caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
8564e2fea559c58fecab3c7c01acf498bbe7820a
clang-p2996/llvm/test/CodeGen/AMDGPU/wqm.mir
Carl Ritson 976f3b3c9e [AMDGPU] Only allow implicit WQM in pixel shaders 
Implicit derivatives are only valid in pixel shaders,
hence only implicitly enable WQM for pixel shaders.
This avoids unintended WQM in other shader types (e.g. compute)
when image sampling instructions are used.

Reviewed By: rampitec

Differential Revision: https://reviews.llvm.org/D114414
2021-11-24 20:04:42 +09:00

446 lines
14 KiB
YAML
Raw Blame History

# RUN: llc -march=amdgcn -mcpu=fiji -verify-machineinstrs -run-pass si-wqm -o - %s | FileCheck %s
--- |
define amdgpu_ps void @test_strict_wwm_scc() {
ret void
}
define amdgpu_ps void @test_strict_wwm_scc2() {
ret void
}
define amdgpu_ps void @no_cfg() {
ret void
}
define amdgpu_ps void @copy_exec() {
ret void
}
define amdgpu_ps void @scc_always_live() {
ret void
}
define amdgpu_ps void @test_wwm_set_inactive_propagation() {
ret void
}
define amdgpu_ps void @test_wqm_lr_phi() {
ret void
}
define amdgpu_cs void @no_wqm_in_cs() {
ret void
}
define amdgpu_es void @no_wqm_in_es() {
ret void
}
define amdgpu_gs void @no_wqm_in_gs() {
ret void
}
define amdgpu_hs void @no_wqm_in_hs() {
ret void
}
define amdgpu_ls void @no_wqm_in_ls() {
ret void
}
define amdgpu_vs void @no_wqm_in_vs() {
ret void
}
...
---
---
# Check for awareness that s_or_saveexec_b64 clobbers SCC
#
#CHECK: ENTER_STRICT_WWM
#CHECK: S_CMP_LT_I32
#CHECK: S_CSELECT_B32
name: test_strict_wwm_scc
alignment: 1
exposesReturnsTwice: false
legalized: false
regBankSelected: false
selected: false
tracksRegLiveness: true
registers:
- { id: 0, class: sgpr_32, preferred-register: '' }
- { id: 1, class: sgpr_32, preferred-register: '' }
- { id: 2, class: sgpr_32, preferred-register: '' }
- { id: 3, class: vgpr_32, preferred-register: '' }
- { id: 4, class: vgpr_32, preferred-register: '' }
- { id: 5, class: sgpr_32, preferred-register: '' }
- { id: 6, class: vgpr_32, preferred-register: '' }
- { id: 7, class: vgpr_32, preferred-register: '' }
- { id: 8, class: sreg_32_xm0, preferred-register: '' }
- { id: 9, class: sreg_32, preferred-register: '' }
- { id: 10, class: sreg_32, preferred-register: '' }
- { id: 11, class: vgpr_32, preferred-register: '' }
- { id: 12, class: vgpr_32, preferred-register: '' }
liveins:
- { reg: '$sgpr0', virtual-reg: '%0' }
- { reg: '$sgpr1', virtual-reg: '%1' }
- { reg: '$sgpr2', virtual-reg: '%2' }
- { reg: '$vgpr0', virtual-reg: '%3' }
body: |
bb.0:
liveins: $sgpr0, $sgpr1, $sgpr2, $vgpr0
%3 = COPY $vgpr0
%2 = COPY $sgpr2
%1 = COPY $sgpr1
%0 = COPY $sgpr0
S_CMP_LT_I32 0, %0, implicit-def $scc
%12 = V_ADD_CO_U32_e32 %3, %3, implicit-def $vcc, implicit $exec
%5 = S_CSELECT_B32 %2, %1, implicit $scc
%11 = V_ADD_CO_U32_e32 %5, %12, implicit-def $vcc, implicit $exec
$vgpr0 = STRICT_WWM %11, implicit $exec
SI_RETURN_TO_EPILOG $vgpr0
...
