[CodeGen][NPM] Port InitUndef to NPM #138495
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This was referenced May 5, 2025
This was referenced May 5, 2025
|
@llvm/pr-subscribers-backend-amdgpu @llvm/pr-subscribers-backend-aarch64 Author: Akshat Oke (optimisan) ChangesPatch is 24.47 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/138495.diff 12 Files Affected:
diff --git a/llvm/include/llvm/CodeGen/InitUndef.h b/llvm/include/llvm/CodeGen/InitUndef.h
new file mode 100644
index 0000000000000..be1cf4bfc9872
--- /dev/null
+++ b/llvm/include/llvm/CodeGen/InitUndef.h
@@ -0,0 +1,24 @@
+//===- llvm/CodeGen/InitUndef.h ---------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_CODEGEN_INITUNDEF_H
+#define LLVM_CODEGEN_INITUNDEF_H
+
+#include "llvm/CodeGen/MachinePassManager.h"
+
+namespace llvm {
+
+class InitUndefPass : public PassInfoMixin<InitUndefPass> {
+public:
+ PreservedAnalyses run(MachineFunction &MF,
+ MachineFunctionAnalysisManager &MFAM);
+};
+
+} // namespace llvm
+
+#endif // LLVM_CODEGEN_INITUNDEF_H
diff --git a/llvm/include/llvm/InitializePasses.h b/llvm/include/llvm/InitializePasses.h
index bff0526d4177a..07dc86c6fccf2 100644
--- a/llvm/include/llvm/InitializePasses.h
+++ b/llvm/include/llvm/InitializePasses.h
@@ -311,7 +311,7 @@ void initializeTargetTransformInfoWrapperPassPass(PassRegistry &);
void initializeTwoAddressInstructionLegacyPassPass(PassRegistry &);
void initializeTypeBasedAAWrapperPassPass(PassRegistry &);
void initializeTypePromotionLegacyPass(PassRegistry &);
-void initializeInitUndefPass(PassRegistry &);
+void initializeInitUndefLegacyPass(PassRegistry &);
void initializeUniformityInfoWrapperPassPass(PassRegistry &);
void initializeUnifyLoopExitsLegacyPassPass(PassRegistry &);
void initializeUnpackMachineBundlesPass(PassRegistry &);
diff --git a/llvm/include/llvm/Passes/CodeGenPassBuilder.h b/llvm/include/llvm/Passes/CodeGenPassBuilder.h
index 982bb16e71eab..351ef63af05c0 100644
--- a/llvm/include/llvm/Passes/CodeGenPassBuilder.h
+++ b/llvm/include/llvm/Passes/CodeGenPassBuilder.h
@@ -43,6 +43,7 @@
#include "llvm/CodeGen/GlobalMerge.h"
#include "llvm/CodeGen/GlobalMergeFunctions.h"
#include "llvm/CodeGen/IndirectBrExpand.h"
+#include "llvm/CodeGen/InitUndef.h"
#include "llvm/CodeGen/InterleavedAccess.h"
#include "llvm/CodeGen/InterleavedLoadCombine.h"
#include "llvm/CodeGen/JMCInstrumenter.h"
diff --git a/llvm/include/llvm/Passes/MachinePassRegistry.def b/llvm/include/llvm/Passes/MachinePassRegistry.def
index c69573ee3ed97..436b26852ce90 100644
--- a/llvm/include/llvm/Passes/MachinePassRegistry.def
+++ b/llvm/include/llvm/Passes/MachinePassRegistry.def
@@ -148,6 +148,7 @@ MACHINE_FUNCTION_PASS("early-tailduplication", EarlyTailDuplicatePass())
MACHINE_FUNCTION_PASS("fentry-insert", FEntryInserterPass())
MACHINE_FUNCTION_PASS("finalize-isel", FinalizeISelPass())
MACHINE_FUNCTION_PASS("fixup-statepoint-caller-saved", FixupStatepointCallerSavedPass())
+MACHINE_FUNCTION_PASS("init-undef", InitUndefPass())
MACHINE_FUNCTION_PASS("localstackalloc", LocalStackSlotAllocationPass())
