load (<2 x s32>), align 4, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_s64
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s64
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s64), addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX7-LABEL: name: load_local_s64
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (s64), addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX9-LABEL: name: load_local_s64
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s64), addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
; GFX10-LABEL: name: load_local_s64
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (s64), addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s64) = G_LOAD %0 :: (load (s64), align 8, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_s64_align4
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s64_align4
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[LOAD:%[0-9]+]]:vreg_64(s64) = G_LOAD [[COPY]](p3) :: (load (s64), align 4, addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](s64)
; GFX7-LABEL: name: load_local_s64_align4
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ2_B32_:%[0-9]+]]:vreg_64 = DS_READ2_B32 [[COPY]], 0, 1, 0, implicit $m0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_]]
; GFX9-LABEL: name: load_local_s64_align4
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 0, 1, 0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
; GFX10-LABEL: name: load_local_s64_align4
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 0, 1, 0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s64) = G_LOAD %0 :: (load (s64), align 4, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_p3_from_4
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_p3_from_4
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (p3), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX7-LABEL: name: load_local_p3_from_4
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (p3), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX9-LABEL: name: load_local_p3_from_4
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (p3), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
; GFX10-LABEL: name: load_local_p3_from_4
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (p3), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(p3) = G_LOAD %0 :: (load (p3), align 4, addrspace 3)
$vgpr0 = COPY %1
...
---
name: load_local_p5_from_4
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_p5_from_4
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (p5), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX7-LABEL: name: load_local_p5_from_4
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (p5), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX9-LABEL: name: load_local_p5_from_4
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (p5), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
; GFX10-LABEL: name: load_local_p5_from_4
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (p5), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(p5) = G_LOAD %0 :: (load (p5), align 4, addrspace 3)
$vgpr0 = COPY %1
...
---
name: load_local_p1_align8
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_p1_align8
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (p1), addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX7-LABEL: name: load_local_p1_align8
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (p1), addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX9-LABEL: name: load_local_p1_align8
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (p1), addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
; GFX10-LABEL: name: load_local_p1_align8
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (p1), addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(p1) = G_LOAD %0 :: (load (p1), align 8, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_p1_align4
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_p1_align4
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[LOAD:%[0-9]+]]:vreg_64(p1) = G_LOAD [[COPY]](p3) :: (load (p1), align 4, addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](p1)
; GFX7-LABEL: name: load_local_p1_align4
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ2_B32_:%[0-9]+]]:vreg_64 = DS_READ2_B32 [[COPY]], 0, 1, 0, implicit $m0, implicit $exec :: (load (p1), align 4, addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_]]
; GFX9-LABEL: name: load_local_p1_align4
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 0, 1, 0, implicit $exec :: (load (p1), align 4, addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
; GFX10-LABEL: name: load_local_p1_align4
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 0, 1, 0, implicit $exec :: (load (p1), align 4, addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(p1) = G_LOAD %0 :: (load (p1), align 4, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_p999_from_8
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_p999_from_8
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[LOAD:%[0-9]+]]:vreg_64(p999) = G_LOAD [[COPY]](p3) :: (load (p999), addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](p999)
; GFX7-LABEL: name: load_local_p999_from_8
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[LOAD:%[0-9]+]]:vreg_64(p999) = G_LOAD [[COPY]](p3) :: (load (p999), addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](p999)
