265e94e657
Summary: Extend cachepolicy operand in the new VMEM buffer intrinsics to supply information whether the buffer data is swizzled. Also, propagate this information to MIR. Intrinsics updated: int_amdgcn_raw_buffer_load int_amdgcn_raw_buffer_load_format int_amdgcn_raw_buffer_store int_amdgcn_raw_buffer_store_format int_amdgcn_raw_tbuffer_load int_amdgcn_raw_tbuffer_store int_amdgcn_struct_buffer_load int_amdgcn_struct_buffer_load_format int_amdgcn_struct_buffer_store int_amdgcn_struct_buffer_store_format int_amdgcn_struct_tbuffer_load int_amdgcn_struct_tbuffer_store Furthermore, disable merging of VMEM buffer instructions in SI Load/Store optimizer, if the "swizzled" bit on the instruction is on. The default value of the bit is 0, meaning that data in buffer is linear and buffer instructions can be merged. There is no difference in the generated code with this commit. However, in the future it will be expected that front-ends use buffer intrinsics with correct "swizzled" bit set. Reviewers: arsenm, nhaehnle, tpr Reviewed By: nhaehnle Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, arphaman, jfb, Petar.Avramovic, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D68200 llvm-svn: 373491
906 lines
27 KiB
YAML
906 lines
27 KiB
YAML
# RUN: llc -mtriple=amdgcn--amdhsa -mcpu=gfx900 -verify-machineinstrs -run-pass si-fold-operands,dead-mi-elimination -o - %s | FileCheck -check-prefix=GCN %s
|
|
...
|
|
|
|
# GCN-LABEL: name: s_fold_and_imm_regimm_32{{$}}
|
|
# GCN: %10:vgpr_32 = V_MOV_B32_e32 1543, implicit $exec
|
|
# GCN: BUFFER_STORE_DWORD_OFFSET killed %10,
|
|
name: s_fold_and_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_64_xexec }
|
|
- { id: 2, class: sreg_32_xm0 }
|
|
- { id: 3, class: sreg_32_xm0 }
|
|
- { id: 4, class: sreg_32_xm0 }
|
|
- { id: 5, class: sreg_32_xm0 }
|
|
- { id: 6, class: sreg_128 }
|
|
- { id: 7, class: sreg_32_xm0 }
|
|
- { id: 8, class: sreg_32_xm0 }
|
|
- { id: 9, class: sreg_32_xm0 }
|
|
- { id: 10, class: vgpr_32 }
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1
|
|
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%1 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0
|
|
%2 = COPY %1.sub1
|
|
%3 = COPY %1.sub0
|
|
%4 = S_MOV_B32 61440
|
|
%5 = S_MOV_B32 -1
|
|
%6 = REG_SEQUENCE killed %2, 1, killed %3, 2, killed %4, 3, killed %5, 4
|
|
%7 = S_MOV_B32 1234567
|
|
%8 = S_MOV_B32 9999
|
|
%9 = S_AND_B32 killed %7, killed %8, implicit-def dead $scc
|
|
%10 = COPY %9
|
|
BUFFER_STORE_DWORD_OFFSET killed %10, killed %6, 0, 0, 0, 0, 0, 0, 0, implicit $exec
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
|
|
# GCN-LABEL: name: v_fold_and_imm_regimm_32{{$}}
|
|
|
|
# GCN: %9:vgpr_32 = V_MOV_B32_e32 646, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %19, %9,
|
|
|
|
# GCN: %10:vgpr_32 = V_MOV_B32_e32 646, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %19, %10
|
|
|
|
# GCN: %11:vgpr_32 = V_MOV_B32_e32 646, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %19, %11,
|
|
|
|
# GCN: %12:vgpr_32 = V_MOV_B32_e32 1234567, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %19, %12,
|
|
|
|
# GCN: %13:vgpr_32 = V_MOV_B32_e32 63, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %19, %13,
|
|
|
|
name: v_fold_and_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_32_xm0 }
|
|
- { id: 2, class: sgpr_32 }
|
|
- { id: 3, class: vgpr_32 }
|
|
