[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From 8b8932b55c8a6a087d516e174e1d57c9908259bd Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 28095447f6a5a..0b849f3382f63 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -482,14 +482,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 2a7a6edda70a8..06312562060aa 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 7ff7acebedf4e..27cc2a4109879 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -384,17 +385,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [NFC] Formatting PassRegistry.def (PR #144139)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/144139
>From 7fa87f2e42378d656ba743a4971e5c2ffaee8492 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 13 Jun 2025 18:22:10 +
Subject: [PATCH] [NFC] Formatting PassRegistry.def
---
llvm/lib/Passes/PassRegistry.def | 40 ++--
1 file changed, 23 insertions(+), 17 deletions(-)
diff --git a/llvm/lib/Passes/PassRegistry.def b/llvm/lib/Passes/PassRegistry.def
index ec14c6a9211d9..5256f1378b64c 100644
--- a/llvm/lib/Passes/PassRegistry.def
+++ b/llvm/lib/Passes/PassRegistry.def
@@ -63,7 +63,8 @@ MODULE_PASS("coro-early", CoroEarlyPass())
MODULE_PASS("cross-dso-cfi", CrossDSOCFIPass())
MODULE_PASS("ctx-instr-gen",
PGOInstrumentationGen(PGOInstrumentationType::CTXPROF))
-MODULE_PASS("ctx-prof-flatten",
PGOCtxProfFlatteningPass(/*IsPreThinlink=*/false))
+MODULE_PASS("ctx-prof-flatten",
+PGOCtxProfFlatteningPass(/*IsPreThinlink=*/false))
MODULE_PASS("ctx-prof-flatten-prethinlink",
PGOCtxProfFlatteningPass(/*IsPreThinlink=*/true))
MODULE_PASS("noinline-nonprevailing", NoinlineNonPrevailing())
@@ -74,7 +75,8 @@ MODULE_PASS("dot-callgraph", CallGraphDOTPrinterPass())
MODULE_PASS("dxil-upgrade", DXILUpgradePass())
MODULE_PASS("elim-avail-extern", EliminateAvailableExternallyPass())
MODULE_PASS("extract-blocks", BlockExtractorPass({}, false))
-MODULE_PASS("expand-variadics",
ExpandVariadicsPass(ExpandVariadicsMode::Disable))
+MODULE_PASS("expand-variadics",
+ExpandVariadicsPass(ExpandVariadicsMode::Disable))
MODULE_PASS("forceattrs", ForceFunctionAttrsPass())
MODULE_PASS("function-import", FunctionImportPass())
MODULE_PASS("global-merge-func", GlobalMergeFuncPass())
@@ -104,7 +106,10 @@ MODULE_PASS("lower-ifunc", LowerIFuncPass())
MODULE_PASS("simplify-type-tests", SimplifyTypeTestsPass())
MODULE_PASS("lowertypetests", LowerTypeTestsPass())
MODULE_PASS("fatlto-cleanup", FatLtoCleanup())
-MODULE_PASS("pgo-force-function-attrs", PGOForceFunctionAttrsPass(PGOOpt ?
PGOOpt->ColdOptType : PGOOptions::ColdFuncOpt::Default))
+MODULE_PASS("pgo-force-function-attrs",
+PGOForceFunctionAttrsPass(PGOOpt
+ ? PGOOpt->ColdOptType
+ : PGOOptions::ColdFuncOpt::Default))
MODULE_PASS("memprof-context-disambiguation", MemProfContextDisambiguation())
MODULE_PASS("memprof-module", ModuleMemProfilerPass())
MODULE_PASS("mergefunc", MergeFunctionsPass())
@@ -178,7 +183,7 @@ MODULE_PASS_WITH_PARAMS(
parseASanPassOptions, "kernel")
MODULE_PASS_WITH_PARAMS(
"cg-profile", "CGProfilePass",
-[](bool InLTOPostLink) { return CGProfilePass(InLTOPostLink);},
+[](bool InLTOPostLink) { return CGProfilePass(InLTOPostLink); },
parseCGProfilePassOptions, "in-lto-post-link")
MODULE_PASS_WITH_PARAMS(
"global-merge", "GlobalMergePass",
@@ -287,7 +292,8 @@ CGSCC_PASS_WITH_PARAMS(
FUNCTION_ANALYSIS("aa", AAManager())
FUNCTION_ANALYSIS("access-info", LoopAccessAnalysis())
FUNCTION_ANALYSIS("assumptions", AssumptionAnalysis())
-FUNCTION_ANALYSIS("bb-sections-profile-reader",
BasicBlockSectionsProfileReaderAnalysis(TM))
+FUNCTION_ANALYSIS("bb-sections-profile-reader",
+ BasicBlockSectionsProfileReaderAnalysis(TM))
FUNCTION_ANALYSIS("block-freq", BlockFrequencyAnalysis())
FUNCTION_ANALYSIS("branch-prob", BranchProbabilityAnalysis())
FUNCTION_ANALYSIS("cycles", CycleAnalysis())
@@ -377,7 +383,7 @@ FUNCTION_PASS("expand-large-div-rem",
ExpandLargeDivRemPass(TM))
FUNCTION_PASS("expand-fp", ExpandFpPass(TM))
FUNCTION_PASS("expand-memcmp", ExpandMemCmpPass(TM))
FUNCTION_PASS("extra-vector-passes",
- ExtraFunctionPassManager())
+ ExtraFunctionPassManager())
FUNCTION_PASS("fix-irreducible", FixIrreduciblePass())
FUNCTION_PASS("flatten-cfg", FlattenCFGPass())
FUNCTION_PASS("float2int", Float2IntPass())
@@ -548,8 +554,7 @@ FUNCTION_PASS_WITH_PARAMS(
"max-iterations=N")
FUNCTION_PASS_WITH_PARAMS(
"lint", "LintPass",
-[](bool AbortOnError) { return LintPass(AbortOnError); },
-parseLintOptions,
+[](bool AbortOnError) { return LintPass(AbortOnError); }, parseLintOptions,
"abort-on-error")
FUNCTION_PASS_WITH_PARAMS(
"loop-unroll", "LoopUnrollPass",
@@ -576,7 +581,8 @@ FUNCTION_PASS_WITH_PARAMS(
"normalize", "IRNormalizerPass",
[](IRNormalizerOptions Options) { return IRNormalizerPass(Options); },
parseIRNormalizerPassOptions,
-
"no-preserve-order;preserve-order;no-rename-all;rename-all;no-fold-all;fold-all;no-reorder-operands;reorder-operands")
+"no-preserve-order;preserve-order;no-rename-all;rename-all;no-fold-all;"
+"fold-all;no-reorder-operands;reorder-operands")
FUNCTION_PASS_WITH_PARAMS(
"mldst-motion", "MergedLoadStoreMotionPass",
[](MergedLoadStoreMotionOptions Opts) {
@@ -590,7 +596,7 @@ FUNCTION_PASS_WITH_PARAMS(
},
[](StringRe
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From 8b8932b55c8a6a087d516e174e1d57c9908259bd Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 28095447f6a5a..0b849f3382f63 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -482,14 +482,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 2a7a6edda70a8..06312562060aa 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 7ff7acebedf4e..27cc2a4109879 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -384,17 +385,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [NFC] Formatting PassRegistry.def (PR #144139)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/144139
>From 7fa87f2e42378d656ba743a4971e5c2ffaee8492 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 13 Jun 2025 18:22:10 +
Subject: [PATCH] [NFC] Formatting PassRegistry.def
---
llvm/lib/Passes/PassRegistry.def | 40 ++--
1 file changed, 23 insertions(+), 17 deletions(-)
diff --git a/llvm/lib/Passes/PassRegistry.def b/llvm/lib/Passes/PassRegistry.def
index ec14c6a9211d9..5256f1378b64c 100644
--- a/llvm/lib/Passes/PassRegistry.def
+++ b/llvm/lib/Passes/PassRegistry.def
@@ -63,7 +63,8 @@ MODULE_PASS("coro-early", CoroEarlyPass())
MODULE_PASS("cross-dso-cfi", CrossDSOCFIPass())
MODULE_PASS("ctx-instr-gen",
PGOInstrumentationGen(PGOInstrumentationType::CTXPROF))
-MODULE_PASS("ctx-prof-flatten",
PGOCtxProfFlatteningPass(/*IsPreThinlink=*/false))
+MODULE_PASS("ctx-prof-flatten",
+PGOCtxProfFlatteningPass(/*IsPreThinlink=*/false))
MODULE_PASS("ctx-prof-flatten-prethinlink",
PGOCtxProfFlatteningPass(/*IsPreThinlink=*/true))
MODULE_PASS("noinline-nonprevailing", NoinlineNonPrevailing())
@@ -74,7 +75,8 @@ MODULE_PASS("dot-callgraph", CallGraphDOTPrinterPass())
MODULE_PASS("dxil-upgrade", DXILUpgradePass())
MODULE_PASS("elim-avail-extern", EliminateAvailableExternallyPass())
MODULE_PASS("extract-blocks", BlockExtractorPass({}, false))
-MODULE_PASS("expand-variadics",
ExpandVariadicsPass(ExpandVariadicsMode::Disable))
+MODULE_PASS("expand-variadics",
+ExpandVariadicsPass(ExpandVariadicsMode::Disable))
MODULE_PASS("forceattrs", ForceFunctionAttrsPass())
MODULE_PASS("function-import", FunctionImportPass())
MODULE_PASS("global-merge-func", GlobalMergeFuncPass())
@@ -104,7 +106,10 @@ MODULE_PASS("lower-ifunc", LowerIFuncPass())
MODULE_PASS("simplify-type-tests", SimplifyTypeTestsPass())
MODULE_PASS("lowertypetests", LowerTypeTestsPass())
MODULE_PASS("fatlto-cleanup", FatLtoCleanup())
-MODULE_PASS("pgo-force-function-attrs", PGOForceFunctionAttrsPass(PGOOpt ?
