blob: 102fa04011c3e2df8da7ccdc1f2829864a9c9ce6 [file] [edit]
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 6
# RUN: llc -mtriple=amdgpu12.50-amd-amdhsa -run-pass=si-insert-waitcnts -amdgpu-expert-scheduling-mode -o - %s | FileCheck %s
---
name: soft_wait_before_s_barrier
tracksRegLiveness: true
machineFunctionInfo:
isEntryFunction: true
body: |
; CHECK-LABEL: name: soft_wait_before_s_barrier
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $sgpr64_sgpr65 = S_MOV_B64 0
; CHECK-NEXT: V_NOP_e32 implicit $exec
; CHECK-NEXT: GLOBAL_PREFETCH_B8_SADDR $sgpr64_sgpr65, undef $vgpr0, 0, 8, implicit $exec
; CHECK-NEXT: S_SETREG_IMM32_B32 2, 2074, implicit-def $mode, implicit $mode
; CHECK-NEXT: renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
; CHECK-NEXT: renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
; CHECK-NEXT: $vcc_lo = S_MOV_B32 $exec_lo
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_1
; CHECK-NEXT: renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
; CHECK-NEXT: S_WAIT_DSCNT 0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
; CHECK-NEXT: liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_BARRIER_SIGNAL_IMM -1
; CHECK-NEXT: $vgpr1 = V_MOV_B32_e32 $vgpr0, implicit $exec
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_1
; CHECK-NEXT: GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
; CHECK-NEXT: S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: S_ENDPGM 0
bb.0:
successors: %bb.1
renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
$vcc_lo = S_MOV_B32 $exec_lo
renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
bb.1:
successors: %bb.2, %bb.1
liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
S_WAIT_DSCNT_soft 0
S_BARRIER_SIGNAL_IMM -1
$vgpr1 = V_MOV_B32_e32 $vgpr0, implicit $exec
GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc
bb.2:
S_ENDPGM 0
...
---
name: soft_wait_multipath_first_pred_needs_wait
tracksRegLiveness: true
machineFunctionInfo:
isEntryFunction: true
body: |
; CHECK-LABEL: name: soft_wait_multipath_first_pred_needs_wait
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: liveins: $sgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $sgpr64_sgpr65 = S_MOV_B64 0
; CHECK-NEXT: V_NOP_e32 implicit $exec
; CHECK-NEXT: GLOBAL_PREFETCH_B8_SADDR $sgpr64_sgpr65, undef $vgpr0, 0, 8, implicit $exec
; CHECK-NEXT: S_SETREG_IMM32_B32 2, 2074, implicit-def $mode, implicit $mode
; CHECK-NEXT: $vgpr0 = V_MOV_B32_e32 0, implicit $exec
; CHECK-NEXT: S_CMP_EQ_U32 $sgpr0, 0, implicit-def $scc
; CHECK-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: liveins: $vgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_0
; CHECK-NEXT: $vgpr1 = DS_READ_B32_gfx9 $vgpr0, 0, 0, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: liveins: $vgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $vgpr1 = V_MOV_B32_e32 42, implicit $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
; CHECK-NEXT: liveins: $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAIT_DSCNT 0
; CHECK-NEXT: S_BARRIER_SIGNAL_IMM -1
; CHECK-NEXT: $vgpr2 = V_MOV_B32_e32 $vgpr1, implicit $exec
; CHECK-NEXT: S_ENDPGM 0
bb.0:
successors: %bb.1(0x40000000), %bb.2(0x40000000)
liveins: $sgpr0
$vgpr0 = V_MOV_B32_e32 0, implicit $exec
S_CMP_EQ_U32 $sgpr0, 0, implicit-def $scc
S_CBRANCH_SCC1 %bb.2, implicit $scc
bb.1:
successors: %bb.3(0x80000000)
liveins: $vgpr0
$vgpr1 = DS_READ_B32_gfx9 $vgpr0, 0, 0, implicit $exec
S_BRANCH %bb.3
bb.2:
successors: %bb.3(0x80000000)
liveins: $vgpr0
$vgpr1 = V_MOV_B32_e32 42, implicit $exec
bb.3:
liveins: $vgpr0, $vgpr1
S_WAIT_DSCNT_soft 0
S_BARRIER_SIGNAL_IMM -1
$vgpr2 = V_MOV_B32_e32 $vgpr1, implicit $exec
S_ENDPGM 0
...