- { id: 4, class: sreg_64_xexec }
|
|
- { id: 20, class: sreg_32_xm0 }
|
|
- { id: 24, class: vgpr_32 }
|
|
- { id: 25, class: vreg_64 }
|
|
- { id: 26, class: sreg_32_xm0 }
|
|
- { id: 27, class: vgpr_32 }
|
|
- { id: 28, class: vgpr_32 }
|
|
- { id: 29, class: vgpr_32 }
|
|
- { id: 30, class: vgpr_32 }
|
|
- { id: 31, class: vgpr_32 }
|
|
- { id: 32, class: vreg_64 }
|
|
- { id: 33, class: vreg_64 }
|
|
- { id: 34, class: vgpr_32 }
|
|
- { id: 35, class: vgpr_32 }
|
|
- { id: 36, class: vgpr_32 }
|
|
- { id: 37, class: vreg_64 }
|
|
- { id: 44, class: vgpr_32 }
|
|
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
- { reg: '$vgpr0', virtual-reg: '%3' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1, $vgpr0
|
|
|
|
%3 = COPY $vgpr0
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%4 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0
|
|
%31 = V_ASHRREV_I32_e64 31, %3, implicit $exec
|
|
%32 = REG_SEQUENCE %3, 1, %31, 2
|
|
%33 = V_LSHLREV_B64 2, killed %32, implicit $exec
|
|
%20 = COPY %4.sub1
|
|
%44 = V_ADD_I32_e32 %4.sub0, %33.sub0, implicit-def $vcc, implicit $exec
|
|
%36 = COPY killed %20
|
|
%35 = V_ADDC_U32_e32 %33.sub1, %36, implicit-def $vcc, implicit $vcc, implicit $exec
|
|
%37 = REG_SEQUENCE %44, 1, killed %35, 2
|
|
%24 = V_MOV_B32_e32 982, implicit $exec
|
|
%26 = S_MOV_B32 1234567
|
|
%34 = V_MOV_B32_e32 63, implicit $exec
|
|
|
|
%27 = V_AND_B32_e64 %26, %24, implicit $exec
|
|
FLAT_STORE_DWORD %37, %27, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%28 = V_AND_B32_e64 %24, %26, implicit $exec
|
|
FLAT_STORE_DWORD %37, %28, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%29 = V_AND_B32_e32 %26, %24, implicit $exec
|
|
FLAT_STORE_DWORD %37, %29, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%30 = V_AND_B32_e64 %26, %26, implicit $exec
|
|
FLAT_STORE_DWORD %37, %30, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%31 = V_AND_B32_e64 %34, %34, implicit $exec
|
|
FLAT_STORE_DWORD %37, %31, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
|
|
# GCN-LABEL: name: s_fold_shl_imm_regimm_32{{$}}
|
|
# GC1: %13 = V_MOV_B32_e32 4096, implicit $exec
|
|
# GCN: BUFFER_STORE_DWORD_OFFSET killed %13,
|
|
|
|
name: s_fold_shl_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_32_xm0 }
|
|
- { id: 2, class: sgpr_32 }
|
|
- { id: 3, class: vgpr_32 }
|
|
- { id: 4, class: sreg_64_xexec }
|
|
- { id: 5, class: sreg_32_xm0_xexec }
|
|
- { id: 6, class: sreg_32_xm0 }
|
|
- { id: 7, class: sreg_32_xm0 }
|
|
- { id: 8, class: sreg_32_xm0 }
|
|
- { id: 9, class: sreg_32_xm0 }
|
|
- { id: 10, class: sreg_128 }
|
|
- { id: 11, class: sreg_32_xm0 }
|
|
- { id: 12, class: sreg_32_xm0 }
|
|
- { id: 13, class: vgpr_32 }
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1
|
|
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%4 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0
|
|
%5 = S_MOV_B32 1
|
|
%6 = COPY %4.sub1
|
|
%7 = COPY %4.sub0
|
|
%8 = S_MOV_B32 61440
|
|
%9 = S_MOV_B32 -1
|
|
%10 = REG_SEQUENCE killed %7, 1, killed %6, 2, killed %9, 3, killed %8, 4
|
|
%12 = S_LSHL_B32 killed %5, 12, implicit-def dead $scc
|
|
%13 = COPY %12
|
|
BUFFER_STORE_DWORD_OFFSET killed %13, killed %10, 0, 0, 0, 0, 0, 0, 0, implicit $exec
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
# GCN-LABEL: name: v_fold_shl_imm_regimm_32{{$}}
|
|
|
|
# GCN: %11:vgpr_32 = V_MOV_B32_e32 40955904, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %11,
|