PGOOpt->ColdOptType : PGOOptions::ColdFuncOpt::Default))
+MODULE_PASS("pgo-force-function-attrs",
+PGOForceFunctionAttrsPass(PGOOpt
+ ? PGOOpt->ColdOptType
+ : PGOOptions::ColdFuncOpt::Default))
MODULE_PASS("memprof-context-disambiguation", MemProfContextDisambiguation())
MODULE_PASS("memprof-module", ModuleMemProfilerPass())
MODULE_PASS("mergefunc", MergeFunctionsPass())
@@ -178,7 +183,7 @@ MODULE_PASS_WITH_PARAMS(
parseASanPassOptions, "kernel")
MODULE_PASS_WITH_PARAMS(
"cg-profile", "CGProfilePass",
-[](bool InLTOPostLink) { return CGProfilePass(InLTOPostLink);},
+[](bool InLTOPostLink) { return CGProfilePass(InLTOPostLink); },
parseCGProfilePassOptions, "in-lto-post-link")
MODULE_PASS_WITH_PARAMS(
"global-merge", "GlobalMergePass",
@@ -287,7 +292,8 @@ CGSCC_PASS_WITH_PARAMS(
FUNCTION_ANALYSIS("aa", AAManager())
FUNCTION_ANALYSIS("access-info", LoopAccessAnalysis())
FUNCTION_ANALYSIS("assumptions", AssumptionAnalysis())
-FUNCTION_ANALYSIS("bb-sections-profile-reader",
BasicBlockSectionsProfileReaderAnalysis(TM))
+FUNCTION_ANALYSIS("bb-sections-profile-reader",
+ BasicBlockSectionsProfileReaderAnalysis(TM))
FUNCTION_ANALYSIS("block-freq", BlockFrequencyAnalysis())
FUNCTION_ANALYSIS("branch-prob", BranchProbabilityAnalysis())
FUNCTION_ANALYSIS("cycles", CycleAnalysis())
@@ -377,7 +383,7 @@ FUNCTION_PASS("expand-large-div-rem",
ExpandLargeDivRemPass(TM))
FUNCTION_PASS("expand-fp", ExpandFpPass(TM))
FUNCTION_PASS("expand-memcmp", ExpandMemCmpPass(TM))
FUNCTION_PASS("extra-vector-passes",
- ExtraFunctionPassManager())
+ ExtraFunctionPassManager())
FUNCTION_PASS("fix-irreducible", FixIrreduciblePass())
FUNCTION_PASS("flatten-cfg", FlattenCFGPass())
FUNCTION_PASS("float2int", Float2IntPass())
@@ -548,8 +554,7 @@ FUNCTION_PASS_WITH_PARAMS(
"max-iterations=N")
FUNCTION_PASS_WITH_PARAMS(
"lint", "LintPass",
-[](bool AbortOnError) { return LintPass(AbortOnError); },
-parseLintOptions,
+[](bool AbortOnError) { return LintPass(AbortOnError); }, parseLintOptions,
"abort-on-error")
FUNCTION_PASS_WITH_PARAMS(
"loop-unroll", "LoopUnrollPass",
@@ -576,7 +581,8 @@ FUNCTION_PASS_WITH_PARAMS(
"normalize", "IRNormalizerPass",
[](IRNormalizerOptions Options) { return IRNormalizerPass(Options); },
parseIRNormalizerPassOptions,
-
"no-preserve-order;preserve-order;no-rename-all;rename-all;no-fold-all;fold-all;no-reorder-operands;reorder-operands")
+"no-preserve-order;preserve-order;no-rename-all;rename-all;no-fold-all;"
+"fold-all;no-reorder-operands;reorder-operands")
FUNCTION_PASS_WITH_PARAMS(
"mldst-motion", "MergedLoadStoreMotionPass",
[](MergedLoadStoreMotionOptions Opts) {
@@ -590,7 +596,7 @@ FUNCTION_PASS_WITH_PARAMS(
},
[](StringRe
[llvm-branch-commits] [llvm] Increasing tolerance in ApproximatelyEquals (PR #145117)
https://github.com/svkeerthy created
https://github.com/llvm/llvm-project/pull/145117
None
>From d05856c47337b3b6e9086a5ee06b7c39412d9103 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 20 Jun 2025 22:56:46 +
Subject: [PATCH] Increasing tolerance in ApproximatelyEquals
---
llvm/include/llvm/Analysis/IR2Vec.h| 2 +-
llvm/unittests/Analysis/IR2VecTest.cpp | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 06312562060aa..480b834077b86 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -116,7 +116,7 @@ struct Embedding {
/// Returns true if the embedding is approximately equal to the RHS embedding
/// within the specified tolerance.
- bool approximatelyEquals(const Embedding &RHS, double Tolerance = 1e-6)
const;
+ bool approximatelyEquals(const Embedding &RHS, double Tolerance = 1e-4)
const;
void print(raw_ostream &OS) const;
};
diff --git a/llvm/unittests/Analysis/IR2VecTest.cpp
b/llvm/unittests/Analysis/IR2VecTest.cpp
index 05af55b59323b..33ac16828eb6c 100644
--- a/llvm/unittests/Analysis/IR2VecTest.cpp
+++ b/llvm/unittests/Analysis/IR2VecTest.cpp
@@ -154,14 +154,14 @@ TEST(EmbeddingTest, ApproximatelyEqual) {
EXPECT_TRUE(E1.approximatelyEquals(E2)); // Diff = 1e-7
Embedding E3 = {1.2, 2.2, 3.2}; // Diff = 2e-5
- EXPECT_FALSE(E1.approximatelyEquals(E3));
+ EXPECT_FALSE(E1.approximatelyEquals(E3, 1e-6));
EXPECT_TRUE(E1.approximatelyEquals(E3, 3e-5));
Embedding E_clearly_within = {1.005, 2.005, 3.005}; // Diff =
5e-7
EXPECT_TRUE(E1.approximatelyEquals(E_clearly_within));
Embedding E_clearly_outside = {1.1, 2.1, 3.1}; // Diff = 1e-5
- EXPECT_FALSE(E1.approximatelyEquals(E_clearly_outside));
+ EXPECT_FALSE(E1.approximatelyEquals(E_clearly_outside, 1e-6));
Embedding E4 = {1.0, 2.0, 3.5}; // Large diff
EXPECT_FALSE(E1.approximatelyEquals(E4, 0.01));
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] Overloading operator+ for Embeddngs (PR #145118)
https://github.com/svkeerthy created
https://github.com/llvm/llvm-project/pull/145118
None
>From cbd2c6e77eefb4ba7b8acbf6ea12f21486e7dbc8 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 20 Jun 2025 23:00:40 +
Subject: [PATCH] Overloading operator+ for Embeddngs
---
llvm/include/llvm/Analysis/IR2Vec.h| 1 +
llvm/lib/Analysis/IR2Vec.cpp | 8
llvm/unittests/Analysis/IR2VecTest.cpp | 18 ++
3 files changed, 27 insertions(+)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 480b834077b86..f6c40d36f8026 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -106,6 +106,7 @@ struct Embedding {
const std::vector &getData() const { return Data; }
/// Arithmetic operators
+ Embedding operator+(const Embedding &RHS) const;
Embedding &operator+=(const Embedding &RHS);
Embedding &operator-=(const Embedding &RHS);
Embedding &operator*=(double Factor);
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 27cc2a4109879..d5d27db8bd2bf 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -71,6 +71,14 @@ inline bool fromJSON(const llvm::json::Value &E, Embedding
&Out,
// Embedding
//===--===//
+Embedding Embedding::operator+(const Embedding &RHS) const {
+ assert(this->size() == RHS.size() && "Vectors must have the same dimension");
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), RHS.begin(), Result.begin(),
+ std::plus());
+ return Result;
+}
+
Embedding &Embedding::operator+=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
diff --git a/llvm/unittests/Analysis/IR2VecTest.cpp
b/llvm/unittests/Analysis/IR2VecTest.cpp
index 33ac16828eb6c..50eb7f73c6f50 100644
--- a/llvm/unittests/Analysis/IR2VecTest.cpp
+++ b/llvm/unittests/Analysis/IR2VecTest.cpp
@@ -109,6 +109,18 @@ TEST(EmbeddingTest, ConstructorsAndAccessors) {
}
}
+TEST(EmbeddingTest, AddVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 + E2;
+ EXPECT_THAT(E3, ElementsAre(1.5, 3.5, 2.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, AddVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -180,6 +192,12 @@ TEST(EmbeddingTest, AccessOutOfBounds) {
EXPECT_DEATH(E[4] = 4.0, "Index out of bounds");
}
+TEST(EmbeddingTest, MismatchedDimensionsAddVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0};
+ Embedding E2 = {1.0};
+ EXPECT_DEATH(E1 + E2, "Vectors must have the same dimension");
+}
+
TEST(EmbeddingTest, MismatchedDimensionsAddVectors) {
Embedding E1 = {1.0, 2.0};
Embedding E2 = {1.0};
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] Increasing tolerance in ApproximatelyEquals (PR #145117)
svkeerthy wrote: > [!WARNING] > This pull request is not mergeable via GitHub because a downstack PR is > open. Once all requirements are satisfied, merge this PR as a stack href="https://app.graphite.dev/github/pr/llvm/llvm-project/145117?utm_source=stack-comment-downstack-mergeability-warning"; > >on Graphite. > https://graphite.dev/docs/merge-pull-requests";>Learn more * **#145119** https://app.graphite.dev/github/pr/llvm/llvm-project/145119?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#145118** https://app.graphite.dev/github/pr/llvm/llvm-project/145118?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#145117** https://app.graphite.dev/github/pr/llvm/llvm-project/145117?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> 👈 https://app.graphite.dev/github/pr/llvm/llvm-project/145117?utm_source=stack-comment-view-in-graphite"; target="_blank">(View in Graphite) * **#143999** https://app.graphite.dev/github/pr/llvm/llvm-project/143999?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143986** https://app.graphite.dev/github/pr/llvm/llvm-project/143986?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143479** https://app.graphite.dev/github/pr/llvm/llvm-project/143479?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/>: 1 other dependent PR ([#144139](https://github.com/llvm/llvm-project/pull/144139) https://app.graphite.dev/github/pr/llvm/llvm-project/144139?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/>) * **#143476** https://app.graphite.dev/github/pr/llvm/llvm-project/143476?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143200** https://app.graphite.dev/github/pr/llvm/llvm-project/143200?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143197** https://app.graphite.dev/github/pr/llvm/llvm-project/143197?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * `main` This stack of pull requests is managed by https://graphite.dev?utm-source=stack-comment";>Graphite. Learn more about https://stacking.dev/?utm_source=stack-comment";>stacking. https://github.com/llvm/llvm-project/pull/145117 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Scale embeddings once in vocab analysis instead of repetitive scaling (PR #143986)
@@ -104,7 +106,10 @@ MODULE_PASS("lower-ifunc", LowerIFuncPass())
MODULE_PASS("simplify-type-tests", SimplifyTypeTestsPass())
MODULE_PASS("lowertypetests", LowerTypeTestsPass())
MODULE_PASS("fatlto-cleanup", FatLtoCleanup())
-MODULE_PASS("pgo-force-function-attrs", PGOForceFunctionAttrsPass(PGOOpt ?