---
# Second test for awareness that s_or_saveexec_b64 clobbers SCC
# Because entry block is treated differently.
#
#CHECK: %bb.1
#CHECK: S_CMP_LT_I32
#CHECK: COPY $scc
#CHECK: ENTER_STRICT_WWM
#CHECK: $scc = COPY
#CHECK: S_CSELECT_B32
name: test_strict_wwm_scc2
tracksRegLiveness: true
body: |
bb.0:
liveins: $sgpr0, $sgpr1, $sgpr2, $vgpr0
%3:vgpr_32 = COPY $vgpr0
%2:sgpr_32 = COPY $sgpr2
%1:sgpr_32 = COPY $sgpr1
%0:sgpr_32 = COPY $sgpr0
%13:sgpr_128 = IMPLICIT_DEF
bb.1:
S_CMP_LT_I32 0, %0:sgpr_32, implicit-def $scc
%10:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN %3:vgpr_32, %13:sgpr_128, 0, 0, 0, 0, 0, implicit $exec
%12:vgpr_32 = V_ADD_CO_U32_e32 %3:vgpr_32, %3:vgpr_32, implicit-def $vcc, implicit $exec
%5:sgpr_32 = S_CSELECT_B32 %2:sgpr_32, %1:sgpr_32, implicit $scc
%11:vgpr_32 = V_ADD_CO_U32_e32 %5:sgpr_32, %12:vgpr_32, implicit-def $vcc, implicit $exec
$vgpr0 = STRICT_WWM %11:vgpr_32, implicit $exec
$vgpr1 = COPY %10:vgpr_32
SI_RETURN_TO_EPILOG $vgpr0, $vgpr1
...
---
# V_SET_INACTIVE, when its second operand is undef, is replaced by a
# COPY by si-wqm. Ensure the instruction is removed.
#CHECK-NOT: V_SET_INACTIVE
name: no_cfg
alignment: 1
exposesReturnsTwice: false
legalized: false
regBankSelected: false
selected: false
failedISel: false
tracksRegLiveness: true
hasWinCFI: false
registers:
- { id: 0, class: sgpr_32, preferred-register: '' }
- { id: 1, class: sgpr_32, preferred-register: '' }
- { id: 2, class: sgpr_32, preferred-register: '' }
- { id: 3, class: sgpr_32, preferred-register: '' }
- { id: 4, class: sgpr_32, preferred-register: '' }
- { id: 5, class: sgpr_128, preferred-register: '' }
- { id: 6, class: sgpr_128, preferred-register: '' }
- { id: 7, class: sreg_32, preferred-register: '' }
- { id: 8, class: vreg_64, preferred-register: '' }
- { id: 9, class: sreg_32, preferred-register: '' }
- { id: 10, class: vgpr_32, preferred-register: '' }
- { id: 11, class: vgpr_32, preferred-register: '' }
- { id: 12, class: sreg_32, preferred-register: '' }
- { id: 13, class: vgpr_32, preferred-register: '' }
- { id: 14, class: vgpr_32, preferred-register: '' }
- { id: 15, class: vgpr_32, preferred-register: '' }
- { id: 16, class: vgpr_32, preferred-register: '' }
liveins:
- { reg: '$sgpr0', virtual-reg: '%0' }
- { reg: '$sgpr1', virtual-reg: '%1' }
- { reg: '$sgpr2', virtual-reg: '%2' }
- { reg: '$sgpr3', virtual-reg: '%3' }
body: |
bb.0:
liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3
%3:sgpr_32 = COPY $sgpr3
%2:sgpr_32 = COPY $sgpr2
%1:sgpr_32 = COPY $sgpr1
%0:sgpr_32 = COPY $sgpr0
%6:sgpr_128 = REG_SEQUENCE %0, %subreg.sub0, %1, %subreg.sub1, %2, %subreg.sub2, %3, %subreg.sub3
%5:sgpr_128 = COPY %6
%7:sreg_32 = S_MOV_B32 0
%8:vreg_64 = BUFFER_LOAD_DWORDX2_OFFSET %6, %7, 0, 0, 0, 0, implicit $exec
%16:vgpr_32 = COPY %8.sub1
%11:vgpr_32 = COPY %16
%10:vgpr_32 = V_SET_INACTIVE_B32 %11, undef %12:sreg_32, implicit $exec, implicit-def $scc
%14:vgpr_32 = COPY %7
%13:vgpr_32 = V_MOV_B32_dpp %14, killed %10, 323, 12, 15, 0, implicit $exec
early-clobber %15:vgpr_32 = STRICT_WWM killed %13, implicit $exec
BUFFER_STORE_DWORD_OFFSET_exact killed %15, %6, %7, 4, 0, 0, 0, implicit $exec
S_ENDPGM 0
...