|
|
|
# GCN: %12:vgpr_32 = V_MOV_B32_e32 24, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %12,
|
|
|
|
# GCN: %13:vgpr_32 = V_MOV_B32_e32 4096, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %13,
|
|
|
|
# GCN: %14:vgpr_32 = V_MOV_B32_e32 24, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %14,
|
|
|
|
# GCN: %15:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %15,
|
|
|
|
# GCN: %22:vgpr_32 = V_MOV_B32_e32 4096, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %22,
|
|
|
|
# GCN: %23:vgpr_32 = V_MOV_B32_e32 1, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %23,
|
|
|
|
# GCN: %25:vgpr_32 = V_MOV_B32_e32 2, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %25,
|
|
|
|
# GCN: %26:vgpr_32 = V_MOV_B32_e32 7927808, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %26,
|
|
|
|
# GCN: %28:vgpr_32 = V_MOV_B32_e32 -8, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %28,
|
|
|
|
name: v_fold_shl_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_32_xm0 }
|
|
- { id: 2, class: vgpr_32 }
|
|
- { id: 3, class: sreg_64_xexec }
|
|
- { id: 4, class: sreg_64_xexec }
|
|
- { id: 5, class: sreg_32_xm0 }
|
|
- { id: 6, class: vgpr_32 }
|
|
- { id: 7, class: sreg_32_xm0 }
|
|
- { id: 8, class: sreg_64 }
|
|
- { id: 9, class: sreg_32_xm0 }
|
|
- { id: 10, class: vgpr_32 }
|
|
- { id: 11, class: vgpr_32 }
|
|
- { id: 12, class: vgpr_32 }
|
|
- { id: 13, class: vgpr_32 }
|
|
- { id: 14, class: vgpr_32 }
|
|
- { id: 15, class: vgpr_32 }
|
|
- { id: 16, class: vreg_64 }
|
|
- { id: 17, class: vreg_64 }
|
|
- { id: 18, class: vgpr_32 }
|
|
- { id: 19, class: vgpr_32 }
|
|
- { id: 20, class: vreg_64 }
|
|
- { id: 21, class: vgpr_32 }
|
|
- { id: 22, class: vgpr_32 }
|
|
- { id: 23, class: vgpr_32 }
|
|
- { id: 24, class: vgpr_32 }
|
|
- { id: 25, class: vgpr_32 }
|
|
- { id: 26, class: vgpr_32 }
|
|
- { id: 27, class: sreg_32_xm0 }
|
|
- { id: 28, class: vgpr_32 }
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
- { reg: '$vgpr0', virtual-reg: '%2' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1, $vgpr0
|
|
|
|
%2 = COPY $vgpr0
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%3 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0 :: (non-temporal dereferenceable invariant load 8 from `i64 addrspace(4)* undef`)
|
|
%15 = V_ASHRREV_I32_e64 31, %2, implicit $exec
|
|
%16 = REG_SEQUENCE %2, 1, %15, 2
|
|
%17 = V_LSHLREV_B64 2, killed %16, implicit $exec
|
|
%9 = COPY %3.sub1
|
|
%21 = V_ADD_I32_e32 %3.sub0, %17.sub0, implicit-def $vcc, implicit $exec
|
|
%19 = COPY killed %9
|
|
%18 = V_ADDC_U32_e32 %17.sub1, %19, implicit-def $vcc, implicit $vcc, implicit $exec
|
|
%20 = REG_SEQUENCE %21, 1, killed %18, 2
|
|
%10 = V_MOV_B32_e32 9999, implicit $exec
|
|
%24 = V_MOV_B32_e32 3871, implicit $exec
|
|
%6 = V_MOV_B32_e32 1, implicit $exec
|
|
%7 = S_MOV_B32 1
|
|
%27 = S_MOV_B32 -4
|
|
|
|
%11 = V_LSHLREV_B32_e64 12, %10, implicit $exec
|
|
FLAT_STORE_DWORD %20, %11, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%12 = V_LSHLREV_B32_e64 %7, 12, implicit $exec
|
|
FLAT_STORE_DWORD %20, %12, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%13 = V_LSHL_B32_e64 %7, 12, implicit $exec
|
|
FLAT_STORE_DWORD %20, %13, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%14 = V_LSHL_B32_e64 12, %7, implicit $exec
|
|