MACHINE_FUNCTION_PASS("machine-cp", MachineCopyPropagationPass())
MACHINE_FUNCTION_PASS("machine-cse", MachineCSEPass())
@@ -304,7 +305,6 @@ DUMMY_MACHINE_FUNCTION_PASS("fs-profile-loader", MIRProfileLoaderNewPass)
DUMMY_MACHINE_FUNCTION_PASS("funclet-layout", FuncletLayoutPass)
DUMMY_MACHINE_FUNCTION_PASS("gc-empty-basic-blocks", GCEmptyBasicBlocksPass)
DUMMY_MACHINE_FUNCTION_PASS("implicit-null-checks", ImplicitNullChecksPass)
-DUMMY_MACHINE_FUNCTION_PASS("init-undef-pass", InitUndefPass)
DUMMY_MACHINE_FUNCTION_PASS("instruction-select", InstructionSelectPass)
DUMMY_MACHINE_FUNCTION_PASS("irtranslator", IRTranslatorPass)
DUMMY_MACHINE_FUNCTION_PASS("kcfi", MachineKCFIPass)
diff --git a/llvm/lib/CodeGen/CodeGen.cpp b/llvm/lib/CodeGen/CodeGen.cpp
index 5250534d8a4e4..aa3591cb6be58 100644
--- a/llvm/lib/CodeGen/CodeGen.cpp
+++ b/llvm/lib/CodeGen/CodeGen.cpp
@@ -54,7 +54,7 @@ void llvm::initializeCodeGen(PassRegistry &Registry) {
initializeIfConverterPass(Registry);
initializeImplicitNullChecksPass(Registry);
initializeIndirectBrExpandLegacyPassPass(Registry);
- initializeInitUndefPass(Registry);
+ initializeInitUndefLegacyPass(Registry);
initializeInterleavedLoadCombinePass(Registry);
initializeInterleavedAccessPass(Registry);
initializeJMCInstrumenterPass(Registry);
diff --git a/llvm/lib/CodeGen/InitUndef.cpp b/llvm/lib/CodeGen/InitUndef.cpp
index 6c0e9f9e930b9..500a73be7c0f5 100644
--- a/llvm/lib/CodeGen/InitUndef.cpp
+++ b/llvm/lib/CodeGen/InitUndef.cpp
@@ -38,6 +38,7 @@
//
//===----------------------------------------------------------------------===//
+#include "llvm/CodeGen/InitUndef.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/CodeGen/DetectDeadLanes.h"
@@ -59,20 +60,12 @@ using namespace llvm;
namespace {
-class InitUndef : public MachineFunctionPass {
- const TargetInstrInfo *TII;
- MachineRegisterInfo *MRI;
- const TargetSubtargetInfo *ST;
- const TargetRegisterInfo *TRI;
-
- // Newly added vregs, assumed to be fully rewritten
- SmallSet<Register, 8> NewRegs;
- SmallVector<MachineInstr *, 8> DeadInsts;
-
+class InitUndefLegacy : public MachineFunctionPass {
public:
static char ID;
- InitUndef() : MachineFunctionPass(ID) {}
+ InitUndefLegacy() : MachineFunctionPass(ID) {}
+
bool runOnMachineFunction(MachineFunction &MF) override;
void getAnalysisUsage(AnalysisUsage &AU) const override {
@@ -81,6 +74,20 @@ class InitUndef : public MachineFunctionPass {
}
StringRef getPassName() const override { return INIT_UNDEF_NAME; }
+};
+
+class InitUndef {
+ const TargetInstrInfo *TII;
+ MachineRegisterInfo *MRI;
+ const TargetSubtargetInfo *ST;
+ const TargetRegisterInfo *TRI;
+
+ // Newly added vregs, assumed to be fully rewritten
+ SmallSet<Register, 8> NewRegs;
+ SmallVector<MachineInstr *, 8> DeadInsts;
+
+public:
+ bool run(MachineFunction &MF);
private:
bool processBasicBlock(MachineFunction &MF, MachineBasicBlock &MBB,
@@ -93,9 +100,9 @@ class InitUndef : public MachineFunctionPass {
} // end anonymous namespace
-char InitUndef::ID = 0;