; GFX9-LABEL: name: load_local_p999_from_8
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX9-NEXT: [[LOAD:%[0-9]+]]:vreg_64(p999) = G_LOAD [[COPY]](p3) :: (load (p999), addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](p999)
; GFX10-LABEL: name: load_local_p999_from_8
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX10-NEXT: [[LOAD:%[0-9]+]]:vreg_64(p999) = G_LOAD [[COPY]](p3) :: (load (p999), addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](p999)
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(p999) = G_LOAD %0 :: (load (p999), align 8, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_v2p3
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_v2p3
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[LOAD:%[0-9]+]]:vreg_64(<2 x p3>) = G_LOAD [[COPY]](p3) :: (load (<2 x p3>), addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x p3>)
; GFX7-LABEL: name: load_local_v2p3
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[LOAD:%[0-9]+]]:vreg_64(<2 x p3>) = G_LOAD [[COPY]](p3) :: (load (<2 x p3>), addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x p3>)
; GFX9-LABEL: name: load_local_v2p3
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX9-NEXT: [[LOAD:%[0-9]+]]:vreg_64(<2 x p3>) = G_LOAD [[COPY]](p3) :: (load (<2 x p3>), addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x p3>)
; GFX10-LABEL: name: load_local_v2p3
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX10-NEXT: [[LOAD:%[0-9]+]]:vreg_64(<2 x p3>) = G_LOAD [[COPY]](p3) :: (load (<2 x p3>), addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](<2 x p3>)
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(<2 x p3>) = G_LOAD %0 :: (load (<2 x p3>), align 8, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
---
name: load_local_v2s16
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_v2s16
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (<2 x s16>), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX7-LABEL: name: load_local_v2s16
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B32_:%[0-9]+]]:vgpr_32 = DS_READ_B32 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (<2 x s16>), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_B32_]]
; GFX9-LABEL: name: load_local_v2s16
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (<2 x s16>), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
; GFX10-LABEL: name: load_local_v2s16
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B32_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_B32_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (<2 x s16>), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(<2 x s16>) = G_LOAD %0 :: (load (<2 x s16>), align 4, addrspace 3)
$vgpr0 = COPY %1
...
---
name: load_local_v4s16
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_v4s16
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (<4 x s16>), addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX7-LABEL: name: load_local_v4s16
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_B64_:%[0-9]+]]:vreg_64 = DS_READ_B64 [[COPY]], 0, 0, implicit $m0, implicit $exec :: (load (<4 x s16>), addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_]]
; GFX9-LABEL: name: load_local_v4s16
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (<4 x s16>), addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
; GFX10-LABEL: name: load_local_v4s16
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_B64_gfx9_:%[0-9]+]]:vreg_64 = DS_READ_B64_gfx9 [[COPY]], 0, 0, implicit $exec :: (load (<4 x s16>), addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ_B64_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(<4 x s16>) = G_LOAD %0 :: (load (<4 x s16>), align 8, addrspace 3)
$vgpr0_vgpr1 = COPY %1
...
# ---
# name: load_local_v6s16
# legalized: true
# regBankSelected: true
# tracksRegLiveness: true
# machineFunctionInfo:
# scratchRSrcReg: $sgpr0_sgpr1_sgpr2_sgpr3
# stackPtrOffsetReg: $sgpr32
# body: |
# bb.0:
# liveins: $vgpr0
# %0:vgpr(p3) = COPY $vgpr0
# %1:vgpr(<6 x s16>) = G_LOAD %0 :: (load (<6 x s16>), align 4, addrspace 3)
# $vgpr0_vgpr1_vgpr2 = COPY %1
# ...
################################################################################
### Stress addressing modes
################################################################################
---
name: load_local_s32_from_1_gep_65535
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s32_from_1_gep_65535
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 65535, implicit $exec
; GFX6-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[V_ADD_CO_U32_e64_]], 0, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX7-LABEL: name: load_local_s32_from_1_gep_65535
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[COPY]], 65535, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX9-LABEL: name: load_local_s32_from_1_gep_65535
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[COPY]], 65535, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
; GFX10-LABEL: name: load_local_s32_from_1_gep_65535
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[COPY]], 65535, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_CONSTANT i32 65535
%2:vgpr(p3) = G_PTR_ADD %0, %1
%3:vgpr(s32) = G_LOAD %2 :: (load (s8), align 1, addrspace 3)
$vgpr0 = COPY %3
...