---
name: soft_wait_second_pred_needs_wait
tracksRegLiveness: true
machineFunctionInfo:
isEntryFunction: true
body: |
; CHECK-LABEL: name: soft_wait_second_pred_needs_wait
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: liveins: $sgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $sgpr64_sgpr65 = S_MOV_B64 0
; CHECK-NEXT: V_NOP_e32 implicit $exec
; CHECK-NEXT: GLOBAL_PREFETCH_B8_SADDR $sgpr64_sgpr65, undef $vgpr0, 0, 8, implicit $exec
; CHECK-NEXT: S_SETREG_IMM32_B32 2, 2074, implicit-def $mode, implicit $mode
; CHECK-NEXT: $vgpr0 = V_MOV_B32_e32 0, implicit $exec
; CHECK-NEXT: S_CMP_EQ_U32 $sgpr0, 0, implicit-def $scc
; CHECK-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: liveins: $vgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $vgpr1 = V_MOV_B32_e32 42, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: liveins: $vgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_0
; CHECK-NEXT: $vgpr1 = DS_READ_B32_gfx9 $vgpr0, 0, 0, implicit $exec
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
; CHECK-NEXT: liveins: $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAIT_DSCNT 0
; CHECK-NEXT: $vgpr2 = V_MOV_B32_e32 $vgpr1, implicit $exec
; CHECK-NEXT: S_ENDPGM 0
bb.0:
successors: %bb.1(0x40000000), %bb.2(0x40000000)
liveins: $sgpr0
$vgpr0 = V_MOV_B32_e32 0, implicit $exec
S_CMP_EQ_U32 $sgpr0, 0, implicit-def $scc
S_CBRANCH_SCC1 %bb.2, implicit $scc
bb.1:
successors: %bb.3(0x80000000)
liveins: $vgpr0
$vgpr1 = V_MOV_B32_e32 42, implicit $exec
S_BRANCH %bb.3
bb.2:
successors: %bb.3(0x80000000)
liveins: $vgpr0
$vgpr1 = DS_READ_B32_gfx9 $vgpr0, 0, 0, implicit $exec
bb.3:
liveins: $vgpr0, $vgpr1
S_WAIT_DSCNT_soft 0
$vgpr2 = V_MOV_B32_e32 $vgpr1, implicit $exec
S_ENDPGM 0
...