FLAT_STORE_DWORD %20, %14, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%15 = V_LSHL_B32_e64 12, %24, implicit $exec
|
|
FLAT_STORE_DWORD %20, %15, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%22 = V_LSHL_B32_e64 %6, 12, implicit $exec
|
|
FLAT_STORE_DWORD %20, %22, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%23 = V_LSHL_B32_e64 %6, 32, implicit $exec
|
|
FLAT_STORE_DWORD %20, %23, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%25 = V_LSHL_B32_e32 %6, %6, implicit $exec
|
|
FLAT_STORE_DWORD %20, %25, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%26 = V_LSHLREV_B32_e32 11, %24, implicit $exec
|
|
FLAT_STORE_DWORD %20, %26, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%28 = V_LSHL_B32_e32 %27, %6, implicit $exec
|
|
FLAT_STORE_DWORD %20, %28, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
|
|
# GCN-LABEL: name: s_fold_ashr_imm_regimm_32{{$}}
|
|
# GCN: %11:vgpr_32 = V_MOV_B32_e32 243, implicit $exec
|
|
# GCN: BUFFER_STORE_DWORD_OFFSET killed %11, killed %8,
|
|
name: s_fold_ashr_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_32_xm0 }
|
|
- { id: 4, class: sreg_64_xexec }
|
|
- { id: 5, class: sreg_32_xm0_xexec }
|
|
- { id: 6, class: sreg_32_xm0 }
|
|
- { id: 7, class: sreg_32_xm0 }
|
|
- { id: 8, class: sreg_32_xm0 }
|
|
- { id: 9, class: sreg_32_xm0 }
|
|
- { id: 10, class: sreg_128 }
|
|
- { id: 11, class: sreg_32_xm0 }
|
|
- { id: 12, class: sreg_32_xm0 }
|
|
- { id: 13, class: vgpr_32 }
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1
|
|
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%4 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0
|
|
%5 = S_MOV_B32 999123
|
|
%6 = COPY %4.sub1
|
|
%7 = COPY %4.sub0
|
|
%8 = S_MOV_B32 61440
|
|
%9 = S_MOV_B32 -1
|
|
%10 = REG_SEQUENCE killed %7, 1, killed %6, 2, killed %9, 3, killed %8, 4
|
|
%12 = S_ASHR_I32 killed %5, 12, implicit-def dead $scc
|
|
%13 = COPY %12
|
|
BUFFER_STORE_DWORD_OFFSET killed %13, killed %10, 0, 0, 0, 0, 0, 0, 0, implicit $exec
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
|
|
# GCN-LABEL: name: v_fold_ashr_imm_regimm_32{{$}}
|
|
# GCN: %11:vgpr_32 = V_MOV_B32_e32 3903258, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %11,
|
|
|
|
# GCN: %12:vgpr_32 = V_MOV_B32_e32 62452139, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %12,
|
|
|
|
# GCN: %13:vgpr_32 = V_MOV_B32_e32 1678031, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %13,
|
|
|
|
# GCN: %14:vgpr_32 = V_MOV_B32_e32 3, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %14,
|
|
|
|
# GCN: %15:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %15,
|
|
|
|
# GCN: %22:vgpr_32 = V_MOV_B32_e32 62500, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %22,
|
|
|
|
# GCN: %23:vgpr_32 = V_MOV_B32_e32 500000, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %23,
|
|
|
|
# GCN: %25:vgpr_32 = V_MOV_B32_e32 1920, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %25,
|
|
|
|
# GCN: %26:vgpr_32 = V_MOV_B32_e32 487907, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %26,
|
|
|
|
# GCN: %28:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %28,
|
|
|
|
name: v_fold_ashr_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_32_xm0 }
|
|
- { id: 2, class: vgpr_32 }
|
|
- { id: 3, class: sreg_64_xexec }
|
|
- { id: 4, class: sreg_64_xexec }
|
|
- { id: 5, class: sreg_32_xm0 }
|
|
- { id: 6, class: vgpr_32 }
|
|
- { id: 7, class: sreg_32_xm0 }