---
# Ensure that strict_wwm is not put around an EXEC copy
#CHECK-LABEL: name: copy_exec
#CHECK: %7:sreg_64 = COPY $exec
#CHECK-NEXT: %14:sreg_64 = ENTER_STRICT_WWM -1, implicit-def $exec, implicit-def $scc, implicit $exec
#CHECK-NEXT: %8:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
#CHECK-NEXT: $exec = EXIT_STRICT_WWM %14
#CHECK-NEXT: %9:vgpr_32 = V_MBCNT_LO_U32_B32_e64 %7.sub0, 0, implicit $exec
name: copy_exec
tracksRegLiveness: true
body: |
bb.0:
liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3
%3:sgpr_32 = COPY $sgpr3
%2:sgpr_32 = COPY $sgpr2
%1:sgpr_32 = COPY $sgpr1
%0:sgpr_32 = COPY $sgpr0
%4:sgpr_128 = REG_SEQUENCE %0, %subreg.sub0, %1, %subreg.sub1, %2, %subreg.sub2, %3, %subreg.sub3
%5:sreg_32 = S_MOV_B32 0
%6:vreg_64 = BUFFER_LOAD_DWORDX2_OFFSET %4, %5, 0, 0, 0, 0, implicit $exec
%8:sreg_64 = COPY $exec
%9:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
%10:vgpr_32 = V_MBCNT_LO_U32_B32_e64 %8.sub0:sreg_64, 0, implicit $exec
%11:vgpr_32 = V_MOV_B32_dpp %9:vgpr_32, %10:vgpr_32, 312, 15, 15, 0, implicit $exec
%12:sreg_32 = V_READLANE_B32 %11:vgpr_32, 63
early-clobber %13:sreg_32 = STRICT_WWM %9:vgpr_32, implicit $exec
%14:vgpr_32 = COPY %13
BUFFER_STORE_DWORD_OFFSET_exact killed %14, %4, %5, 4, 0, 0, 0, implicit $exec
S_ENDPGM 0
...
---
# Check exit of WQM is still inserted correctly when SCC is live until block end.
# Critially this tests that compilation does not fail.
#CHECK-LABEL: name: scc_always_live
#CHECK: %8:vreg_128 = IMAGE_SAMPLE_V4_V2 %7
#CHECK-NEXT: S_CMP_EQ_U32 %2, 0, implicit-def $scc
#CHECK-NEXT: undef %9.sub0:vreg_64 = nsz arcp nofpexcept V_ADD_F32_e64
#CHECK-NEXT: %9.sub1:vreg_64 = nsz arcp nofpexcept V_MUL_F32_e32
#CHECK-NEXT: %14:sreg_32_xm0 = COPY $scc
#CHECK-NEXT: $exec = S_AND_B64 $exec, %13, implicit-def $scc
#CHECK-NEXT: $scc = COPY %14
#CHECK-NEXT: %10:vgpr_32 = nsz arcp nofpexcept V_ADD_F32_e64
#CHECK-NEXT: %11:vreg_128 = IMAGE_SAMPLE_V4_V2
#CHECK-NEXT: S_CBRANCH_SCC0 %bb.2
name: scc_always_live
tracksRegLiveness: true
body: |
bb.0:
liveins: $sgpr1, $sgpr2, $vgpr1, $vgpr2
$m0 = COPY $sgpr1
%0:vgpr_32 = COPY $vgpr1
%1:vgpr_32 = COPY $vgpr2
%8:sgpr_32 = COPY $sgpr2
%100:sgpr_256 = IMPLICIT_DEF
%101:sgpr_128 = IMPLICIT_DEF