# Test that soft wait in RequiredWait does not incorrectly simplify VM_VSRC.
# When processing from the path without DS operations first:
# - soft S_WAIT_DSCNT goes to RequiredWait (AllowWaitDeletion=false)
# - hard S_WAITCNT_DEPCTR with VM_VSRC=0 goes to Wait
# If we incorrectly used RequiredWait in simplifyVmVsrc's CheckWait,
# the VM_VSRC would be simplified away because DS_CNT=0 is present.
# Then when the soft wait is deleted (not needed from second path),
# we'd lose the required VM_VSRC wait.
---
name: soft_wait_does_not_simplify_vm_vsrc
tracksRegLiveness: true
machineFunctionInfo:
isEntryFunction: true
body: |
; CHECK-LABEL: name: soft_wait_does_not_simplify_vm_vsrc
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: liveins: $sgpr0, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $sgpr64_sgpr65 = S_MOV_B64 0
; CHECK-NEXT: V_NOP_e32 implicit $exec
; CHECK-NEXT: GLOBAL_PREFETCH_B8_SADDR $sgpr64_sgpr65, undef $vgpr0, 0, 8, implicit $exec
; CHECK-NEXT: S_SETREG_IMM32_B32 2, 2074, implicit-def $mode, implicit $mode
; CHECK-NEXT: S_CMP_EQ_U32 $sgpr0, 0, implicit-def $scc
; CHECK-NEXT: S_CBRANCH_SCC1 %bb.2, implicit $scc
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: liveins: $vgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $vgpr1 = DS_READ_B32_gfx9 $vgpr0, 0, 0, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.3(0x80000000)
; CHECK-NEXT: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_LOAD_V4_V1_gfx12 $vgpr0, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 15, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
; CHECK-NEXT: liveins: $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAITCNT_DEPCTR .VmVsrc_0
; CHECK-NEXT: S_WAIT_LOADCNT_DSCNT .Loadcnt_0_Dscnt_0
; CHECK-NEXT: $vgpr4 = V_ADD_F32_e32 $vgpr0, $vgpr1, implicit $mode, implicit $exec
; CHECK-NEXT: S_ENDPGM 0
bb.0:
successors: %bb.1(0x40000000), %bb.2(0x40000000)
liveins: $sgpr0, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0
S_CMP_EQ_U32 $sgpr0, 0, implicit-def $scc
S_CBRANCH_SCC1 %bb.2, implicit $scc
bb.1:
successors: %bb.3(0x80000000)
liveins: $vgpr0
; DS operation - requires DS_CNT wait
$vgpr1 = DS_READ_B32_gfx9 $vgpr0, 0, 0, implicit $exec
S_BRANCH %bb.3
bb.2:
successors: %bb.3(0x80000000)
liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0
$vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_LOAD_V4_V1_gfx12 $vgpr0, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 15, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
bb.3:
liveins: $vgpr0, $vgpr1
S_WAIT_DSCNT_soft 0
S_WAITCNT_DEPCTR .VmVsrc_0
$vgpr4 = V_ADD_F32_e32 $vgpr0, $vgpr1, implicit $mode, implicit $exec
S_ENDPGM 0
...
---
name: soft_wait_no_vmsrc_simplify
tracksRegLiveness: true
machineFunctionInfo:
isEntryFunction: true
body: |
; CHECK-LABEL: name: soft_wait_no_vmsrc_simplify
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $sgpr64_sgpr65 = S_MOV_B64 0
; CHECK-NEXT: V_NOP_e32 implicit $exec
; CHECK-NEXT: GLOBAL_PREFETCH_B8_SADDR $sgpr64_sgpr65, undef $vgpr0, 0, 8, implicit $exec
; CHECK-NEXT: S_SETREG_IMM32_B32 2, 2074, implicit-def $mode, implicit $mode
; CHECK-NEXT: renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
; CHECK-NEXT: renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
; CHECK-NEXT: $vcc_lo = S_MOV_B32 $exec_lo
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_1
; CHECK-NEXT: renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
; CHECK-NEXT: S_WAIT_DSCNT 0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
; CHECK-NEXT: liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_0
; CHECK-NEXT: GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
; CHECK-NEXT: S_WAITCNT_DEPCTR .VmVsrc_0
; CHECK-NEXT: $vgpr17 = V_MOV_B32_e32 $vgpr0, implicit $exec
; CHECK-NEXT: S_BARRIER_SIGNAL_IMM -1
; CHECK-NEXT: S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: S_ENDPGM 0
bb.0:
successors: %bb.1
renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
$vcc_lo = S_MOV_B32 $exec_lo
renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
bb.1:
successors: %bb.2, %bb.1
liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
S_WAIT_DSCNT_soft 0
$vgpr17 = V_MOV_B32_e32 $vgpr0, implicit $exec
S_BARRIER_SIGNAL_IMM -1
S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc
bb.2:
S_ENDPGM 0
...