PGOOpt->ColdOptType : PGOOptions::ColdFuncOpt::Default))
+MODULE_PASS("pgo-force-function-attrs",
+PGOForceFunctionAttrsPass(PGOOpt
svkeerthy wrote:
Yeah, will do. Missed the unrelated formatting changes.
https://github.com/llvm/llvm-project/pull/143986
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Scale embeddings once in vocab analysis instead of repetitive scaling (PR #143986)
@@ -259,32 +306,40 @@ Error IR2VecVocabAnalysis::readVocabulary() {
return createFileError(VocabFile, BufOrError.getError());
auto Content = BufOrError.get()->getBuffer();
- json::Path::Root Path("");
+
Expected ParsedVocabValue = json::parse(Content);
if (!ParsedVocabValue)
return ParsedVocabValue.takeError();
- bool Res = json::fromJSON(*ParsedVocabValue, Vocabulary, Path);
- if (!Res)
-return createStringError(errc::illegal_byte_sequence,
- "Unable to parse the vocabulary");
+ ir2vec::Vocab OpcodeVocab, TypeVocab, ArgVocab;
+ unsigned OpcodeDim, TypeDim, ArgDim;
+ if (auto Err = parseVocabSection("Opcodes", *ParsedVocabValue, OpcodeVocab,
svkeerthy wrote:
Correct. Will put it in the doc.
https://github.com/llvm/llvm-project/pull/143986
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From a2bec77ad03e20cd76b6870149863049a96c4f9e Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 +++-
llvm/include/llvm/Analysis/InlineAdvisor.h| 4 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 8 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 145 --
8 files changed, 338 insertions(+), 24 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..50ba3c13da70f 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,10 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocabIfRequested(Module &M,
+ ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..a166621243cad 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,12 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+// Dimensions of embeddings are not known in the compile time (until vocab is
+// read). Hence macros cannot be used here.
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +160,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatureIndex
Feature) {
constexpr size_t NumberOfFeatures =
static_cast(FeatureIndex::NumberOfFeatures);
-LLVM_ABI extern const std::vector FeatureMap;
+LLVM_ABI extern std::vector FeatureMap;
LLVM_ABI extern const char *const DecisionName;
LLVM_ABI extern const TensorSpec InlineDecisionSpec;
diff --git a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
index 580dd5e95d760..8262dd0846ede 100644
--- a/llvm/include/llvm/Analysis/MLInlin
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From d71dd503f4794abf8a396ddb8a5deeafe0d75c83 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 28095447f6a5a..0b849f3382f63 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -482,14 +482,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 2a7a6edda70a8..06312562060aa 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 7ff7acebedf4e..27cc2a4109879 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -384,17 +385,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From a2bec77ad03e20cd76b6870149863049a96c4f9e Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 +++-
llvm/include/llvm/Analysis/InlineAdvisor.h| 4 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 8 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 145 --
8 files changed, 338 insertions(+), 24 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..50ba3c13da70f 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,10 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocabIfRequested(Module &M,
+ ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..a166621243cad 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,12 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+// Dimensions of embeddings are not known in the compile time (until vocab is
+// read). Hence macros cannot be used here.
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +160,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatureIndex
Feature) {
constexpr size_t NumberOfFeatures =
static_cast(FeatureIndex::NumberOfFeatures);
-LLVM_ABI extern const std::vector FeatureMap;
+LLVM_ABI extern std::vector FeatureMap;
LLVM_ABI extern const char *const DecisionName;
LLVM_ABI extern const TensorSpec InlineDecisionSpec;
diff --git a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
index 580dd5e95d760..8262dd0846ede 100644
--- a/llvm/include/llvm/Analysis/MLInlin
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From d71dd503f4794abf8a396ddb8a5deeafe0d75c83 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 28095447f6a5a..0b849f3382f63 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -482,14 +482,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 2a7a6edda70a8..06312562060aa 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 7ff7acebedf4e..27cc2a4109879 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -384,17 +385,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From 0d921416a0f81e5634705dc9dfc5363d721a55bf Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 4f8fb3f59ca19..e7bba9995b75b 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -479,14 +479,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index f1aaf4cd2e013..6efa6eac56af9 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index de9c2db9531e8..308c3d86a7668 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -388,17 +389,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From ea224dfb11b37573f5dbdd34ca118fee5a9808c1 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 28095447f6a5a..0b849f3382f63 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -482,14 +482,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 2a7a6edda70a8..06312562060aa 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 7ff7acebedf4e..27cc2a4109879 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -384,17 +385,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [NFC] Formatting PassRegistry.def (PR #144139)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/144139
>From cd6a0f4fbfa87df8bed4efcdf066530523f5ec0d Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 13 Jun 2025 18:22:10 +
Subject: [PATCH] [NFC] Formatting PassRegistry.def
---
llvm/lib/Passes/PassRegistry.def | 40 ++--
1 file changed, 23 insertions(+), 17 deletions(-)
diff --git a/llvm/lib/Passes/PassRegistry.def b/llvm/lib/Passes/PassRegistry.def
index f761d0dab09a8..b1570162d3434 100644
--- a/llvm/lib/Passes/PassRegistry.def
+++ b/llvm/lib/Passes/PassRegistry.def
@@ -63,7 +63,8 @@ MODULE_PASS("coro-early", CoroEarlyPass())
MODULE_PASS("cross-dso-cfi", CrossDSOCFIPass())
MODULE_PASS("ctx-instr-gen",
PGOInstrumentationGen(PGOInstrumentationType::CTXPROF))
-MODULE_PASS("ctx-prof-flatten",
PGOCtxProfFlatteningPass(/*IsPreThinlink=*/false))
+MODULE_PASS("ctx-prof-flatten",
+PGOCtxProfFlatteningPass(/*IsPreThinlink=*/false))
MODULE_PASS("ctx-prof-flatten-prethinlink",
PGOCtxProfFlatteningPass(/*IsPreThinlink=*/true))
MODULE_PASS("noinline-nonprevailing", NoinlineNonPrevailing())
@@ -74,7 +75,8 @@ MODULE_PASS("dot-callgraph", CallGraphDOTPrinterPass())
MODULE_PASS("dxil-upgrade", DXILUpgradePass())
MODULE_PASS("elim-avail-extern", EliminateAvailableExternallyPass())
MODULE_PASS("extract-blocks", BlockExtractorPass({}, false))
-MODULE_PASS("expand-variadics",
ExpandVariadicsPass(ExpandVariadicsMode::Disable))
+MODULE_PASS("expand-variadics",
+ExpandVariadicsPass(ExpandVariadicsMode::Disable))
MODULE_PASS("forceattrs", ForceFunctionAttrsPass())
MODULE_PASS("function-import", FunctionImportPass())
MODULE_PASS("global-merge-func", GlobalMergeFuncPass())
@@ -104,7 +106,10 @@ MODULE_PASS("lower-ifunc", LowerIFuncPass())
MODULE_PASS("simplify-type-tests", SimplifyTypeTestsPass())
MODULE_PASS("lowertypetests", LowerTypeTestsPass())
MODULE_PASS("fatlto-cleanup", FatLtoCleanup())
-MODULE_PASS("pgo-force-function-attrs", PGOForceFunctionAttrsPass(PGOOpt ?
PGOOpt->ColdOptType : PGOOptions::ColdFuncOpt::Default))
+MODULE_PASS("pgo-force-function-attrs",
+PGOForceFunctionAttrsPass(PGOOpt
+ ? PGOOpt->ColdOptType
+ : PGOOptions::ColdFuncOpt::Default))
MODULE_PASS("memprof-context-disambiguation", MemProfContextDisambiguation())
MODULE_PASS("memprof-module", ModuleMemProfilerPass())
MODULE_PASS("mergefunc", MergeFunctionsPass())
@@ -178,7 +183,7 @@ MODULE_PASS_WITH_PARAMS(
parseASanPassOptions, "kernel")
MODULE_PASS_WITH_PARAMS(
"cg-profile", "CGProfilePass",
-[](bool InLTOPostLink) { return CGProfilePass(InLTOPostLink);},
+[](bool InLTOPostLink) { return CGProfilePass(InLTOPostLink); },
parseCGProfilePassOptions, "in-lto-post-link")
MODULE_PASS_WITH_PARAMS(
"global-merge", "GlobalMergePass",
@@ -287,7 +292,8 @@ CGSCC_PASS_WITH_PARAMS(
FUNCTION_ANALYSIS("aa", AAManager())
FUNCTION_ANALYSIS("access-info", LoopAccessAnalysis())
FUNCTION_ANALYSIS("assumptions", AssumptionAnalysis())
-FUNCTION_ANALYSIS("bb-sections-profile-reader",
BasicBlockSectionsProfileReaderAnalysis(TM))
+FUNCTION_ANALYSIS("bb-sections-profile-reader",
+ BasicBlockSectionsProfileReaderAnalysis(TM))
FUNCTION_ANALYSIS("block-freq", BlockFrequencyAnalysis())
FUNCTION_ANALYSIS("branch-prob", BranchProbabilityAnalysis())
FUNCTION_ANALYSIS("cycles", CycleAnalysis())
@@ -377,7 +383,7 @@ FUNCTION_PASS("expand-large-div-rem",
ExpandLargeDivRemPass(TM))
FUNCTION_PASS("expand-fp", ExpandFpPass(TM))
FUNCTION_PASS("expand-memcmp", ExpandMemCmpPass(TM))
FUNCTION_PASS("extra-vector-passes",
- ExtraFunctionPassManager())
+ ExtraFunctionPassManager())
FUNCTION_PASS("fix-irreducible", FixIrreduciblePass())
FUNCTION_PASS("flatten-cfg", FlattenCFGPass())
FUNCTION_PASS("float2int", Float2IntPass())
@@ -548,8 +554,7 @@ FUNCTION_PASS_WITH_PARAMS(
"max-iterations=N")
FUNCTION_PASS_WITH_PARAMS(
"lint", "LintPass",
-[](bool AbortOnError) { return LintPass(AbortOnError); },
-parseLintOptions,
+[](bool AbortOnError) { return LintPass(AbortOnError); }, parseLintOptions,
"abort-on-error")
FUNCTION_PASS_WITH_PARAMS(
"loop-unroll", "LoopUnrollPass",
@@ -576,7 +581,8 @@ FUNCTION_PASS_WITH_PARAMS(
"normalize", "IRNormalizerPass",
[](IRNormalizerOptions Options) { return IRNormalizerPass(Options); },
parseIRNormalizerPassOptions,
-
"no-preserve-order;preserve-order;no-rename-all;rename-all;no-fold-all;fold-all;no-reorder-operands;reorder-operands")
+"no-preserve-order;preserve-order;no-rename-all;rename-all;no-fold-all;"
+"fold-all;no-reorder-operands;reorder-operands")
FUNCTION_PASS_WITH_PARAMS(
"mldst-motion", "MergedLoadStoreMotionPass",
[](MergedLoadStoreMotionOptions Opts) {
@@ -590,7 +596,7 @@ FUNCTION_PASS_WITH_PARAMS(
},
[](StringRe
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From d3468ab37c05d4796661f13d61fffe31bcf47ba5 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 163 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 3e7e31aac0f6d..dbce8ebe5e103 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
@@ -226,10 +234,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..2ad65c2f40c33 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +253,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!B
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From d3468ab37c05d4796661f13d61fffe31bcf47ba5 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 163 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 3e7e31aac0f6d..dbce8ebe5e103 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
@@ -226,10 +234,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..2ad65c2f40c33 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +253,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!B
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From 96e4a8b7faac3855a552ca52ff227bfc677cc6be Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 12 ++-
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 164 insertions(+), 67 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index b12a4eeaec3e8..7976cc7470d5b 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,7 +56,12 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
-/// Embedding is a datavtype that wraps std::vector. It provides
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
+/// Embedding is a datatype that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
/// in the sense that it does not allow the user to change the size of the