%2:vgpr_32 = V_INTERP_P1_F32 %0:vgpr_32, 3, 2, implicit $mode, implicit $m0, implicit $exec
%3:vgpr_32 = V_INTERP_P1_F32 %1:vgpr_32, 3, 2, implicit $mode, implicit $m0, implicit $exec
undef %7.sub0:vreg_64 = COPY %2:vgpr_32
%7.sub1:vreg_64 = COPY %3:vgpr_32
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %7:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
S_CMP_EQ_U32 %8:sgpr_32, 0, implicit-def $scc
undef %5.sub0:vreg_64 = nsz arcp nofpexcept V_ADD_F32_e64 0, %4.sub0:vreg_128, 0, %3:vgpr_32, 1, 0, implicit $mode, implicit $exec
%5.sub1:vreg_64 = nsz arcp nofpexcept V_MUL_F32_e32 %2, %3, implicit $mode, implicit $exec
%6:vgpr_32 = nsz arcp nofpexcept V_ADD_F32_e64 0, %2:vgpr_32, 0, %3:vgpr_32, 1, 0, implicit $mode, implicit $exec
%9:vreg_128 = IMAGE_SAMPLE_V4_V2 %5:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
S_CBRANCH_SCC0 %bb.2, implicit $scc
bb.1:
%10:sreg_32 = S_MOV_B32 0
BUFFER_STORE_DWORD_OFFSET_exact %6:vgpr_32, %101:sgpr_128, %10:sreg_32, 4, 0, 0, 0, implicit $exec
S_ENDPGM 0
bb.2:
$vgpr0 = COPY %4.sub0:vreg_128
$vgpr1 = COPY %4.sub1:vreg_128
$vgpr2 = COPY %9.sub0:vreg_128
$vgpr3 = COPY %9.sub1:vreg_128
SI_RETURN_TO_EPILOG $vgpr0, $vgpr1, $vgpr2, $vgpr3
...
---
# Check that unnecessary instruction do not get marked for WWM
#
#CHECK-NOT: ENTER_STRICT_WWM
#CHECK: BUFFER_LOAD_DWORDX2
#CHECK-NOT: ENTER_STRICT_WWM
#CHECK: V_SET_INACTIVE_B32
#CHECK: V_SET_INACTIVE_B32
#CHECK: ENTER_STRICT_WWM
#CHECK: V_MAX
name: test_wwm_set_inactive_propagation
tracksRegLiveness: true
body: |
bb.0:
liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0
%0:sgpr_128 = COPY $sgpr0_sgpr1_sgpr2_sgpr3
%1:vgpr_32 = COPY $vgpr0
%2:vreg_64 = BUFFER_LOAD_DWORDX2_OFFEN %1:vgpr_32, %0:sgpr_128, 0, 0, 0, 0, 0, implicit $exec
%2.sub0:vreg_64 = V_SET_INACTIVE_B32 %2.sub0:vreg_64, 0, implicit $exec, implicit-def $scc
%2.sub1:vreg_64 = V_SET_INACTIVE_B32 %2.sub1:vreg_64, 0, implicit $exec, implicit-def $scc
%3:vreg_64 = nnan nsz arcp contract reassoc nofpexcept V_MAX_F64_e64 0, %2:vreg_64, 0, %2:vreg_64, 0, 0, implicit $mode, implicit $exec
$vgpr0 = STRICT_WWM %3.sub0:vreg_64, implicit $exec
$vgpr1 = STRICT_WWM %3.sub1:vreg_64, implicit $exec
SI_RETURN_TO_EPILOG $vgpr0, $vgpr1
...