-INITIALIZE_PASS(InitUndef, DEBUG_TYPE, INIT_UNDEF_NAME, false, false)
-char &llvm::InitUndefID = InitUndef::ID;
+char InitUndefLegacy::ID = 0;
+INITIALIZE_PASS(InitUndefLegacy, DEBUG_TYPE, INIT_UNDEF_NAME, false, false)
+char &llvm::InitUndefID = InitUndefLegacy::ID;
static bool isEarlyClobberMI(MachineInstr &MI) {
return llvm::any_of(MI.all_defs(), [](const MachineOperand &DefMO) {
@@ -246,7 +253,20 @@ bool InitUndef::processBasicBlock(MachineFunction &MF, MachineBasicBlock &MBB,
return Changed;
}
-bool InitUndef::runOnMachineFunction(MachineFunction &MF) {
+bool InitUndefLegacy::runOnMachineFunction(MachineFunction &MF) {
+ return InitUndef().run(MF);
+}
+
+PreservedAnalyses InitUndefPass::run(MachineFunction &MF,
+ MachineFunctionAnalysisManager &MFAM) {
+ if (!InitUndef().run(MF))
+ return PreservedAnalyses::all();
+ auto PA = getMachineFunctionPassPreservedAnalyses();
+ PA.preserveSet<CFGAnalyses>();
+ return PA;
+}
+
+bool InitUndef::run(MachineFunction &MF) {
ST = &MF.getSubtarget();
// The pass is only needed if early-clobber defs and undef ops cannot be
diff --git a/llvm/lib/Passes/PassBuilder.cpp b/llvm/lib/Passes/PassBuilder.cpp
index 7740f622ede7c..5f7ce13ad8a3e 100644
--- a/llvm/lib/Passes/PassBuilder.cpp
+++ b/llvm/lib/Passes/PassBuilder.cpp
@@ -103,6 +103,7 @@
#include "llvm/CodeGen/GlobalMergeFunctions.h"
#include "llvm/CodeGen/HardwareLoops.h"
#include "llvm/CodeGen/IndirectBrExpand.h"
+#include "llvm/CodeGen/InitUndef.h"
#include "llvm/CodeGen/InterleavedAccess.h"
#include "llvm/CodeGen/InterleavedLoadCombine.h"
#include "llvm/CodeGen/JMCInstrumenter.h"
diff --git a/llvm/test/CodeGen/AArch64/init-undef.mir b/llvm/test/CodeGen/AArch64/init-undef.mir
index c9d23006d3523..cee41c053251d 100644
--- a/llvm/test/CodeGen/AArch64/init-undef.mir
+++ b/llvm/test/CodeGen/AArch64/init-undef.mir
@@ -2,6 +2,9 @@
# RUN: llc -mtriple=aarch64-- -aarch64-enable-subreg-liveness-tracking=false -run-pass=init-undef -o - %s | FileCheck %s
# RUN: llc -mtriple=aarch64-- -aarch64-enable-subreg-liveness-tracking=true -run-pass=init-undef -o - %s | FileCheck %s
+# RUN: llc -mtriple=aarch64-- -aarch64-enable-subreg-liveness-tracking=false -passes=init-undef -o - %s | FileCheck %s
+# RUN: llc -mtriple=aarch64-- -aarch64-enable-subreg-liveness-tracking=true -passes=init-undef -o - %s | FileCheck %s
+
---
name: test_stxp_undef
body: |
diff --git a/llvm/test/CodeGen/AMDGPU/llc-pipeline-npm.ll b/llvm/test/CodeGen/AMDGPU/llc-pipeline-npm.ll
index 91c15565762de..e00b7ff83e322 100644
--- a/llvm/test/CodeGen/AMDGPU/llc-pipeline-npm.ll
+++ b/llvm/test/CodeGen/AMDGPU/llc-pipeline-npm.ll
@@ -9,9 +9,9 @@