---
name: load_local_s32_from_1_gep_65535_known_bits_base_address
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s32_from_1_gep_65535_known_bits_base_address
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 2147483647, implicit $exec
; GFX6-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY]], [[V_MOV_B32_e32_]], implicit $exec
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[V_AND_B32_e64_]], 65535, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX7-LABEL: name: load_local_s32_from_1_gep_65535_known_bits_base_address
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 2147483647, implicit $exec
; GFX7-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY]], [[V_MOV_B32_e32_]], implicit $exec
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[V_AND_B32_e64_]], 65535, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX9-LABEL: name: load_local_s32_from_1_gep_65535_known_bits_base_address
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 2147483647, implicit $exec
; GFX9-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY]], [[V_MOV_B32_e32_]], implicit $exec
; GFX9-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[V_AND_B32_e64_]], 65535, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
; GFX10-LABEL: name: load_local_s32_from_1_gep_65535_known_bits_base_address
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 2147483647, implicit $exec
; GFX10-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY]], [[V_MOV_B32_e32_]], implicit $exec
; GFX10-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[V_AND_B32_e64_]], 65535, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
%0:vgpr(s32) = COPY $vgpr0
%1:vgpr(s32) = G_CONSTANT i32 2147483647
%2:vgpr(s32) = G_AND %0, %1
%3:vgpr(p3) = G_INTTOPTR %2
%4:vgpr(s32) = G_CONSTANT i32 65535
%5:vgpr(p3) = G_PTR_ADD %3, %4
%6:vgpr(s32) = G_LOAD %5 :: (load (s8), align 1, addrspace 3)
$vgpr0 = COPY %6
...
---
name: load_local_s32_from_1_gep_65536
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s32_from_1_gep_65536
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 65536, implicit $exec
; GFX6-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[V_ADD_CO_U32_e64_]], 0, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX7-LABEL: name: load_local_s32_from_1_gep_65536
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 65536, implicit $exec
; GFX7-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[V_ADD_CO_U32_e64_]], 0, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX9-LABEL: name: load_local_s32_from_1_gep_65536
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 65536, implicit $exec
; GFX9-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX9-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[V_ADD_U32_e64_]], 0, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
; GFX10-LABEL: name: load_local_s32_from_1_gep_65536
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 65536, implicit $exec
; GFX10-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX10-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[V_ADD_U32_e64_]], 0, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_CONSTANT i32 65536
%2:vgpr(p3) = G_PTR_ADD %0, %1
%3:vgpr(s32) = G_LOAD %2 :: (load (s8), align 1, addrspace 3)
$vgpr0 = COPY %3
...
---
name: load_local_s32_from_1_gep_m1
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0
; GFX6-LABEL: name: load_local_s32_from_1_gep_m1
; GFX6: liveins: $vgpr0
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX6-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec
; GFX6-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[V_ADD_CO_U32_e64_]], 0, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX6-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX7-LABEL: name: load_local_s32_from_1_gep_m1
; GFX7: liveins: $vgpr0
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec
; GFX7-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ_U8_:%[0-9]+]]:vgpr_32 = DS_READ_U8 [[V_ADD_CO_U32_e64_]], 0, 0, implicit $m0, implicit $exec :: (load (s8), addrspace 3)
; GFX7-NEXT: $vgpr0 = COPY [[DS_READ_U8_]]
; GFX9-LABEL: name: load_local_s32_from_1_gep_m1
; GFX9: liveins: $vgpr0
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec
; GFX9-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX9-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[V_ADD_U32_e64_]], 0, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX9-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
; GFX10-LABEL: name: load_local_s32_from_1_gep_m1
; GFX10: liveins: $vgpr0
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec
; GFX10-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX10-NEXT: [[DS_READ_U8_gfx9_:%[0-9]+]]:vgpr_32 = DS_READ_U8_gfx9 [[V_ADD_U32_e64_]], 0, 0, implicit $exec :: (load (s8), addrspace 3)
; GFX10-NEXT: $vgpr0 = COPY [[DS_READ_U8_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_CONSTANT i32 -1
%2:vgpr(p3) = G_PTR_ADD %0, %1
%3:vgpr(s32) = G_LOAD %2 :: (load (s8), align 1, addrspace 3)
$vgpr0 = COPY %3
...