---
# Check that WQM marking occurs correctly through phi nodes in live range graph.
# If not then initial V_MOV will not be in WQM.
#
#CHECK-LABEL: name: test_wqm_lr_phi
#CHECK: COPY $exec
#CHECK-NEXT: S_WQM
#CHECK-NEXT: V_MOV_B32_e32 -10
#CHECK-NEXT: V_MOV_B32_e32 0
name: test_wqm_lr_phi
tracksRegLiveness: true
body: |
bb.0:
undef %0.sub0:vreg_64 = V_MOV_B32_e32 -10, implicit $exec
%0.sub1:vreg_64 = V_MOV_B32_e32 0, implicit $exec
%1:sreg_64 = S_GETPC_B64
%2:sgpr_256 = S_LOAD_DWORDX8_IMM %1:sreg_64, 32, 0
bb.1:
$vcc = V_CMP_LT_U32_e64 4, 4, implicit $exec
S_CBRANCH_VCCNZ %bb.3, implicit $vcc
S_BRANCH %bb.2
bb.2:
%0.sub0:vreg_64 = V_ADD_U32_e32 1, %0.sub1, implicit $exec
S_BRANCH %bb.3
bb.3:
%0.sub1:vreg_64 = V_ADD_U32_e32 1, %0.sub1, implicit $exec
S_BRANCH %bb.4
bb.4:
%3:sgpr_128 = IMPLICIT_DEF
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %0:vreg_64, %2:sgpr_256, %3:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from custom "ImageResource")
$vgpr0 = COPY %4.sub0:vreg_128
$vgpr1 = COPY %4.sub1:vreg_128
SI_RETURN_TO_EPILOG $vgpr0, $vgpr1
...
---
#CHECK-LABEL: name: no_wqm_in_cs
#CHECK-NOT: S_WQM
name: no_wqm_in_cs
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr1, $vgpr2
undef %0.sub0:vreg_64 = COPY $vgpr1
%0.sub1:vreg_64 = COPY $vgpr2
%100:sgpr_256 = IMPLICIT_DEF
%101:sgpr_128 = IMPLICIT_DEF
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %0:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
...
---
#CHECK-LABEL: name: no_wqm_in_es
#CHECK-NOT: S_WQM
name: no_wqm_in_es
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr1, $vgpr2
undef %0.sub0:vreg_64 = COPY $vgpr1
%0.sub1:vreg_64 = COPY $vgpr2
%100:sgpr_256 = IMPLICIT_DEF
%101:sgpr_128 = IMPLICIT_DEF
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %0:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
...
---
#CHECK-LABEL: name: no_wqm_in_gs
#CHECK-NOT: S_WQM
name: no_wqm_in_gs
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr1, $vgpr2
undef %0.sub0:vreg_64 = COPY $vgpr1
%0.sub1:vreg_64 = COPY $vgpr2
%100:sgpr_256 = IMPLICIT_DEF
%101:sgpr_128 = IMPLICIT_DEF
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %0:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
...
---
#CHECK-LABEL: name: no_wqm_in_hs
#CHECK-NOT: S_WQM
name: no_wqm_in_hs
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr1, $vgpr2
undef %0.sub0:vreg_64 = COPY $vgpr1
%0.sub1:vreg_64 = COPY $vgpr2
%100:sgpr_256 = IMPLICIT_DEF
%101:sgpr_128 = IMPLICIT_DEF
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %0:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
...
---
#CHECK-LABEL: name: no_wqm_in_ls
#CHECK-NOT: S_WQM
name: no_wqm_in_ls
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr1, $vgpr2
undef %0.sub0:vreg_64 = COPY $vgpr1
%0.sub1:vreg_64 = COPY $vgpr2
%100:sgpr_256 = IMPLICIT_DEF
%101:sgpr_128 = IMPLICIT_DEF
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %0:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
...
---
#CHECK-LABEL: name: no_wqm_in_vs
#CHECK-NOT: S_WQM
name: no_wqm_in_vs
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr1, $vgpr2
undef %0.sub0:vreg_64 = COPY $vgpr1
%0.sub1:vreg_64 = COPY $vgpr2
%100:sgpr_256 = IMPLICIT_DEF
%101:sgpr_128 = IMPLICIT_DEF
%4:vreg_128 = IMAGE_SAMPLE_V4_V2 %0:vreg_64, %100:sgpr_256, %101:sgpr_128, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
...
Reference in New Issue View Git Blame Copy Permalink