@@ -226,10 +234,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..2ad65c2f40c33 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +253,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From 96e4a8b7faac3855a552ca52ff227bfc677cc6be Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 12 ++-
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 164 insertions(+), 67 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index b12a4eeaec3e8..7976cc7470d5b 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,7 +56,12 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
-/// Embedding is a datavtype that wraps std::vector. It provides
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
+/// Embedding is a datatype that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
/// in the sense that it does not allow the user to change the size of the
@@ -226,10 +234,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..2ad65c2f40c33 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +253,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143476 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143476 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143479 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] reachable BB (PR #143476)
svkeerthy wrote: > [!WARNING] > This pull request is not mergeable via GitHub because a downstack PR is > open. Once all requirements are satisfied, merge this PR as a stack href="https://app.graphite.dev/github/pr/llvm/llvm-project/143476?utm_source=stack-comment-downstack-mergeability-warning"; > >on Graphite. > https://graphite.dev/docs/merge-pull-requests";>Learn more * **#143476** https://app.graphite.dev/github/pr/llvm/llvm-project/143476?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> 👈 https://app.graphite.dev/github/pr/llvm/llvm-project/143476?utm_source=stack-comment-view-in-graphite"; target="_blank">(View in Graphite) * **#143200** https://app.graphite.dev/github/pr/llvm/llvm-project/143200?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143197** https://app.graphite.dev/github/pr/llvm/llvm-project/143197?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * `main` This stack of pull requests is managed by https://graphite.dev?utm-source=stack-comment";>Graphite. Learn more about https://stacking.dev/?utm_source=stack-comment";>stacking. https://github.com/llvm/llvm-project/pull/143476 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy ready_for_review https://github.com/llvm/llvm-project/pull/143479 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] reachable BB (PR #143476)
https://github.com/svkeerthy created
https://github.com/llvm/llvm-project/pull/143476
None
>From 79060df5f6ebb7f13bd88dd3128800790433a224 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 15 ++---
llvm/test/Analysis/IR2Vec/dbg-inst.ll| 13
llvm/test/Analysis/IR2Vec/unreachable.ll | 42
3 files changed, 66 insertions(+), 4 deletions(-)
create mode 100644 llvm/test/Analysis/IR2Vec/dbg-inst.ll
create mode 100644 llvm/test/Analysis/IR2Vec/unreachable.ll
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 2ad65c2f40c33..8a392e0709c7f 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,10 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/PostOrderIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
+#include "llvm/IR/Function.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -193,7 +196,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -218,9 +222,12 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ ReversePostOrderTraversal RPOT(&F);
+ for (const BasicBlock *BB : RPOT) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
diff --git a/llvm/test/Analysis/IR2Vec/dbg-inst.ll
b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
new file mode 100644
index 0..0f486b0ba6a52
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
@@ -0,0 +1,13 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define void @bar2(ptr %foo) {
+ store i32 0, ptr %foo, align 4
+ tail call void @llvm.dbg.value(metadata !{}, i64 0, metadata !{}, metadata
!{})
+ ret void
+}
+
+declare void @llvm.dbg.value(metadata, i64, metadata, metadata) nounwind
readnone
+
+; CHECK: Instruction vectors:
+; CHECK-NEXT: Instruction: store i32 0, ptr %foo, align 4 [ 7.00 8.00 9.00
]
+; CHECK-NEXT: Instruction: ret void [ 0.00 0.00 0.00 ]
diff --git a/llvm/test/Analysis/IR2Vec/unreachable.ll
b/llvm/test/Analysis/IR2Vec/unreachable.ll
new file mode 100644
index 0..370fe6881d6ce
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/unreachable.ll
@@ -0,0 +1,42 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define dso_local i32 @abc(i32 noundef %a, i32 noundef %b) #0 {
+entry:
+ %retval = alloca i32, align 4
+ %a.addr = alloca i32, align 4
+ %b.addr = alloca i32, align 4
+ store i32 %a, ptr %a.addr, align 4
+ store i32 %b, ptr %b.addr, align 4
+ %0 = load i32, ptr %a.addr, align 4
+ %1 = load i32, ptr %b.addr, align 4
+ %cmp = icmp sgt i32 %0, %1
+ br i1 %cmp, label %if.then, label %if.else
+
+if.then: ; preds = %entry
+ %2 = load i32, ptr %b.addr, align 4
+ store i32 %2, ptr %retval, align 4
+ br label %return
+
+if.else: ; preds = %entry
+ %3 = load i32, ptr %a.addr, align 4
+ store i32 %3, ptr %retval, align 4
+ br label %return
+
+unreachable: ; Unreachable
+ store i32 0, ptr %retval, align 4
+ br label %return
+
+return: ; preds = %if.else, %if.then
+ %4 = load i32, ptr %retval, align 4
+ ret i32 %4
+}
+
+; CHECK: Basic block vectors:
+; CHECK-NEXT: Basic block: entry:
+; CHECK-NEXT: [ 25.00 32.00 39.00 ]
+; CHECK-NEXT: Basic block: if.then:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: if.else:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: return:
+; CHECK-NEXT: [ 4.00 5.00 6.00 ]
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy created
https://github.com/llvm/llvm-project/pull/143479
None
>From 58c85ec93a7d81e89ef9d9dd594aa50056e36559 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/IR2Vec.h | 2 +-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 6 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/IR2Vec.cpp | 4 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
10 files changed, 361 insertions(+), 38 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 7976cc7470d5b..3f44c650e640d 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -239,7 +239,7 @@ class IR2VecVocabAnalysis : public
AnalysisInfoMixin {
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
- explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..91d3378565fc5 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,10 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +158,7 @@ inlineCostFeatureToMlFeature(InlineCos
[llvm-branch-commits] [llvm] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
svkeerthy wrote: > [!WARNING] > This pull request is not mergeable via GitHub because a downstack PR is > open. Once all requirements are satisfied, merge this PR as a stack href="https://app.graphite.dev/github/pr/llvm/llvm-project/143479?utm_source=stack-comment-downstack-mergeability-warning"; > >on Graphite. > https://graphite.dev/docs/merge-pull-requests";>Learn more * **#143479** https://app.graphite.dev/github/pr/llvm/llvm-project/143479?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> 👈 https://app.graphite.dev/github/pr/llvm/llvm-project/143479?utm_source=stack-comment-view-in-graphite"; target="_blank">(View in Graphite) * **#143476** https://app.graphite.dev/github/pr/llvm/llvm-project/143476?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143200** https://app.graphite.dev/github/pr/llvm/llvm-project/143200?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143197** https://app.graphite.dev/github/pr/llvm/llvm-project/143197?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * `main` This stack of pull requests is managed by https://graphite.dev?utm-source=stack-comment";>Graphite. Learn more about https://stacking.dev/?utm_source=stack-comment";>stacking. https://github.com/llvm/llvm-project/pull/143479 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy ready_for_review https://github.com/llvm/llvm-project/pull/143476 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143476 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From 750cb2f730cb1f8d38fc29fa63b92efd53b82eac Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 78 +-
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 163 insertions(+), 62 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 05f87e311d8a2..953b630d98c2b 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
struct Embedding : public std::vector {
@@ -187,10 +195,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 74f2c4f34644a..0bc3a68448ead 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -254,35 +256,57 @@ Error IR2VecVocabAnalysis::readVocabulary() {
return createStringError(errc::illegal_byte_sequence,
"Unable to parse the vocabulary");
}
- assert(Vocabulary.size() > 0 && "Vocabulary is empty");
+
+ if (Vocabulary.empty()) {
+return createString
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From 750cb2f730cb1f8d38fc29fa63b92efd53b82eac Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 78 +-
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 163 insertions(+), 62 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 05f87e311d8a2..953b630d98c2b 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
struct Embedding : public std::vector {
@@ -187,10 +195,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 74f2c4f34644a..0bc3a68448ead 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -254,35 +256,57 @@ Error IR2VecVocabAnalysis::readVocabulary() {
return createStringError(errc::illegal_byte_sequence,
"Unable to parse the vocabulary");
}
- assert(Vocabulary.size() > 0 && "Vocabulary is empty");
+
+ if (Vocabulary.empty()) {
+return createString
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
@@ -83,9 +85,10 @@ void Embedding::scaleAndAdd(const Embedding &Src, float
Factor) {
bool Embedding::approximatelyEquals(const Embedding &RHS,
double Tolerance) const {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
- for (size_t i = 0; i < this->size(); ++i)
+ for (size_t i = 0; i < this->size(); ++i) {
svkeerthy wrote:
Yes, done
https://github.com/llvm/llvm-project/pull/143200
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From c1842ec938d27261a0b18bc9bcbb20328713f20a Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 163 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 05f87e311d8a2..953b630d98c2b 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
struct Embedding : public std::vector {
@@ -187,10 +195,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 74f2c4f34644a..d6c7d6524c61d 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -240,9 +242,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!BufOrError)
return createF
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From 7f2012cd56db0fc6e1c430a8d5b38d360b33145f Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 163 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 14f28999b174c..3d32942670785 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
@@ -224,10 +232,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..2ad65c2f40c33 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +253,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!B
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From 7f2012cd56db0fc6e1c430a8d5b38d360b33145f Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 163 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 14f28999b174c..3d32942670785 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
@@ -224,10 +232,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..2ad65c2f40c33 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +253,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!B
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
@@ -218,9 +222,12 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ ReversePostOrderTraversal RPOT(&F);
svkeerthy wrote:
Yeah. Also, the current MLInliner considers only reachable blocks (tests have
some examples). This check ensures parity, and it makes sense to consider only
reachable blocks in general. If we need the embedding of an unreachable block,
it can still be obtained by directly invoking `getBBVector()`.