|
|
- { id: 8, class: sreg_32_xm0 }
|
|
- { id: 9, class: sreg_32_xm0 }
|
|
- { id: 10, class: vgpr_32 }
|
|
- { id: 11, class: vgpr_32 }
|
|
- { id: 12, class: vgpr_32 }
|
|
- { id: 13, class: vgpr_32 }
|
|
- { id: 14, class: vgpr_32 }
|
|
- { id: 15, class: vgpr_32 }
|
|
- { id: 16, class: vreg_64 }
|
|
- { id: 17, class: vreg_64 }
|
|
- { id: 18, class: vgpr_32 }
|
|
- { id: 19, class: vgpr_32 }
|
|
- { id: 20, class: vreg_64 }
|
|
- { id: 21, class: vgpr_32 }
|
|
- { id: 22, class: vgpr_32 }
|
|
- { id: 23, class: vgpr_32 }
|
|
- { id: 24, class: vgpr_32 }
|
|
- { id: 25, class: vgpr_32 }
|
|
- { id: 26, class: vgpr_32 }
|
|
- { id: 27, class: sreg_32_xm0 }
|
|
- { id: 28, class: vgpr_32 }
|
|
- { id: 32, class: sreg_32_xm0 }
|
|
- { id: 33, class: sreg_32_xm0 }
|
|
- { id: 34, class: vgpr_32 }
|
|
- { id: 35, class: vgpr_32 }
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
- { reg: '$vgpr0', virtual-reg: '%2' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1, $vgpr0
|
|
|
|
%2 = COPY $vgpr0
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%3 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0
|
|
%15 = V_ASHRREV_I32_e64 31, %2, implicit $exec
|
|
%16 = REG_SEQUENCE %2, 1, %15, 2
|
|
%17 = V_LSHLREV_B64 2, killed %16, implicit $exec
|
|
%9 = COPY %3.sub1
|
|
%21 = V_ADD_I32_e32 %3.sub0, %17.sub0, implicit-def $vcc, implicit $exec
|
|
%19 = COPY killed %9
|
|
%18 = V_ADDC_U32_e32 %17.sub1, %19, implicit-def $vcc, implicit $vcc, implicit $exec
|
|
%20 = REG_SEQUENCE %21, 1, killed %18, 2
|
|
%10 = V_MOV_B32_e32 999234234, implicit $exec
|
|
%24 = V_MOV_B32_e32 3871, implicit $exec
|
|
%6 = V_MOV_B32_e32 1000000, implicit $exec
|
|
%7 = S_MOV_B32 13424252
|
|
%8 = S_MOV_B32 4
|
|
%27 = S_MOV_B32 -4
|
|
%32 = S_MOV_B32 1
|
|
%33 = S_MOV_B32 3841
|
|
%34 = V_MOV_B32_e32 3841, implicit $exec
|
|
%35 = V_MOV_B32_e32 2, implicit $exec
|
|
|
|
%11 = V_ASHRREV_I32_e64 8, %10, implicit $exec
|
|
FLAT_STORE_DWORD %20, %11, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%12 = V_ASHRREV_I32_e64 %8, %10, implicit $exec
|
|
FLAT_STORE_DWORD %20, %12, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%13 = V_ASHR_I32_e64 %7, 3, implicit $exec
|
|
FLAT_STORE_DWORD %20, %13, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%14 = V_ASHR_I32_e64 7, %32, implicit $exec
|
|
FLAT_STORE_DWORD %20, %14, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%15 = V_ASHR_I32_e64 %27, %24, implicit $exec
|
|
FLAT_STORE_DWORD %20, %15, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%22 = V_ASHR_I32_e64 %6, 4, implicit $exec
|
|
FLAT_STORE_DWORD %20, %22, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%23 = V_ASHR_I32_e64 %6, %33, implicit $exec
|
|
FLAT_STORE_DWORD %20, %23, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%25 = V_ASHR_I32_e32 %34, %34, implicit $exec
|
|
FLAT_STORE_DWORD %20, %25, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%26 = V_ASHRREV_I32_e32 11, %10, implicit $exec
|
|
FLAT_STORE_DWORD %20, %26, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%28 = V_ASHR_I32_e32 %27, %35, implicit $exec
|
|
FLAT_STORE_DWORD %20, %28, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
|
|
# GCN-LABEL: name: s_fold_lshr_imm_regimm_32{{$}}
|
|
# GCN: %11:vgpr_32 = V_MOV_B32_e32 1048332, implicit $exec
|
|
# GCN: BUFFER_STORE_DWORD_OFFSET killed %11, killed %8,
|
|
name: s_fold_lshr_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_32_xm0 }
|
|
- { id: 4, class: sreg_64_xexec }
|
|
- { id: 5, class: sreg_32_xm0_xexec }
|
|
- { id: 6, class: sreg_32_xm0 }
|
|
- { id: 7, class: sreg_32_xm0 }
|
|
- { id: 8, class: sreg_32_xm0 }
|
|
- { id: 9, class: sreg_32_xm0 }
|
|
- { id: 10, class: sreg_128 }
|
|
- { id: 11, class: sreg_32_xm0 }
|
|
- { id: 12, class: sreg_32_xm0 }
|
|
- { id: 13, class: vgpr_32 }
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1
|
|
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%4 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0
|
|
%5 = S_MOV_B32 -999123
|
|
%6 = COPY %4.sub1
|
|
%7 = COPY %4.sub0
|
|
%8 = S_MOV_B32 61440
|
|
%9 = S_MOV_B32 -1
|
|
%10 = REG_SEQUENCE killed %7, 1, killed %6, 2, killed %9, 3, killed %8, 4
|
|
%12 = S_LSHR_B32 killed %5, 12, implicit-def dead $scc
|
|
%13 = COPY %12
|
|
BUFFER_STORE_DWORD_OFFSET killed %13, killed %10, 0, 0, 0, 0, 0, 0, 0, implicit $exec
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
|
|
# GCN-LABEL: name: v_fold_lshr_imm_regimm_32{{$}}
|
|
# GCN: %11:vgpr_32 = V_MOV_B32_e32 3903258, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %11,
|
|
|
|
# GCN: %12:vgpr_32 = V_MOV_B32_e32 62452139, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %12,
|
|
|
|
# GCN: %13:vgpr_32 = V_MOV_B32_e32 1678031, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %13,
|
|
|
|
# GCN: %14:vgpr_32 = V_MOV_B32_e32 3, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %14,
|
|
|
|
# GCN: %15:vgpr_32 = V_MOV_B32_e32 1, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %15,
|
|
|
|
# GCN: %22:vgpr_32 = V_MOV_B32_e32 62500, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %22,
|
|
|
|
# GCN: %23:vgpr_32 = V_MOV_B32_e32 500000, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %23,
|
|
|
|
# GCN: %25:vgpr_32 = V_MOV_B32_e32 1920, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %25,
|
|
|
|
# GCN: %26:vgpr_32 = V_MOV_B32_e32 487907, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %26,
|
|
|
|
# GCN: %28:vgpr_32 = V_MOV_B32_e32 1073741823, implicit $exec
|
|
# GCN: FLAT_STORE_DWORD %20, %28,
|
|
|
|
name: v_fold_lshr_imm_regimm_32
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: sgpr_64 }
|
|
- { id: 1, class: sreg_32_xm0 }
|
|
- { id: 2, class: vgpr_32 }
|
|
- { id: 3, class: sreg_64_xexec }
|
|
- { id: 4, class: sreg_64_xexec }
|
|
- { id: 5, class: sreg_32_xm0 }
|
|
- { id: 6, class: vgpr_32 }
|
|
- { id: 7, class: sreg_32_xm0 }
|
|
- { id: 8, class: sreg_32_xm0 }
|
|
- { id: 9, class: sreg_32_xm0 }
|
|
- { id: 10, class: vgpr_32 }
|
|
- { id: 11, class: vgpr_32 }
|
|
- { id: 12, class: vgpr_32 }
|
|
- { id: 13, class: vgpr_32 }
|
|
- { id: 14, class: vgpr_32 }
|
|
- { id: 15, class: vgpr_32 }
|
|
- { id: 16, class: vreg_64 }
|
|
- { id: 17, class: vreg_64 }
|
|
- { id: 18, class: vgpr_32 }
|
|
- { id: 19, class: vgpr_32 }
|
|
- { id: 20, class: vreg_64 }
|
|
- { id: 21, class: vgpr_32 }
|
|
- { id: 22, class: vgpr_32 }
|
|
- { id: 23, class: vgpr_32 }
|
|
- { id: 24, class: vgpr_32 }
|
|
- { id: 25, class: vgpr_32 }
|
|
- { id: 26, class: vgpr_32 }
|
|
- { id: 27, class: sreg_32_xm0 }
|
|
- { id: 28, class: vgpr_32 }
|
|
- { id: 32, class: sreg_32_xm0 }
|
|
- { id: 33, class: sreg_32_xm0 }
|
|
- { id: 34, class: vgpr_32 }
|
|
- { id: 35, class: vgpr_32 }
|
|
liveins:
|
|
- { reg: '$sgpr0_sgpr1', virtual-reg: '%0' }