---
name: soft_wait_no_vmsrc_simplify2
tracksRegLiveness: true
machineFunctionInfo:
isEntryFunction: true
body: |
; CHECK-LABEL: name: soft_wait_no_vmsrc_simplify2
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $sgpr64_sgpr65 = S_MOV_B64 0
; CHECK-NEXT: V_NOP_e32 implicit $exec
; CHECK-NEXT: GLOBAL_PREFETCH_B8_SADDR $sgpr64_sgpr65, undef $vgpr0, 0, 8, implicit $exec
; CHECK-NEXT: S_SETREG_IMM32_B32 2, 2074, implicit-def $mode, implicit $mode
; CHECK-NEXT: renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
; CHECK-NEXT: renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
; CHECK-NEXT: $vcc_lo = S_MOV_B32 $exec_lo
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_1
; CHECK-NEXT: renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
; CHECK-NEXT: S_WAIT_DSCNT 0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
; CHECK-NEXT: liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAITCNT_DEPCTR .VmVsrc_0
; CHECK-NEXT: $vgpr17 = V_MOV_B32_e32 $vgpr0, implicit $exec
; CHECK-NEXT: S_BARRIER_SIGNAL_IMM -1
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_0
; CHECK-NEXT: GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
; CHECK-NEXT: S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: S_ENDPGM 0
bb.0:
successors: %bb.1
renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
$vcc_lo = S_MOV_B32 $exec_lo
renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
bb.1:
successors: %bb.2, %bb.1
liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
S_WAIT_DSCNT_soft 0
$vgpr17 = V_MOV_B32_e32 $vgpr0, implicit $exec
S_BARRIER_SIGNAL_IMM -1
GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc
bb.2:
S_ENDPGM 0
...
---
name: soft_wait_promote_simplify
tracksRegLiveness: true
machineFunctionInfo:
isEntryFunction: true
body: |
; CHECK-LABEL: name: soft_wait_promote_simplify
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $sgpr64_sgpr65 = S_MOV_B64 0
; CHECK-NEXT: V_NOP_e32 implicit $exec
; CHECK-NEXT: GLOBAL_PREFETCH_B8_SADDR $sgpr64_sgpr65, undef $vgpr0, 0, 8, implicit $exec
; CHECK-NEXT: S_SETREG_IMM32_B32 2, 2074, implicit-def $mode, implicit $mode
; CHECK-NEXT: renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
; CHECK-NEXT: renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
; CHECK-NEXT: $vcc_lo = S_MOV_B32 $exec_lo
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_1
; CHECK-NEXT: renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
; CHECK-NEXT: S_WAIT_DSCNT 0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
; CHECK-NEXT: liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_WAIT_DSCNT 0
; CHECK-NEXT: S_WAITCNT_DEPCTR .VmVsrc_0
; CHECK-NEXT: $vgpr17 = V_MOV_B32_e32 $vgpr0, implicit $exec
; CHECK-NEXT: S_BARRIER_SIGNAL_IMM -1
; CHECK-NEXT: S_WAITCNT_DEPCTR .VaVdst_0
; CHECK-NEXT: GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
; CHECK-NEXT: renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
; CHECK-NEXT: S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc_lo
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: S_ENDPGM 0
bb.0:
successors: %bb.1
renamable $vgpr17 = V_MOV_B32_e32 0, implicit $exec
renamable $vgpr32_vgpr33 = V_MOV_B64_e32 0, implicit $exec
$vcc_lo = S_MOV_B32 $exec_lo
renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
bb.1:
successors: %bb.2, %bb.1
liveins: $vgpr32_vgpr33, $vgpr17, $vgpr0, $vcc_lo
S_WAIT_DSCNT_soft 0
$vgpr17 = V_MOV_B32_e32 $vgpr0, implicit $exec
S_BARRIER_SIGNAL_IMM -1
GLOBAL_LOAD_ASYNC_TO_LDS_B64 $vgpr17, $vgpr32_vgpr33, 0, 0, implicit-def dead $asynccnt, implicit $exec, implicit $asynccnt
renamable $vgpr0 = DS_READ_B32_gfx9 renamable $vgpr17, 0, 0, implicit $exec
S_CBRANCH_VCCNZ %bb.1, implicit $vcc_lo, implicit $vcc
bb.2:
S_ENDPGM 0
...