https://github.com/llvm/llvm-project/pull/143476
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
@@ -142,6 +142,10 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
svkeerthy wrote:
Done
https://github.com/llvm/llvm-project/pull/143479
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
@@ -186,6 +186,20 @@ MLInlineAdvisor::MLInlineAdvisor(
EdgeCount += getLocalCalls(KVP.first->getFunction());
}
NodeCount = AllNodes.size();
+
+ if (auto IR2VecVocabResult = MAM.getCachedResult(M)) {
+if (!IR2VecVocabResult->isValid()) {
+ M.getContext().emitError("IR2VecVocabAnalysis is not valid");
+ return;
+}
+// Add the IR2Vec features to the feature map
+auto IR2VecDim = IR2VecVocabResult->getDimension();
+FeatureMap.push_back(
+TensorSpec::createSpec("callee_embedding", {IR2VecDim}));
+FeatureMap.push_back(
+TensorSpec::createSpec("caller_embedding", {IR2VecDim}));
+UseIR2Vec = true;
svkeerthy wrote:
Okay. Done.
https://github.com/llvm/llvm-project/pull/143479
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From b40003fb48e9c135dd1b1a06ba10f3cc2cd81d64 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 8 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
8 files changed, 360 insertions(+), 35 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..a166621243cad 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,12 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+// Dimensions of embeddings are not known in the compile time (until vocab is
+// read). Hence macros cannot be used here.
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +160,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatureIndex
Feature) {
constexpr size_t NumberOfFeatures =
static_cast(FeatureIndex::NumberOfFeatures);
-LLVM_ABI extern const std::vector FeatureMap;
+LLVM_ABI extern std::vector FeatureMap;
LLVM_ABI extern const char *const DecisionName;
LLVM_ABI extern const TensorSpec InlineDecisionSpec;
diff --git a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
index 580dd5e95d760..8262dd0846ede 100644
--- a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
@@
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From c200f67b92e6b6b3c79d5dcf17a5aa2105871de7 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 11 ++
llvm/lib/Analysis/IR2Vec.cpp | 82 +--
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 164 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 8bf21b0e75d67..de67955d85d7c 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+extern cl::opt OpcWeight;
+extern cl::opt TypeWeight;
+extern cl::opt ArgWeight;
+
/// Embedding is a datatype that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
@@ -226,10 +234,13 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(const ir2vec::Vocab &Vocab);
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..0f7303c1b0917 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,17 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional, cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+cl::opt TypeWeight("ir2vec-type-weight", cl::Optional, cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional, cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +250,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!BufOrError)
return createFileError(VocabFile, BufOrError.getError());
- }
+
auto Content = BufOrError.get()->getBuffer();
json::Path::Roo
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From 716f3d2db74a7d3f21cfaf1340e62425c16d5fe1 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 13 +---
llvm/test/Analysis/IR2Vec/dbg-inst.ll| 13
llvm/test/Analysis/IR2Vec/unreachable.ll | 42
3 files changed, 64 insertions(+), 4 deletions(-)
create mode 100644 llvm/test/Analysis/IR2Vec/dbg-inst.ll
create mode 100644 llvm/test/Analysis/IR2Vec/unreachable.ll
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 0f7303c1b0917..fa38c35796a0e 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,9 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +192,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +218,11 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ for (const BasicBlock *BB : depth_first(&F)) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
diff --git a/llvm/test/Analysis/IR2Vec/dbg-inst.ll
b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
new file mode 100644
index 0..0f486b0ba6a52
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
@@ -0,0 +1,13 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define void @bar2(ptr %foo) {
+ store i32 0, ptr %foo, align 4
+ tail call void @llvm.dbg.value(metadata !{}, i64 0, metadata !{}, metadata
!{})
+ ret void
+}
+
+declare void @llvm.dbg.value(metadata, i64, metadata, metadata) nounwind
readnone
+
+; CHECK: Instruction vectors:
+; CHECK-NEXT: Instruction: store i32 0, ptr %foo, align 4 [ 7.00 8.00 9.00
]
+; CHECK-NEXT: Instruction: ret void [ 0.00 0.00 0.00 ]
diff --git a/llvm/test/Analysis/IR2Vec/unreachable.ll
b/llvm/test/Analysis/IR2Vec/unreachable.ll
new file mode 100644
index 0..370fe6881d6ce
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/unreachable.ll
@@ -0,0 +1,42 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define dso_local i32 @abc(i32 noundef %a, i32 noundef %b) #0 {
+entry:
+ %retval = alloca i32, align 4
+ %a.addr = alloca i32, align 4
+ %b.addr = alloca i32, align 4
+ store i32 %a, ptr %a.addr, align 4
+ store i32 %b, ptr %b.addr, align 4
+ %0 = load i32, ptr %a.addr, align 4
+ %1 = load i32, ptr %b.addr, align 4
+ %cmp = icmp sgt i32 %0, %1
+ br i1 %cmp, label %if.then, label %if.else
+
+if.then: ; preds = %entry
+ %2 = load i32, ptr %b.addr, align 4
+ store i32 %2, ptr %retval, align 4
+ br label %return
+
+if.else: ; preds = %entry
+ %3 = load i32, ptr %a.addr, align 4
+ store i32 %3, ptr %retval, align 4
+ br label %return
+
+unreachable: ; Unreachable
+ store i32 0, ptr %retval, align 4
+ br label %return
+
+return: ; preds = %if.else, %if.then
+ %4 = load i32, ptr %retval, align 4
+ ret i32 %4
+}
+
+; CHECK: Basic block vectors:
+; CHECK-NEXT: Basic block: entry:
+; CHECK-NEXT: [ 25.00 32.00 39.00 ]
+; CHECK-NEXT: Basic block: if.then:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: if.else:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: return:
+; CHECK-NEXT: [ 4.00 5.00 6.00 ]
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
@@ -294,8 +294,8 @@ Error IR2VecVocabAnalysis::readVocabulary() {
return Error::success();
}
-IR2VecVocabAnalysis::IR2VecVocabAnalysis(Vocab &&Vocabulary)
-: Vocabulary(std::move(Vocabulary)) {}
+IR2VecVocabAnalysis::IR2VecVocabAnalysis(Vocab Vocabulary)
+: Vocabulary(Vocabulary) {}
svkeerthy wrote:
Yes, moved it to earlier PR (#143200)
https://github.com/llvm/llvm-project/pull/143479
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From 716f3d2db74a7d3f21cfaf1340e62425c16d5fe1 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 13 +---
llvm/test/Analysis/IR2Vec/dbg-inst.ll| 13
llvm/test/Analysis/IR2Vec/unreachable.ll | 42
3 files changed, 64 insertions(+), 4 deletions(-)
create mode 100644 llvm/test/Analysis/IR2Vec/dbg-inst.ll
create mode 100644 llvm/test/Analysis/IR2Vec/unreachable.ll
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 0f7303c1b0917..fa38c35796a0e 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,9 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +192,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +218,11 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ for (const BasicBlock *BB : depth_first(&F)) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
diff --git a/llvm/test/Analysis/IR2Vec/dbg-inst.ll
b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
new file mode 100644
index 0..0f486b0ba6a52
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
@@ -0,0 +1,13 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define void @bar2(ptr %foo) {
+ store i32 0, ptr %foo, align 4
+ tail call void @llvm.dbg.value(metadata !{}, i64 0, metadata !{}, metadata
!{})
+ ret void
+}
+
+declare void @llvm.dbg.value(metadata, i64, metadata, metadata) nounwind
readnone
+
+; CHECK: Instruction vectors:
+; CHECK-NEXT: Instruction: store i32 0, ptr %foo, align 4 [ 7.00 8.00 9.00
]
+; CHECK-NEXT: Instruction: ret void [ 0.00 0.00 0.00 ]
diff --git a/llvm/test/Analysis/IR2Vec/unreachable.ll
b/llvm/test/Analysis/IR2Vec/unreachable.ll
new file mode 100644
index 0..370fe6881d6ce
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/unreachable.ll
@@ -0,0 +1,42 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define dso_local i32 @abc(i32 noundef %a, i32 noundef %b) #0 {
+entry:
+ %retval = alloca i32, align 4
+ %a.addr = alloca i32, align 4
+ %b.addr = alloca i32, align 4
+ store i32 %a, ptr %a.addr, align 4
+ store i32 %b, ptr %b.addr, align 4
+ %0 = load i32, ptr %a.addr, align 4
+ %1 = load i32, ptr %b.addr, align 4
+ %cmp = icmp sgt i32 %0, %1
+ br i1 %cmp, label %if.then, label %if.else
+
+if.then: ; preds = %entry
+ %2 = load i32, ptr %b.addr, align 4
+ store i32 %2, ptr %retval, align 4
+ br label %return
+
+if.else: ; preds = %entry
+ %3 = load i32, ptr %a.addr, align 4
+ store i32 %3, ptr %retval, align 4
+ br label %return
+
+unreachable: ; Unreachable
+ store i32 0, ptr %retval, align 4
+ br label %return
+
+return: ; preds = %if.else, %if.then
+ %4 = load i32, ptr %retval, align 4
+ ret i32 %4
+}
+
+; CHECK: Basic block vectors:
+; CHECK-NEXT: Basic block: entry:
+; CHECK-NEXT: [ 25.00 32.00 39.00 ]
+; CHECK-NEXT: Basic block: if.then:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: if.else:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: return:
+; CHECK-NEXT: [ 4.00 5.00 6.00 ]
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From b40003fb48e9c135dd1b1a06ba10f3cc2cd81d64 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 8 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
8 files changed, 360 insertions(+), 35 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..a166621243cad 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,12 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+// Dimensions of embeddings are not known in the compile time (until vocab is
+// read). Hence macros cannot be used here.
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +160,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatureIndex
Feature) {
constexpr size_t NumberOfFeatures =
static_cast(FeatureIndex::NumberOfFeatures);
-LLVM_ABI extern const std::vector FeatureMap;
+LLVM_ABI extern std::vector FeatureMap;
LLVM_ABI extern const char *const DecisionName;
LLVM_ABI extern const TensorSpec InlineDecisionSpec;
diff --git a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
index 580dd5e95d760..8262dd0846ede 100644
--- a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
@@
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
@@ -199,6 +199,29 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
#undef CHECK_OPERAND
}
}
+
+ if (IR2VecVocab) {
+// We instantiate the IR2Vec embedder each time, as having an unique
+// pointer to the embedder as member of the class would make it
+// non-copyable. Instantiating the embedder in itself is not costly.
+auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
svkeerthy wrote:
Yes, we can do that. I'll put up a patch to remove it.
https://github.com/llvm/llvm-project/pull/143479
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
@@ -193,7 +196,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
svkeerthy wrote:
Right. We just want to ensure we consider only non-debug, non-pseudo
instructions. Some of the old ll files in tests still have debug instructions.
https://github.com/llvm/llvm-project/pull/143476
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
@@ -20,33 +21,102 @@
#include "llvm/Support/Compiler.h"
#include "llvm/Support/SourceMgr.h"
#include "llvm/Transforms/Utils/Cloning.h"
+#include "gmock/gmock.h"
#include "gtest/gtest.h"
#include
using namespace llvm;
+using namespace testing;
namespace llvm {
LLVM_ABI extern cl::opt EnableDetailedFunctionProperties;
LLVM_ABI extern cl::opt BigBasicBlockInstructionThreshold;
LLVM_ABI extern cl::opt MediumBasicBlockInstrutionThreshold;
+LLVM_ABI extern cl::opt ir2vec::OpcWeight;
+LLVM_ABI extern cl::opt ir2vec::TypeWeight;
+LLVM_ABI extern cl::opt ir2vec::ArgWeight;
} // namespace llvm
namespace {
class FunctionPropertiesAnalysisTest : public testing::Test {
public:
FunctionPropertiesAnalysisTest() {
+createTestVocabulary(1);
+MAM.registerPass([&] { return IR2VecVocabAnalysis(Vocabulary); });
+MAM.registerPass([&] { return PassInstrumentationAnalysis(); });
+FAM.registerPass([&] { return ModuleAnalysisManagerFunctionProxy(MAM); });
FAM.registerPass([&] { return DominatorTreeAnalysis(); });
FAM.registerPass([&] { return LoopAnalysis(); });
FAM.registerPass([&] { return PassInstrumentationAnalysis(); });
+
+ir2vec::OpcWeight = 1.0;
+ir2vec::TypeWeight = 1.0;
+ir2vec::ArgWeight = 1.0;
+ }
+
+private:
+ float OriginalOpcWeight = ir2vec::OpcWeight;
+ float OriginalTypeWeight = ir2vec::TypeWeight;
+ float OriginalArgWeight = ir2vec::ArgWeight;
+
+ void createTestVocabulary(unsigned Dim) {
+Vocabulary["add"] = ir2vec::Embedding(Dim, 0.1);
svkeerthy wrote:
Currently the constructor of vocab analysis could only take in Vocabulary map
and not JSON. It has to be via flags if we need to use JSON. This design of
constructor would help in later patches where vocab map would be autogenerated
during the build time.
https://github.com/llvm/llvm-project/pull/143479
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From 173c3b1479f50923429d02e7cd666d3d033f0d65 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 13 +---
llvm/test/Analysis/IR2Vec/dbg-inst.ll| 13
llvm/test/Analysis/IR2Vec/unreachable.ll | 42
3 files changed, 64 insertions(+), 4 deletions(-)
create mode 100644 llvm/test/Analysis/IR2Vec/dbg-inst.ll
create mode 100644 llvm/test/Analysis/IR2Vec/unreachable.ll
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 0f7303c1b0917..fa38c35796a0e 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,9 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +192,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +218,11 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ for (const BasicBlock *BB : depth_first(&F)) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
diff --git a/llvm/test/Analysis/IR2Vec/dbg-inst.ll
b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
new file mode 100644
index 0..0f486b0ba6a52
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
@@ -0,0 +1,13 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define void @bar2(ptr %foo) {
+ store i32 0, ptr %foo, align 4
+ tail call void @llvm.dbg.value(metadata !{}, i64 0, metadata !{}, metadata
!{})
+ ret void
+}
+
+declare void @llvm.dbg.value(metadata, i64, metadata, metadata) nounwind
readnone
+
+; CHECK: Instruction vectors:
+; CHECK-NEXT: Instruction: store i32 0, ptr %foo, align 4 [ 7.00 8.00 9.00
]
+; CHECK-NEXT: Instruction: ret void [ 0.00 0.00 0.00 ]
diff --git a/llvm/test/Analysis/IR2Vec/unreachable.ll
b/llvm/test/Analysis/IR2Vec/unreachable.ll
new file mode 100644
index 0..370fe6881d6ce
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/unreachable.ll
@@ -0,0 +1,42 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define dso_local i32 @abc(i32 noundef %a, i32 noundef %b) #0 {
+entry:
+ %retval = alloca i32, align 4
+ %a.addr = alloca i32, align 4
+ %b.addr = alloca i32, align 4
+ store i32 %a, ptr %a.addr, align 4
+ store i32 %b, ptr %b.addr, align 4
+ %0 = load i32, ptr %a.addr, align 4
+ %1 = load i32, ptr %b.addr, align 4
+ %cmp = icmp sgt i32 %0, %1
+ br i1 %cmp, label %if.then, label %if.else
+
+if.then: ; preds = %entry
+ %2 = load i32, ptr %b.addr, align 4
+ store i32 %2, ptr %retval, align 4
+ br label %return
+
+if.else: ; preds = %entry
+ %3 = load i32, ptr %a.addr, align 4
+ store i32 %3, ptr %retval, align 4
+ br label %return
+
+unreachable: ; Unreachable
+ store i32 0, ptr %retval, align 4
+ br label %return
+
+return: ; preds = %if.else, %if.then
+ %4 = load i32, ptr %retval, align 4
+ ret i32 %4
+}
+
+; CHECK: Basic block vectors:
+; CHECK-NEXT: Basic block: entry:
+; CHECK-NEXT: [ 25.00 32.00 39.00 ]
+; CHECK-NEXT: Basic block: if.then:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: if.else:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: return:
+; CHECK-NEXT: [ 4.00 5.00 6.00 ]
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From 5d7a2b0f4e15da7c4322954bda4bf0e9a3795a66 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 8 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
8 files changed, 360 insertions(+), 35 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..a166621243cad 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,12 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+// Dimensions of embeddings are not known in the compile time (until vocab is
+// read). Hence macros cannot be used here.
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +160,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatureIndex
Feature) {
constexpr size_t NumberOfFeatures =
static_cast(FeatureIndex::NumberOfFeatures);
-LLVM_ABI extern const std::vector FeatureMap;
+LLVM_ABI extern std::vector FeatureMap;
LLVM_ABI extern const char *const DecisionName;
LLVM_ABI extern const TensorSpec InlineDecisionSpec;
diff --git a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
index 580dd5e95d760..8262dd0846ede 100644
--- a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
@@
[llvm-branch-commits] [llvm] Vocab changes1 (PR #143200)
svkeerthy wrote: > [!WARNING] > This pull request is not mergeable via GitHub because a downstack PR is > open. Once all requirements are satisfied, merge this PR as a stack href="https://app.graphite.dev/github/pr/llvm/llvm-project/143200?utm_source=stack-comment-downstack-mergeability-warning"; > >on Graphite. > https://graphite.dev/docs/merge-pull-requests";>Learn more * **#143200** https://app.graphite.dev/github/pr/llvm/llvm-project/143200?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> 👈 https://app.graphite.dev/github/pr/llvm/llvm-project/143200?utm_source=stack-comment-view-in-graphite"; target="_blank">(View in Graphite) * **#143197** https://app.graphite.dev/github/pr/llvm/llvm-project/143197?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * `main` This stack of pull requests is managed by https://graphite.dev?utm-source=stack-comment";>Graphite. Learn more about https://stacking.dev/?utm_source=stack-comment";>stacking. https://github.com/llvm/llvm-project/pull/143200 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] Vocab changes1 (PR #143200)
https://github.com/svkeerthy created
https://github.com/llvm/llvm-project/pull/143200
None
>From d62a38bb0784972e363a7408ba362c31e404f02c Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 81 ++-
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 165 insertions(+), 63 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 930b13f079796..1bd80ed65d434 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+LLVM_ABI extern cl::opt OpcWeight;
+LLVM_ABI extern cl::opt TypeWeight;
+LLVM_ABI extern cl::opt ArgWeight;
+
/// Embedding is a ADT that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
struct Embedding : public std::vector {
@@ -187,10 +195,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 8ee8e5b0ff74e..e751f10b8 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,20 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
+ cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
+ cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+LLVM_ABI cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
+ cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -83,9 +85,10 @@ void Embedding::scaleAndAdd(const Embedding &Src, float
Factor) {
bool Embedding::approximatelyEquals(const Embedding &RHS,
double Tolerance) const {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
- for (size_t
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143200 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy ready_for_review https://github.com/llvm/llvm-project/pull/143200 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143200 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143200 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
@@ -20,33 +21,102 @@
#include "llvm/Support/Compiler.h"
#include "llvm/Support/SourceMgr.h"
#include "llvm/Transforms/Utils/Cloning.h"
+#include "gmock/gmock.h"
#include "gtest/gtest.h"
#include
using namespace llvm;
+using namespace testing;
namespace llvm {
LLVM_ABI extern cl::opt EnableDetailedFunctionProperties;
LLVM_ABI extern cl::opt BigBasicBlockInstructionThreshold;
LLVM_ABI extern cl::opt MediumBasicBlockInstrutionThreshold;
+LLVM_ABI extern cl::opt ir2vec::OpcWeight;
+LLVM_ABI extern cl::opt ir2vec::TypeWeight;
+LLVM_ABI extern cl::opt ir2vec::ArgWeight;
} // namespace llvm
namespace {
class FunctionPropertiesAnalysisTest : public testing::Test {
public:
FunctionPropertiesAnalysisTest() {
+createTestVocabulary(1);
+MAM.registerPass([&] { return IR2VecVocabAnalysis(Vocabulary); });
+MAM.registerPass([&] { return PassInstrumentationAnalysis(); });
+FAM.registerPass([&] { return ModuleAnalysisManagerFunctionProxy(MAM); });
FAM.registerPass([&] { return DominatorTreeAnalysis(); });
FAM.registerPass([&] { return LoopAnalysis(); });
FAM.registerPass([&] { return PassInstrumentationAnalysis(); });
+
+ir2vec::OpcWeight = 1.0;
+ir2vec::TypeWeight = 1.0;
+ir2vec::ArgWeight = 1.0;
+ }
+
+private:
+ float OriginalOpcWeight = ir2vec::OpcWeight;
+ float OriginalTypeWeight = ir2vec::TypeWeight;
+ float OriginalArgWeight = ir2vec::ArgWeight;
+
+ void createTestVocabulary(unsigned Dim) {
+Vocabulary["add"] = ir2vec::Embedding(Dim, 0.1);
svkeerthy wrote:
Simplified it using lambda. Please let me know if this is better. (Just thought
this would reduce the number of lines)
https://github.com/llvm/llvm-project/pull/143479
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From b7ec65230c731c77dea1f4a484df6731819729b5 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 +++-
llvm/include/llvm/Analysis/InlineAdvisor.h| 4 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 8 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 145 --
8 files changed, 338 insertions(+), 24 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..50ba3c13da70f 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,10 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocabIfRequested(Module &M,
+ ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..a166621243cad 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,12 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+// Dimensions of embeddings are not known in the compile time (until vocab is
+// read). Hence macros cannot be used here.
