maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[AMDGPU] Add support for in-order bvh in waitcnt pass 

bvh should be handled separately from vmem and vmem with sampler instructions
for waitcnt handling.

Differential Revision: https://reviews.llvm.org/D114794
This commit is contained in:
David Stuttard 2021-11-30 09:29:14 +00:00
parent 17a3385c35
commit 0e8590f065
2 changed files with 8 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
View File

@ -150,6 +150,8 @@ enum VmemType {
VMEM_NOSAMPLER, VMEM_NOSAMPLER,
// MIMG instructions with a sampler. // MIMG instructions with a sampler.
VMEM_SAMPLER, VMEM_SAMPLER,
// BVH instructions
VMEM_BVH
}; };
VmemType getVmemType(const MachineInstr &Inst) { VmemType getVmemType(const MachineInstr &Inst) {
@ -157,9 +159,10 @@ VmemType getVmemType(const MachineInstr &Inst) {
if (!SIInstrInfo::isMIMG(Inst)) if (!SIInstrInfo::isMIMG(Inst))
return VMEM_NOSAMPLER; return VMEM_NOSAMPLER;
const AMDGPU::MIMGInfo *Info = AMDGPU::getMIMGInfo(Inst.getOpcode()); const AMDGPU::MIMGInfo *Info = AMDGPU::getMIMGInfo(Inst.getOpcode());
return AMDGPU::getMIMGBaseOpcodeInfo(Info->BaseOpcode)->Sampler const AMDGPU::MIMGBaseOpcodeInfo *BaseInfo =
? VMEM_SAMPLER AMDGPU::getMIMGBaseOpcodeInfo(Info->BaseOpcode);
: VMEM_NOSAMPLER; return BaseInfo->BVH ? VMEM_BVH
: BaseInfo->Sampler ? VMEM_SAMPLER : VMEM_NOSAMPLER;
} }
void addWait(AMDGPU::Waitcnt &Wait, InstCounterType T, unsigned Count) { void addWait(AMDGPU::Waitcnt &Wait, InstCounterType T, unsigned Count) {

2
llvm/test/CodeGen/AMDGPU/waitcnt-bvh.mir
View File

@ -23,6 +23,7 @@ body: |
; GCN: S_WAITCNT 0 ; GCN: S_WAITCNT 0
; GCN-NEXT: S_WAITCNT_VSCNT undef $sgpr_null, 0 ; GCN-NEXT: S_WAITCNT_VSCNT undef $sgpr_null, 0
; GCN-NEXT: $vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_BVH_INTERSECT_RAY_sa_gfx10 killed $vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19, renamable $sgpr0_sgpr1_sgpr2_sgpr3, implicit $exec :: (dereferenceable load (s128) from custom "ImageResource") ; GCN-NEXT: $vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_BVH_INTERSECT_RAY_sa_gfx10 killed $vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19, renamable $sgpr0_sgpr1_sgpr2_sgpr3, implicit $exec :: (dereferenceable load (s128) from custom "ImageResource")
; GCN-NEXT: S_WAITCNT 16240
; GCN-NEXT: $vgpr0 = BUFFER_LOAD_DWORD_OFFEN $vgpr16, $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, 0, 0, 0, 0, implicit $exec ; GCN-NEXT: $vgpr0 = BUFFER_LOAD_DWORD_OFFEN $vgpr16, $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, 0, 0, 0, 0, implicit $exec
; GCN-NEXT: S_ENDPGM 0 ; GCN-NEXT: S_ENDPGM 0
$vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_BVH_INTERSECT_RAY_sa_gfx10 killed $vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19, renamable $sgpr0_sgpr1_sgpr2_sgpr3, implicit $exec :: (dereferenceable load (s128) from custom "ImageResource") $vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_BVH_INTERSECT_RAY_sa_gfx10 killed $vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19, renamable $sgpr0_sgpr1_sgpr2_sgpr3, implicit $exec :: (dereferenceable load (s128) from custom "ImageResource")
@ -52,6 +53,7 @@ body: |
; GCN: S_WAITCNT 0 ; GCN: S_WAITCNT 0
; GCN-NEXT: S_WAITCNT_VSCNT undef $sgpr_null, 0 ; GCN-NEXT: S_WAITCNT_VSCNT undef $sgpr_null, 0
; GCN-NEXT: $vgpr0 = BUFFER_LOAD_DWORD_OFFEN $vgpr20, $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, 0, 0, 0, 0, implicit $exec ; GCN-NEXT: $vgpr0 = BUFFER_LOAD_DWORD_OFFEN $vgpr20, $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, 0, 0, 0, 0, implicit $exec
; GCN-NEXT: S_WAITCNT 16240
; GCN-NEXT: $vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_BVH_INTERSECT_RAY_sa_gfx10 killed $vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3, implicit $exec :: (dereferenceable load (s128) from custom "ImageResource") ; GCN-NEXT: $vgpr0_vgpr1_vgpr2_vgpr3 = IMAGE_BVH_INTERSECT_RAY_sa_gfx10 killed $vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15_vgpr16_vgpr17_vgpr18_vgpr19, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3, implicit $exec :: (dereferenceable load (s128) from custom "ImageResource")
; GCN-NEXT: S_ENDPGM 0 ; GCN-NEXT: S_ENDPGM 0
$vgpr0 = BUFFER_LOAD_DWORD_OFFEN $vgpr20, $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, 0, 0, 0, 0, implicit $exec $vgpr0 = BUFFER_LOAD_DWORD_OFFEN $vgpr20, $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, 0, 0, 0, 0, implicit $exec