; GCN-O0: require<MachineModuleAnalysis>,require<profile-summary>,require<collector-metadata>,pre-isel-intrinsic-lowering,function(expand-large-div-rem,expand-fp),amdgpu-remove-incompatible-functions,amdgpu-printf-runtime-binding,amdgpu-lower-ctor-dtor,expand-variadics,amdgpu-always-inline,always-inline,amdgpu-export-kernel-runtime-handles,amdgpu-sw-lower-lds,amdgpu-lower-module-lds,function(atomic-expand,verify,gc-lowering,lower-constant-intrinsics,UnreachableBlockElimPass,ee-instrument<post-inline>,scalarize-masked-mem-intrin,ExpandReductionsPass,amdgpu-lower-kernel-arguments),amdgpu-lower-buffer-fat-pointers,cgscc(function(lower-switch,lower-invoke,UnreachableBlockElimPass,amdgpu-unify-divergent-exit-nodes,fix-irreducible,unify-loop-exits,StructurizeCFGPass,amdgpu-annotate-uniform,si-annotate-control-flow,amdgpu-rewrite-undef-for-phi,lcssa,require<uniformity>,callbr-prepare,safe-stack,stack-protector,verify)),cgscc(function(machine-function(amdgpu-isel,si-fix-sgpr-copies,si-i1-copies,finalize-isel,localstackalloc,phi-node-elimination,two-address-instruction,regallocfast,si-fix-vgpr-copies,remove-redundant-debug-values,fixup-statepoint-caller-saved,prolog-epilog,post-ra-pseudos,fentry-insert,xray-instrumentation,patchable-function,si-memory-legalizer,si-insert-waitcnts,si-late-branch-lowering,post-RA-hazard-rec,AMDGPUWaitSGPRHazardsPass,branch-relaxation,remove-loads-into-fake-uses,live-debug-values,machine-sanmd,stack-frame-layout,verify),invalidate<machine-function-info>))
-; GCN-O2: require<MachineModuleAnalysis>,require<profile-summary>,require<collector-metadata>,pre-isel-intrinsic-lowering,function(expand-large-div-rem,expand-fp),amdgpu-remove-incompatible-functions,amdgpu-printf-runtime-binding,amdgpu-lower-ctor-dtor,function(amdgpu-image-intrinsic-opt),expand-variadics,amdgpu-always-inline,always-inline,amdgpu-export-kernel-runtime-handles,amdgpu-sw-lower-lds,amdgpu-lower-module-lds,function(infer-address-spaces,amdgpu-atomic-optimizer,atomic-expand,amdgpu-promote-alloca,separate-const-offset-from-gep<>,slsr,early-cse<>,nary-reassociate,early-cse<>,amdgpu-codegenprepare,verify,loop-mssa(loop-reduce),mergeicmps,expand-memcmp,gc-lowering,lower-constant-intrinsics,UnreachableBlockElimPass,consthoist,ReplaceWithVeclib,partially-inline-libcalls,ee-instrument<post-inline>,scalarize-masked-mem-intrin,ExpandReductionsPass,early-cse<>,amdgpu-lower-kernel-arguments),amdgpu-lower-buffer-fat-pointers,cgscc(function(codegenprepare,load-store-vectorizer,lower-switch,lower-invoke,UnreachableBlockElimPass,flatten-cfg,sink,amdgpu-late-codegenprepare,amdgpu-unify-divergent-exit-nodes,fix-irreducible,unify-loop-exits,StructurizeCFGPass,amdgpu-annotate-uniform,si-annotate-control-flow,amdgpu-rewrite-undef-for-phi,lcssa)),amdgpu-perf-hint,cgscc(function(require<uniformity>,callbr-prepare,safe-stack,stack-protector,verify)),cgscc(function(machine-function(amdgpu-isel,si-fix-sgpr-copies,si-i1-copies,finalize-isel,early-tailduplication,opt-phis,stack-coloring,localstackalloc,dead-mi-elimination,early-machinelicm,machine-cse,machine-sink,peephole-opt,dead-mi-elimination,si-fold-operands,gcn-dpp-combine,si-load-store-opt,si-peephole-sdwa,early-machinelicm,machine-cse,si-fold-operands,dead-mi-elimination,si-shrink-instructions,detect-dead-lanes,dead-mi-elimination,InitUndefPass,ProcessImplicitDefsPass,unreachable-mbb-elimination,require<live-vars>,si-opt-vgpr-liverange,require<machine-loops>,phi-node-elimination,si-lower-control-flow,two-address-instruction,register-coalescer,rename-independent-subregs,amdgpu-rewrite-partial-reg-uses,machine-scheduler,amdgpu-pre-ra-optimizations,si-wqm,si-optimize-exec-masking-pre-ra,si-form-memory-clauses,amdgpu-pre-ra-long-branch-reg,greedy<sgpr>,virt-reg-rewriter<no-clear-vregs>,stack-slot-coloring,si-lower-sgpr-spills,si-pre-allocate-wwm-regs,greedy<wwm>,si-lower-wwm-copies,virt-reg-rewriter<no-clear-vregs>,amdgpu-reserve-wwm-regs,greedy<vgpr>,amdgpu-nsa-reassign,virt-reg-rewriter,amdgpu-mark-last-scratch-load,machine-cp,machinelicm,si-fix-vgpr-copies,si-optimize-exec-masking,remove-redundant-debug-values,fixup-statepoint-caller-saved,PostRAMachineSinkingPass,shrink-wrap,prolog-epilog,branch-folder,tailduplication,machine-latecleanup,machine-cp,post-ra-pseudos,postmisched,block-placement,fentry-insert,xray-instrumentation,patchable-function,gcn-create-vopd,si-memory-legalizer,si-insert-waitcnts,si-late-branch-lowering,si-pre-emit-peephole,post-RA-hazard-rec,AMDGPUWaitSGPRHazardsPass,amdgpu-insert-delay-alu,branch-relaxation,remove-loads-into-fake-uses,live-debug-values,machine-sanmd,stack-frame-layout,verify),invalidate<machine-function-info>))
+; GCN-O2: require<MachineModuleAnalysis>,require<profile-summary>,require<collector-metadata>,pre-isel-intrinsic-lowering,function(expand-large-div-rem,expand-fp),amdgpu-remove-incompatible-functions,amdgpu-printf-runtime-binding,amdgpu-lower-ctor-dtor,function(amdgpu-image-intrinsic-opt),expand-variadics,amdgpu-always-inline,always-inline,amdgpu-export-kernel-runtime-handles,amdgpu-sw-lower-lds,amdgpu-lower-module-lds,function(infer-address-spaces,amdgpu-atomic-optimizer,atomic-expand,amdgpu-promote-alloca,separate-const-offset-from-gep<>,slsr,early-cse<>,nary-reassociate,early-cse<>,amdgpu-codegenprepare,verify,loop-mssa(loop-reduce),mergeicmps,expand-memcmp,gc-lowering,lower-constant-intrinsics,UnreachableBlockElimPass,consthoist,ReplaceWithVeclib,partially-inline-libcalls,ee-instrument<post-inline>,scalarize-masked-mem-intrin,ExpandReductionsPass,early-cse<>,amdgpu-lower-kernel-arguments),amdgpu-lower-buffer-fat-pointers,cgscc(function(codegenprepare,load-store-vectorizer,lower-switch,lower-invoke,UnreachableBlockElimPass,flatten-cfg,sink,amdgpu-late-codegenprepare,amdgpu-unify-divergent-exit-nodes,fix-irreducible,unify-loop-exits,StructurizeCFGPass,amdgpu-annotate-uniform,si-annotate-control-flow,amdgpu-rewrite-undef-for-phi,lcssa)),amdgpu-perf-hint,cgscc(function(require<uniformity>,callbr-prepare,safe-stack,stack-protector,verify)),cgscc(function(machine-function(amdgpu-isel,si-fix-sgpr-copies,si-i1-copies,finalize-isel,early-tailduplication,opt-phis,stack-coloring,localstackalloc,dead-mi-elimination,early-machinelicm,machine-cse,machine-sink,peephole-opt,dead-mi-elimination,si-fold-operands,gcn-dpp-combine,si-load-store-opt,si-peephole-sdwa,early-machinelicm,machine-cse,si-fold-operands,dead-mi-elimination,si-shrink-instructions,detect-dead-lanes,dead-mi-elimination,init-undef,ProcessImplicitDefsPass,unreachable-mbb-elimination,require<live-vars>,si-opt-vgpr-liverange,require<machine-loops>,phi-node-elimination,si-lower-control-flow,two-address-instruction,register-coalescer,rename-independent-subregs,amdgpu-rewrite-partial-reg-uses,machine-scheduler,amdgpu-pre-ra-optimizations,si-wqm,si-optimize-exec-masking-pre-ra,si-form-memory-clauses,amdgpu-pre-ra-long-branch-reg,greedy<sgpr>,virt-reg-rewriter<no-clear-vregs>,stack-slot-coloring,si-lower-sgpr-spills,si-pre-allocate-wwm-regs,greedy<wwm>,si-lower-wwm-copies,virt-reg-rewriter<no-clear-vregs>,amdgpu-reserve-wwm-regs,greedy<vgpr>,amdgpu-nsa-reassign,virt-reg-rewriter,amdgpu-mark-last-scratch-load,machine-cp,machinelicm,si-fix-vgpr-copies,si-optimize-exec-masking,remove-redundant-debug-values,fixup-statepoint-caller-saved,PostRAMachineSinkingPass,shrink-wrap,prolog-epilog,branch-folder,tailduplication,machine-latecleanup,machine-cp,post-ra-pseudos,postmisched,block-placement,fentry-insert,xray-instrumentation,patchable-function,gcn-create-vopd,si-memory-legalizer,si-insert-waitcnts,si-late-branch-lowering,si-pre-emit-peephole,post-RA-hazard-rec,AMDGPUWaitSGPRHazardsPass,amdgpu-insert-delay-alu,branch-relaxation,remove-loads-into-fake-uses,live-debug-values,machine-sanmd,stack-frame-layout,verify),invalidate<machine-function-info>))
-; GCN-O3: require<MachineModuleAnalysis>,require<profile-summary>,require<collector-metadata>,pre-isel-intrinsic-lowering,function(expand-large-div-rem,expand-fp),amdgpu-remove-incompatible-functions,amdgpu-printf-runtime-binding,amdgpu-lower-ctor-dtor,function(amdgpu-image-intrinsic-opt),expand-variadics,amdgpu-always-inline,always-inline,amdgpu-export-kernel-runtime-handles,amdgpu-sw-lower-lds,amdgpu-lower-module-lds,function(infer-address-spaces,amdgpu-atomic-optimizer,atomic-expand,amdgpu-promote-alloca,separate-const-offset-from-gep<>,slsr,gvn<>,nary-reassociate,early-cse<>,amdgpu-codegenprepare,verify,loop-mssa(loop-reduce),mergeicmps,expand-memcmp,gc-lowering,lower-constant-intrinsics,UnreachableBlockElimPass,consthoist,ReplaceWithVeclib,partially-inline-libcalls,ee-instrument<post-inline>,scalarize-masked-mem-intrin,ExpandReductionsPass,gvn<>,amdgpu-lower-kernel-arguments),amdgpu-lower-buffer-fat-pointers,cgscc(function(codegenprepare,load-store-vectorizer,lower-switch,lower-invoke,UnreachableBlockElimPass,flatten-cfg,sink,amdgpu-late-codegenprepare,amdgpu-unify-divergent-exit-nodes,fix-irreducible,unify-loop-exits,StructurizeCFGPass,amdgpu-annotate-uniform,si-annotate-control-flow,amdgpu-rewrite-undef-for-phi,lcssa)),amdgpu-perf-hint,cgscc(function(require<uniformity>,callbr-prepare,safe-stack,stack-protector,verify)),cgscc(function(machine-function(amdgpu-isel,si-fix-sgpr-copies,si-i1-copies,finalize-isel,early-tailduplication,opt-phis,stack-coloring,localstackalloc,dead-mi-elimination,early-machinelicm,machine-cse,machine-sink,peephole-opt,dead-mi-elimination,si-fold-operands,gcn-dpp-combine,si-load-store-opt,si-peephole-sdwa,early-machinelicm,machine-cse,si-fold-operands,dead-mi-elimination,si-shrink-instructions,detect-dead-lanes,dead-mi-elimination,InitUndefPass,ProcessImplicitDefsPass,unreachable-mbb-elimination,require<live-vars>,si-opt-vgpr-liverange,require<machine-loops>,phi-node-elimination,si-lower-control-flow,two-address-instruction,register-coalescer,rename-independent-subregs,amdgpu-rewrite-partial-reg-uses,machine-scheduler,amdgpu-pre-ra-optimizations,si-wqm,si-optimize-exec-masking-pre-ra,si-form-memory-clauses,amdgpu-pre-ra-long-branch-reg,greedy<sgpr>,virt-reg-rewriter<no-clear-vregs>,stack-slot-coloring,si-lower-sgpr-spills,si-pre-allocate-wwm-regs,greedy<wwm>,si-lower-wwm-copies,virt-reg-rewriter<no-clear-vregs>,amdgpu-reserve-wwm-regs,greedy<vgpr>,amdgpu-nsa-reassign,virt-reg-rewriter,amdgpu-mark-last-scratch-load,machine-cp,machinelicm,si-fix-vgpr-copies,si-optimize-exec-masking,remove-redundant-debug-values,fixup-statepoint-caller-saved,PostRAMachineSinkingPass,shrink-wrap,prolog-epilog,branch-folder,tailduplication,machine-latecleanup,machine-cp,post-ra-pseudos,postmisched,block-placement,fentry-insert,xray-instrumentation,patchable-function,gcn-create-vopd,si-memory-legalizer,si-insert-waitcnts,si-late-branch-lowering,si-pre-emit-peephole,post-RA-hazard-rec,AMDGPUWaitSGPRHazardsPass,amdgpu-insert-delay-alu,branch-relaxation,remove-loads-into-fake-uses,live-debug-values,machine-sanmd,stack-frame-layout,verify),invalidate<machine-function-info>))
+; GCN-O3: require<MachineModuleAnalysis>,require<profile-summary>,require<collector-metadata>,pre-isel-intrinsic-lowering,function(expand-large-div-rem,expand-fp),amdgpu-remove-incompatible-functions,amdgpu-printf-runtime-binding,amdgpu-lower-ctor-dtor,function(amdgpu-image-intrinsic-opt),expand-variadics,amdgpu-always-inline,always-inline,amdgpu-export-kernel-runtime-handles,amdgpu-sw-lower-lds,amdgpu-lower-module-lds,function(infer-address-spaces,amdgpu-atomic-optimizer,atomic-expand,amdgpu-promote-alloca,separate-const-offset-from-gep<>,slsr,gvn<>,nary-reassociate,early-cse<>,amdgpu-codegenprepare,verify,loop-mssa(loop-reduce),mergeicmps,expa...
[truncated]
|
arsenm
approved these changes
May 5, 2025
optimisan
force-pushed
the
users/optimisan/port/init-undef
branch
from
9cf4254 to
8b0a590
Compare
optimisan
force-pushed
the
users/optimisan/pb/optimized-ra-pipeline
branch
from
fec3016 to
5f81915
Compare
optimisan
force-pushed
the
users/optimisan/pb/optimized-ra-pipeline
branch
from
5f81915 to
dc9a316
Compare
optimisan
force-pushed
the
users/optimisan/port/init-undef
branch
from
8b0a590 to
4c9a533
Compare
This was referenced May 12, 2025
optimisan
force-pushed
the
users/optimisan/pb/optimized-ra-pipeline
branch
from
dc9a316 to
3d1996a
Compare
optimisan
force-pushed
the
users/optimisan/port/init-undef
branch
from
4c9a533 to
86719da
Compare
optimisan
force-pushed
the
users/optimisan/pb/optimized-ra-pipeline
branch
from
3d1996a to
a013bbb
Compare
optimisan
force-pushed
the
users/optimisan/port/init-undef
branch
from
86719da to
4f07154
Compare
optimisan
force-pushed
the
users/optimisan/port/init-undef
branch
from
4f07154 to
9aa8fa5
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.