---
name: load_local_s64_align4_from_1_gep_1016
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0_vgpr1
; GFX6-LABEL: name: load_local_s64_align4_from_1_gep_1016
; GFX6: liveins: $vgpr0_vgpr1
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX6-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 1016
; GFX6-NEXT: [[PTR_ADD:%[0-9]+]]:vgpr(p3) = G_PTR_ADD [[COPY]], [[C]](s32)
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[LOAD:%[0-9]+]]:vreg_64(s64) = G_LOAD [[PTR_ADD]](p3) :: (load (s64), align 4, addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](s64)
; GFX7-LABEL: name: load_local_s64_align4_from_1_gep_1016
; GFX7: liveins: $vgpr0_vgpr1
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ2_B32_:%[0-9]+]]:vreg_64 = DS_READ2_B32 [[COPY]], 254, 255, 0, implicit $m0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_]]
; GFX9-LABEL: name: load_local_s64_align4_from_1_gep_1016
; GFX9: liveins: $vgpr0_vgpr1
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 254, 255, 0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
; GFX10-LABEL: name: load_local_s64_align4_from_1_gep_1016
; GFX10: liveins: $vgpr0_vgpr1
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[COPY]], 254, 255, 0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_CONSTANT i32 1016
%2:vgpr(p3) = G_PTR_ADD %0, %1
%3:vgpr(s64) = G_LOAD %2 :: (load (s64), align 4, addrspace 3)
$vgpr0_vgpr1 = COPY %3
...
---
name: load_local_s64_align4_from_1_gep_1020
legalized: true
regBankSelected: true
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0_vgpr1
; GFX6-LABEL: name: load_local_s64_align4_from_1_gep_1020
; GFX6: liveins: $vgpr0_vgpr1
; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:vgpr(p3) = COPY $vgpr0
; GFX6-NEXT: [[C:%[0-9]+]]:vgpr(s32) = G_CONSTANT i32 1020
; GFX6-NEXT: [[PTR_ADD:%[0-9]+]]:vgpr(p3) = G_PTR_ADD [[COPY]], [[C]](s32)
; GFX6-NEXT: $m0 = S_MOV_B32 -1
; GFX6-NEXT: [[LOAD:%[0-9]+]]:vreg_64(s64) = G_LOAD [[PTR_ADD]](p3) :: (load (s64), align 4, addrspace 3)
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[LOAD]](s64)
; GFX7-LABEL: name: load_local_s64_align4_from_1_gep_1020
; GFX7: liveins: $vgpr0_vgpr1
; GFX7-NEXT: {{ $}}
; GFX7-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX7-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1020, implicit $exec
; GFX7-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX7-NEXT: $m0 = S_MOV_B32 -1
; GFX7-NEXT: [[DS_READ2_B32_:%[0-9]+]]:vreg_64 = DS_READ2_B32 [[V_ADD_CO_U32_e64_]], 0, 1, 0, implicit $m0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX7-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_]]
; GFX9-LABEL: name: load_local_s64_align4_from_1_gep_1020
; GFX9: liveins: $vgpr0_vgpr1
; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX9-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1020, implicit $exec
; GFX9-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX9-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[V_ADD_U32_e64_]], 0, 1, 0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
; GFX10-LABEL: name: load_local_s64_align4_from_1_gep_1020
; GFX10: liveins: $vgpr0_vgpr1
; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX10-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1020, implicit $exec
; GFX10-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY]], [[V_MOV_B32_e32_]], 0, implicit $exec
; GFX10-NEXT: [[DS_READ2_B32_gfx9_:%[0-9]+]]:vreg_64 = DS_READ2_B32_gfx9 [[V_ADD_U32_e64_]], 0, 1, 0, implicit $exec :: (load (s64), align 4, addrspace 3)
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[DS_READ2_B32_gfx9_]]
%0:vgpr(p3) = COPY $vgpr0
%1:vgpr(s32) = G_CONSTANT i32 1020
%2:vgpr(p3) = G_PTR_ADD %0, %1
%3:vgpr(s64) = G_LOAD %2 :: (load (s64), align 4, addrspace 3)
$vgpr0_vgpr1 = COPY %3
...
Codebase Browser
llvm