|
|
- { reg: '$vgpr0', virtual-reg: '%2' }
|
|
frameInfo:
|
|
isFrameAddressTaken: false
|
|
isReturnAddressTaken: false
|
|
hasStackMap: false
|
|
hasPatchPoint: false
|
|
stackSize: 0
|
|
offsetAdjustment: 0
|
|
maxAlignment: 0
|
|
adjustsStack: false
|
|
hasCalls: false
|
|
maxCallFrameSize: 0
|
|
hasOpaqueSPAdjustment: false
|
|
hasVAStart: false
|
|
hasMustTailInVarArgFunc: false
|
|
body: |
|
|
bb.0:
|
|
liveins: $sgpr0_sgpr1, $vgpr0
|
|
|
|
%2 = COPY $vgpr0
|
|
%0 = COPY $sgpr0_sgpr1
|
|
%3 = S_LOAD_DWORDX2_IMM %0, 36, 0, 0
|
|
%15 = V_ASHRREV_I32_e64 31, %2, implicit $exec
|
|
%16 = REG_SEQUENCE %2, 1, %15, 2
|
|
%17 = V_LSHLREV_B64 2, killed %16, implicit $exec
|
|
%9 = COPY %3.sub1
|
|
%21 = V_ADD_I32_e32 %3.sub0, %17.sub0, implicit-def $vcc, implicit $exec
|
|
%19 = COPY killed %9
|
|
%18 = V_ADDC_U32_e32 %17.sub1, %19, implicit-def $vcc, implicit $vcc, implicit $exec
|
|
%20 = REG_SEQUENCE %21, 1, killed %18, 2
|
|
%10 = V_MOV_B32_e32 999234234, implicit $exec
|
|
%24 = V_MOV_B32_e32 3871, implicit $exec
|
|
%6 = V_MOV_B32_e32 1000000, implicit $exec
|
|
%7 = S_MOV_B32 13424252
|
|
%8 = S_MOV_B32 4
|
|
%27 = S_MOV_B32 -4
|
|
%32 = S_MOV_B32 1
|
|
%33 = S_MOV_B32 3841
|
|
%34 = V_MOV_B32_e32 3841, implicit $exec
|
|
%35 = V_MOV_B32_e32 2, implicit $exec
|
|
|
|
%11 = V_LSHRREV_B32_e64 8, %10, implicit $exec
|
|
FLAT_STORE_DWORD %20, %11, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%12 = V_LSHRREV_B32_e64 %8, %10, implicit $exec
|
|
FLAT_STORE_DWORD %20, %12, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%13 = V_LSHR_B32_e64 %7, 3, implicit $exec
|
|
FLAT_STORE_DWORD %20, %13, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%14 = V_LSHR_B32_e64 7, %32, implicit $exec
|
|
FLAT_STORE_DWORD %20, %14, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%15 = V_LSHR_B32_e64 %27, %24, implicit $exec
|
|
FLAT_STORE_DWORD %20, %15, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%22 = V_LSHR_B32_e64 %6, 4, implicit $exec
|
|
FLAT_STORE_DWORD %20, %22, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%23 = V_LSHR_B32_e64 %6, %33, implicit $exec
|
|
FLAT_STORE_DWORD %20, %23, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%25 = V_LSHR_B32_e32 %34, %34, implicit $exec
|
|
FLAT_STORE_DWORD %20, %25, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%26 = V_LSHRREV_B32_e32 11, %10, implicit $exec
|
|
FLAT_STORE_DWORD %20, %26, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
%28 = V_LSHR_B32_e32 %27, %35, implicit $exec
|
|
FLAT_STORE_DWORD %20, %28, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
# There is only an undef use operand for %1, so there is no
|
|
# corresponding defining instruction
|
|
|
|
# GCN-LABEL: name: undefined_vreg_operand{{$}}
|
|
# GCN: bb.0
|
|
# GCN-NEXT: FLAT_STORE_DWORD undef %3:vreg_64, undef %1:vgpr_32,
|
|
# GCN-NEXT: S_ENDPGM 0
|
|
name: undefined_vreg_operand
|
|
tracksRegLiveness: true
|
|
registers:
|
|
- { id: 0, class: vgpr_32, preferred-register: '' }
|
|
- { id: 1, class: vgpr_32, preferred-register: '' }
|
|
- { id: 2, class: vgpr_32, preferred-register: '' }
|
|
- { id: 3, class: vreg_64, preferred-register: '' }
|
|
body: |
|
|
bb.0:
|
|
%0 = V_MOV_B32_e32 0, implicit $exec
|
|
%2 = V_XOR_B32_e64 killed %0, undef %1, implicit $exec
|
|
FLAT_STORE_DWORD undef %3, %2, 0, 0, 0, 0, implicit $exec, implicit $flat_scr
|
|
S_ENDPGM 0
|
|
|
|
...
|
|
---
|
|
# Make sure there is no crash if one of the operands is a physical register
|
|
# GCN-LABEL: name: constant_fold_physreg_op{{$}}
|
|
# GCN: %3:sreg_64 = S_AND_B64 $exec, 0, implicit-def dead $scc
|
|
|
|
name: constant_fold_physreg_op
|
|
tracksRegLiveness: true
|
|
body: |
|
|
bb.0:
|
|
successors: %bb.1, %bb.3
|
|
liveins: $vgpr0, $sgpr4_sgpr5
|
|
|
|
%19:sreg_64 = IMPLICIT_DEF
|
|
%0:sreg_64 = SI_IF killed %19, %bb.3, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
|
|
S_BRANCH %bb.1
|
|
|
|
bb.1:
|
|
%6:sreg_64 = S_MOV_B64 0
|
|
%7:sreg_64 = S_AND_B64 $exec, killed %6, implicit-def dead $scc
|
|
$vcc = COPY %7
|
|
|
|
bb.3:
|
|
liveins: $vcc
|
|
SI_END_CF %0, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
|
|
S_ENDPGM 0, implicit $vcc
|
|
|
|
...
|
|
---
|
|
# GCN-LABEL: name: constant_fold_lshl_or_reg0_immreg_reg{{$}}
|
|
# GCN: %2:vgpr_32 = COPY $vgpr0, implicit $exec
|
|
# GCN-NEXT: S_ENDPGM 0, implicit %2
|
|
|
|
name: constant_fold_lshl_or_reg0_immreg_reg
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0
|
|
|
|
%0:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
%1:vgpr_32 = V_MOV_B32_e32 16, implicit $exec
|
|
%2:vgpr_32 = V_LSHL_OR_B32 %0,%1, $vgpr0, implicit $exec
|
|
S_ENDPGM 0, implicit %2
|
|
|
|
...
|
|
|
|
---
|
|
|
|
# GCN-LABEL: name: constant_fold_lshl_or_reg0_immreg_imm{{$}}
|
|
# GCN: %2:vgpr_32 = V_MOV_B32_e32 10, implicit $exec
|
|
# GCN-NEXT: S_ENDPGM 0, implicit %2
|
|
|
|
name: constant_fold_lshl_or_reg0_immreg_imm
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
|
|
body: |
|
|
bb.0:
|
|
|
|
%0:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
%1:vgpr_32 = V_MOV_B32_e32 16, implicit $exec
|
|
%2:vgpr_32 = V_LSHL_OR_B32 %0, %1, 10, implicit $exec
|
|
S_ENDPGM 0, implicit %2
|
|
|
|
...
|
|
|
|
---
|
|
|
|
# GCN-LABEL: name: constant_fold_lshl_or_reg0_immreg_immreg{{$}}
|
|
# GCN: %3:vgpr_32 = V_MOV_B32_e32 12345, implicit $exec
|
|
# GCN-NEXT: S_ENDPGM 0, implicit %3
|
|
|
|
name: constant_fold_lshl_or_reg0_immreg_immreg
|
|
alignment: 1
|
|
exposesReturnsTwice: false
|
|
legalized: false
|
|
regBankSelected: false
|
|
selected: false
|
|
tracksRegLiveness: true
|
|
|
|
body: |
|
|
bb.0:
|
|
|
|
%0:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
|
|
%1:vgpr_32 = V_MOV_B32_e32 16, implicit $exec
|
|
%2:vgpr_32 = V_MOV_B32_e32 12345, implicit $exec
|
|
%3:vgpr_32 = V_LSHL_OR_B32 %0, %1, %2, implicit $exec
|
|
S_ENDPGM 0, implicit %3
|
|
|
|
...
|