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +160,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatureIndex
Feature) {
constexpr size_t NumberOfFeatures =
static_cast(FeatureIndex::NumberOfFeatures);
-LLVM_ABI extern const std::vector FeatureMap;
+LLVM_ABI extern std::vector FeatureMap;
LLVM_ABI extern const char *const DecisionName;
LLVM_ABI extern const TensorSpec InlineDecisionSpec;
diff --git a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
index 580dd5e95d760..8262dd0846ede 100644
--- a/llvm/include/llvm/Analysis/MLInlin
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From 6e44bb095fa214a5146af383de0ffeb89606607b Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 13 +
1 file changed, 9 insertions(+), 4 deletions(-)
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 0f7303c1b0917..fa38c35796a0e 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,9 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +192,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +218,11 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ for (const BasicBlock *BB : depth_first(&F)) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From 6e44bb095fa214a5146af383de0ffeb89606607b Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 13 +
1 file changed, 9 insertions(+), 4 deletions(-)
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 0f7303c1b0917..fa38c35796a0e 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,9 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +192,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +218,11 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ for (const BasicBlock *BB : depth_first(&F)) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From b7ec65230c731c77dea1f4a484df6731819729b5 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 +++-
llvm/include/llvm/Analysis/InlineAdvisor.h| 4 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 8 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 145 --
8 files changed, 338 insertions(+), 24 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..50ba3c13da70f 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,10 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocabIfRequested(Module &M,
+ ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..a166621243cad 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,12 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+// Dimensions of embeddings are not known in the compile time (until vocab is
+// read). Hence macros cannot be used here.
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +160,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatureIndex
Feature) {
constexpr size_t NumberOfFeatures =
static_cast(FeatureIndex::NumberOfFeatures);
-LLVM_ABI extern const std::vector FeatureMap;
+LLVM_ABI extern std::vector FeatureMap;
LLVM_ABI extern const char *const DecisionName;
LLVM_ABI extern const TensorSpec InlineDecisionSpec;
diff --git a/llvm/include/llvm/Analysis/MLInlineAdvisor.h
b/llvm/include/llvm/Analysis/MLInlineAdvisor.h
index 580dd5e95d760..8262dd0846ede 100644
--- a/llvm/include/llvm/Analysis/MLInlin
[llvm-branch-commits] [llvm] [IR2Vec] Scale embeddings once in vocab analysis instead of repetitive scaling (PR #143986)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143986 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Scale embeddings once in vocab analysis instead of repetitive scaling (PR #143986)
https://github.com/svkeerthy commented: @albertcohen - Please have a look. I am not able to add you as reviewer. https://github.com/llvm/llvm-project/pull/143986 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy created
https://github.com/llvm/llvm-project/pull/143999
None
>From cc133a17f78f9ed2082930617ed4d94dbbf9bf97 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 19
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 34 insertions(+), 57 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 4f8fb3f59ca19..e7bba9995b75b 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -479,14 +479,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index f1aaf4cd2e013..6efa6eac56af9 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index f51d3252d6606..68026618449d8 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -389,17 +390,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error cre
[llvm-branch-commits] [llvm] Simplifying creation of Embedder (PR #143999)
svkeerthy wrote: > [!WARNING] > This pull request is not mergeable via GitHub because a downstack PR is > open. Once all requirements are satisfied, merge this PR as a stack href="https://app.graphite.dev/github/pr/llvm/llvm-project/143999?utm_source=stack-comment-downstack-mergeability-warning"; > >on Graphite. > https://graphite.dev/docs/merge-pull-requests";>Learn more * **#143999** https://app.graphite.dev/github/pr/llvm/llvm-project/143999?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> 👈 https://app.graphite.dev/github/pr/llvm/llvm-project/143999?utm_source=stack-comment-view-in-graphite"; target="_blank">(View in Graphite) * **#143986** https://app.graphite.dev/github/pr/llvm/llvm-project/143986?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143479** https://app.graphite.dev/github/pr/llvm/llvm-project/143479?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143476** https://app.graphite.dev/github/pr/llvm/llvm-project/143476?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143200** https://app.graphite.dev/github/pr/llvm/llvm-project/143200?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * **#143197** https://app.graphite.dev/github/pr/llvm/llvm-project/143197?utm_source=stack-comment-icon"; target="_blank">https://static.graphite.dev/graphite-32x32-black.png"; alt="Graphite" width="10px" height="10px"/> * `main` This stack of pull requests is managed by https://graphite.dev?utm-source=stack-comment";>Graphite. Learn more about https://stacking.dev/?utm_source=stack-comment";>stacking. https://github.com/llvm/llvm-project/pull/143999 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143999 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
svkeerthy wrote: @albertcohen https://github.com/llvm/llvm-project/pull/143999 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143999 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy ready_for_review https://github.com/llvm/llvm-project/pull/143999 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143999 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From 1a051f18a4bab9f843c4fee84084d5cd31cecea1 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 4f8fb3f59ca19..e7bba9995b75b 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -479,14 +479,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index f1aaf4cd2e013..6efa6eac56af9 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index de9c2db9531e8..308c3d86a7668 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -388,17 +389,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [IR2Vec] Simplifying creation of Embedder (PR #143999)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143999
>From 1a051f18a4bab9f843c4fee84084d5cd31cecea1 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Thu, 12 Jun 2025 23:54:10 +
Subject: [PATCH] Simplifying creation of Embedder
---
llvm/docs/MLGO.rst| 7 +--
llvm/include/llvm/Analysis/IR2Vec.h | 4 +-
.../Analysis/FunctionPropertiesAnalysis.cpp | 10 ++---
llvm/lib/Analysis/IR2Vec.cpp | 17 +++
.../FunctionPropertiesAnalysisTest.cpp| 7 ++-
llvm/unittests/Analysis/IR2VecTest.cpp| 44 +++
6 files changed, 33 insertions(+), 56 deletions(-)
diff --git a/llvm/docs/MLGO.rst b/llvm/docs/MLGO.rst
index 4f8fb3f59ca19..e7bba9995b75b 100644
--- a/llvm/docs/MLGO.rst
+++ b/llvm/docs/MLGO.rst
@@ -479,14 +479,9 @@ embeddings can be computed and accessed via an
``ir2vec::Embedder`` instance.
// Assuming F is an llvm::Function&
// For example, using IR2VecKind::Symbolic:
- Expected> EmbOrErr =
+ std::unique_ptr Emb =
ir2vec::Embedder::create(IR2VecKind::Symbolic, F, Vocabulary);
- if (auto Err = EmbOrErr.takeError()) {
-// Handle error in embedder creation
-return;
- }
- std::unique_ptr Emb = std::move(*EmbOrErr);
3. **Compute and Access Embeddings**:
Call ``getFunctionVector()`` to get the embedding for the function.
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index f1aaf4cd2e013..6efa6eac56af9 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -170,8 +170,8 @@ class Embedder {
virtual ~Embedder() = default;
/// Factory method to create an Embedder object.
- static Expected>
- create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary);
+ static std::unique_ptr create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary);
/// Returns a map containing instructions and the corresponding embeddings
for
/// the function F if it has been computed. If not, it computes the
embeddings
diff --git a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
index 29d3aaf46dc06..dd4eb7f0df053 100644
--- a/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
+++ b/llvm/lib/Analysis/FunctionPropertiesAnalysis.cpp
@@ -204,16 +204,12 @@ void FunctionPropertiesInfo::updateForBB(const BasicBlock
&BB,
// We instantiate the IR2Vec embedder each time, as having an unique
// pointer to the embedder as member of the class would make it
// non-copyable. Instantiating the embedder in itself is not costly.
-auto EmbOrErr = ir2vec::Embedder::create(IR2VecKind::Symbolic,
+auto Embedder = ir2vec::Embedder::create(IR2VecKind::Symbolic,
*BB.getParent(), *IR2VecVocab);
-if (Error Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-BB.getContext().emitError("Error creating IR2Vec embeddings: " +
- EI.message());
- });
+if (!Embedder) {
+ BB.getContext().emitError("Error creating IR2Vec embeddings");
return;
}
-auto Embedder = std::move(*EmbOrErr);
const auto &BBEmbedding = Embedder->getBBVector(BB);
// Subtract BBEmbedding from Function embedding if the direction is -1,
// and add it if the direction is +1.
