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

[AMDGPU] Increase detection range for s_mov, v_cmpx transformation. 

We found that it might be beneficial to have the SIOptimizeExecMasking
pass detect more cases where v_cmp, s_and_saveexec patterns can be
transformed to s_mov, v_cmpx patterns. Currently, the search range
for finding a fitting v_cmp instruction is 5, however, this is doubled
to 10 here.

Reviewed By: foad

Differential Revision: https://reviews.llvm.org/D123367
This commit is contained in:
Thomas Symalla 2022-04-08 09:48:23 +02:00
parent 08920cc043
commit 6d97ca690c
2 changed files with 31 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
llvm/lib/Target/AMDGPU/SIOptimizeExecMasking.cpp
View File

@ -302,12 +302,15 @@ static MachineInstr *
findInstrBackwards(MachineInstr &Origin,
std::function<bool(MachineInstr *)> Pred,
ArrayRef<MCRegister> NonModifiableRegs,
const SIRegisterInfo *TRI, unsigned MaxInstructions = 5) {
const SIRegisterInfo *TRI, unsigned MaxInstructions = 20) {
MachineBasicBlock::reverse_iterator A = Origin.getReverseIterator(),
E = Origin.getParent()->rend();
unsigned CurrentIteration = 0;
for (++A; CurrentIteration < MaxInstructions && A != E; ++A) {
if (A->isDebugInstr())
continue;
if (Pred(&*A))
return &*A;
@ -315,7 +318,7 @@ findInstrBackwards(MachineInstr &Origin,
if (A->modifiesRegister(Reg, TRI))
return nullptr;
}
++CurrentIteration;
}

27
llvm/test/CodeGen/AMDGPU/vcmp-saveexec-to-vcmpx.mir
View File

@ -1,6 +1,5 @@
# RUN: llc -march=amdgcn -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 -run-pass=si-optimize-exec-masking -verify-machineinstrs %s -o - | FileCheck --check-prefixes=GCN,GFX1010 %s
# RUN: llc -march=amdgcn -mcpu=gfx1030 -mattr=-wavefrontsize32,+wavefrontsize64 -run-pass=si-optimize-exec-masking -verify-machineinstrs %s -o - | FileCheck --check-prefixes=GCN,GFX1030 %s
---
# After the Optimize exec masking (post-RA) pass, there's a change of having v_cmpx instructions
@ -62,3 +61,29 @@ body: |
$sgpr2_sgpr3 = COPY $exec, implicit-def $exec
$sgpr2_sgpr3 = S_AND_B64 killed renamable $sgpr2_sgpr3, killed renamable $sgpr0_sgpr1, implicit-def dead $scc
$exec = S_MOV_B64_term killed renamable $sgpr2_sgpr3
...
---
# Check if the sequence will be optimized even with more than 5 (unrelated) instructions inbetween the v_cmp and s_and_saveexec.
# GCN-LABEL: name: vcmp_saveexec_to_mov_vcmpx_check_many_instrs
# GFX1010: V_CMP_LT_F32_e64
# GFX1010: S_AND_SAVEEXEC_B64
# GFX1030: S_MOV_B64
# GFX1030: V_CMPX_LT_F32_nosdst_e64 0, 953267991, 2
name: vcmp_saveexec_to_mov_vcmpx_check_many_instrs
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0, $sgpr2, $vgpr1
renamable $sgpr0_sgpr1 = V_CMP_LT_F32_e64 0, 953267991, 2, $vgpr0, 0, implicit $mode, implicit $exec
$vgpr1 = V_WRITELANE_B32 0, $sgpr2, $vgpr1
$vgpr1 = V_WRITELANE_B32 0, $sgpr2, $vgpr1
$vgpr1 = V_WRITELANE_B32 0, $sgpr2, $vgpr1
$vgpr1 = V_WRITELANE_B32 0, $sgpr2, $vgpr1
$vgpr1 = V_WRITELANE_B32 0, $sgpr2, $vgpr1
$vgpr1 = V_WRITELANE_B32 0, $sgpr2, $vgpr1
$sgpr2_sgpr3 = COPY $exec, implicit-def $exec
$sgpr2_sgpr3 = S_AND_B64 killed renamable $sgpr2_sgpr3, killed renamable $sgpr0_sgpr1, implicit-def dead $scc
$exec = S_MOV_B64_term killed renamable $sgpr2_sgpr3