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index de9c2db9531e8..308c3d86a7668 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -123,13 +123,14 @@ Embedder::Embedder(const Function &F, const Vocab
&Vocabulary)
Dimension(Vocabulary.begin()->second.size()), OpcWeight(::OpcWeight),
TypeWeight(::TypeWeight), ArgWeight(::ArgWeight) {}
-Expected>
-Embedder::create(IR2VecKind Mode, const Function &F, const Vocab &Vocabulary) {
+std::unique_ptr Embedder::create(IR2VecKind Mode, const Function &F,
+ const Vocab &Vocabulary) {
switch (Mode) {
case IR2VecKind::Symbolic:
return std::make_unique(F, Vocabulary);
}
- return make_error("Unknown IR2VecKind", errc::invalid_argument);
+ llvm_unreachable("Unknown IR2Vec kind");
+ return nullptr;
}
// FIXME: Currently lookups are string based. Use numeric Keys
@@ -388,17 +389,13 @@ PreservedAnalyses IR2VecPrinterPass::run(Module &M,
auto Vocab = IR2VecVocabResult.getVocabulary();
for (Function &F : M) {
-Expected> EmbOrErr =
+std::unique_ptr Emb =
Embedder::create(IR2VecKind::Symbolic, F, Vocab);
-if (auto Err = EmbOrErr.takeError()) {
- handleAllErrors(std::move(Err), [&](const ErrorInfoBase &EI) {
-OS << "Error creating IR2Vec embeddings: " << EI.message() << "\n";
- });
+if (!Emb) {
+ OS << "Error creating I
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From 579c3c03fa1117382c810c8d322f5d83689d79dc Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/IR2Vec.h | 2 +-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 6 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/IR2Vec.cpp | 4 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
10 files changed, 361 insertions(+), 38 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 3a6f47ded8ca4..9acffb996283c 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -239,7 +239,7 @@ class IR2VecVocabAnalysis : public
AnalysisInfoMixin {
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
- explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..91d3378565fc5 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,10 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +158,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatu
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From 579c3c03fa1117382c810c8d322f5d83689d79dc Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/IR2Vec.h | 2 +-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 6 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/IR2Vec.cpp | 4 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
10 files changed, 361 insertions(+), 38 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 3a6f47ded8ca4..9acffb996283c 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -239,7 +239,7 @@ class IR2VecVocabAnalysis : public
AnalysisInfoMixin {
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
- explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..91d3378565fc5 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,10 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +158,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatu
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From cd2cdf560d4ffd9171a1104cbc12586bd06f5542 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 13 +---
llvm/test/Analysis/IR2Vec/dbg-inst.ll| 13
llvm/test/Analysis/IR2Vec/unreachable.ll | 42
3 files changed, 64 insertions(+), 4 deletions(-)
create mode 100644 llvm/test/Analysis/IR2Vec/dbg-inst.ll
create mode 100644 llvm/test/Analysis/IR2Vec/unreachable.ll
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 1d17acad56463..4a5218e040b49 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,9 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +192,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +218,11 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ for (const BasicBlock *BB : depth_first(&F)) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
diff --git a/llvm/test/Analysis/IR2Vec/dbg-inst.ll
b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
new file mode 100644
index 0..0f486b0ba6a52
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
@@ -0,0 +1,13 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define void @bar2(ptr %foo) {
+ store i32 0, ptr %foo, align 4
+ tail call void @llvm.dbg.value(metadata !{}, i64 0, metadata !{}, metadata
!{})
+ ret void
+}
+
+declare void @llvm.dbg.value(metadata, i64, metadata, metadata) nounwind
readnone
+
+; CHECK: Instruction vectors:
+; CHECK-NEXT: Instruction: store i32 0, ptr %foo, align 4 [ 7.00 8.00 9.00
]
+; CHECK-NEXT: Instruction: ret void [ 0.00 0.00 0.00 ]
diff --git a/llvm/test/Analysis/IR2Vec/unreachable.ll
b/llvm/test/Analysis/IR2Vec/unreachable.ll
new file mode 100644
index 0..370fe6881d6ce
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/unreachable.ll
@@ -0,0 +1,42 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define dso_local i32 @abc(i32 noundef %a, i32 noundef %b) #0 {
+entry:
+ %retval = alloca i32, align 4
+ %a.addr = alloca i32, align 4
+ %b.addr = alloca i32, align 4
+ store i32 %a, ptr %a.addr, align 4
+ store i32 %b, ptr %b.addr, align 4
+ %0 = load i32, ptr %a.addr, align 4
+ %1 = load i32, ptr %b.addr, align 4
+ %cmp = icmp sgt i32 %0, %1
+ br i1 %cmp, label %if.then, label %if.else
+
+if.then: ; preds = %entry
+ %2 = load i32, ptr %b.addr, align 4
+ store i32 %2, ptr %retval, align 4
+ br label %return
+
+if.else: ; preds = %entry
+ %3 = load i32, ptr %a.addr, align 4
+ store i32 %3, ptr %retval, align 4
+ br label %return
+
+unreachable: ; Unreachable
+ store i32 0, ptr %retval, align 4
+ br label %return
+
+return: ; preds = %if.else, %if.then
+ %4 = load i32, ptr %retval, align 4
+ ret i32 %4
+}
+
+; CHECK: Basic block vectors:
+; CHECK-NEXT: Basic block: entry:
+; CHECK-NEXT: [ 25.00 32.00 39.00 ]
+; CHECK-NEXT: Basic block: if.then:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: if.else:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: return:
+; CHECK-NEXT: [ 4.00 5.00 6.00 ]
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
@@ -218,9 +222,12 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ ReversePostOrderTraversal RPOT(&F);
svkeerthy wrote:
Thanks. Seems like constructing RPO is costly. Changed it to DF Traversal.
https://github.com/llvm/llvm-project/pull/143476
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
@@ -53,7 +56,12 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
-/// Embedding is a datavtype that wraps std::vector. It provides
+
+LLVM_ABI extern cl::opt OpcWeight;
svkeerthy wrote:
#136623 adds LLVM_ABI tags to some globals and externs in Analysis. Removing it
for now. Can be added later if necessary.
https://github.com/llvm/llvm-project/pull/143200
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLInliner][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143479 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/143479 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From d639ce4a4b3ad531e16d7a1c1fa33a53c7e7393a Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 79 --
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 160 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 8bf21b0e75d67..3a6f47ded8ca4 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+extern cl::opt OpcWeight;
+extern cl::opt TypeWeight;
+extern cl::opt ArgWeight;
+
/// Embedding is a datatype that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
@@ -226,10 +234,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..1d17acad56463 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,17 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional, cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+cl::opt TypeWeight("ir2vec-type-weight", cl::Optional, cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional, cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +250,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!BufOrError)
return createFileError(VocabFile, BufOrError.getError());
- }
+
auto Content = BufOrError.get()->getBuffer();
json::Path::Root Path("");
Expected ParsedVocabValue = json::parse(Content
[llvm-branch-commits] [llvm] [IR2Vec] Minor vocab changes and exposing weights (PR #143200)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143200
>From d639ce4a4b3ad531e16d7a1c1fa33a53c7e7393a Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 6 Jun 2025 20:32:32 +
Subject: [PATCH] Vocab changes1
---
llvm/include/llvm/Analysis/IR2Vec.h| 10 ++
llvm/lib/Analysis/IR2Vec.cpp | 79 --
llvm/unittests/Analysis/IR2VecTest.cpp | 137 ++---
3 files changed, 160 insertions(+), 66 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 8bf21b0e75d67..3a6f47ded8ca4 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -31,7 +31,9 @@
#include "llvm/ADT/DenseMap.h"
#include "llvm/IR/PassManager.h"
+#include "llvm/Support/CommandLine.h"
#include "llvm/Support/ErrorOr.h"
+#include "llvm/Support/JSON.h"
#include
namespace llvm {
@@ -43,6 +45,7 @@ class Function;
class Type;
class Value;
class raw_ostream;
+class LLVMContext;
/// IR2Vec computes two kinds of embeddings: Symbolic and Flow-aware.
/// Symbolic embeddings capture the "syntactic" and "statistical correlation"
@@ -53,6 +56,11 @@ class raw_ostream;
enum class IR2VecKind { Symbolic };
namespace ir2vec {
+
+extern cl::opt OpcWeight;
+extern cl::opt TypeWeight;
+extern cl::opt ArgWeight;
+
/// Embedding is a datatype that wraps std::vector. It provides
/// additional functionality for arithmetic and comparison operations.
/// It is meant to be used *like* std::vector but is more restrictive
@@ -226,10 +234,12 @@ class IR2VecVocabResult {
class IR2VecVocabAnalysis : public AnalysisInfoMixin {
ir2vec::Vocab Vocabulary;
Error readVocabulary();
+ void emitError(Error Err, LLVMContext &Ctx);
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 25ce35d4ace37..1d17acad56463 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -16,13 +16,11 @@
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
-#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/Errc.h"
#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/Format.h"
-#include "llvm/Support/JSON.h"
#include "llvm/Support/MemoryBuffer.h"
using namespace llvm;
@@ -33,6 +31,8 @@ using namespace ir2vec;
STATISTIC(VocabMissCounter,
"Number of lookups to entites not present in the vocabulary");
+namespace llvm {
+namespace ir2vec {
static cl::OptionCategory IR2VecCategory("IR2Vec Options");
// FIXME: Use a default vocab when not specified
@@ -40,18 +40,17 @@ static cl::opt
VocabFile("ir2vec-vocab-path", cl::Optional,
cl::desc("Path to the vocabulary file for IR2Vec"), cl::init(""),
cl::cat(IR2VecCategory));
-static cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional,
-cl::init(1.0),
-cl::desc("Weight for opcode embeddings"),
-cl::cat(IR2VecCategory));
-static cl::opt TypeWeight("ir2vec-type-weight", cl::Optional,
- cl::init(0.5),
- cl::desc("Weight for type embeddings"),
- cl::cat(IR2VecCategory));
-static cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional,
-cl::init(0.2),
-cl::desc("Weight for argument embeddings"),
-cl::cat(IR2VecCategory));
+cl::opt OpcWeight("ir2vec-opc-weight", cl::Optional, cl::init(1.0),
+ cl::desc("Weight for opcode embeddings"),
+ cl::cat(IR2VecCategory));
+cl::opt TypeWeight("ir2vec-type-weight", cl::Optional, cl::init(0.5),
+ cl::desc("Weight for type embeddings"),
+ cl::cat(IR2VecCategory));
+cl::opt ArgWeight("ir2vec-arg-weight", cl::Optional, cl::init(0.2),
+ cl::desc("Weight for argument embeddings"),
+ cl::cat(IR2VecCategory));
+} // namespace ir2vec
+} // namespace llvm
AnalysisKey IR2VecVocabAnalysis::Key;
@@ -251,9 +250,9 @@ bool IR2VecVocabResult::invalidate(
// by auto-generating a default vocabulary during the build time.
Error IR2VecVocabAnalysis::readVocabulary() {
auto BufOrError = MemoryBuffer::getFileOrSTDIN(VocabFile, /*IsText=*/true);
- if (!BufOrError) {
+ if (!BufOrError)
return createFileError(VocabFile, BufOrError.getError());
- }
+
auto Content = BufOrError.get()->getBuffer();
json::Path::Root Path("");
Expected ParsedVocabValue = json::parse(Content
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From 6085399a3d002b78a093d951c6a9b6c5bb3fb243 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/IR2Vec.h | 2 +-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 6 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/IR2Vec.cpp | 4 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
10 files changed, 361 insertions(+), 38 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 3a6f47ded8ca4..9acffb996283c 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -239,7 +239,7 @@ class IR2VecVocabAnalysis : public
AnalysisInfoMixin {
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
- explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..91d3378565fc5 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,10 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +158,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatu
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From abee4cd2dd04ce8a99562b79df9ae1eeed10d030 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 15 ++---
llvm/test/Analysis/IR2Vec/dbg-inst.ll| 13
llvm/test/Analysis/IR2Vec/unreachable.ll | 42
3 files changed, 66 insertions(+), 4 deletions(-)
create mode 100644 llvm/test/Analysis/IR2Vec/dbg-inst.ll
create mode 100644 llvm/test/Analysis/IR2Vec/unreachable.ll
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 1d17acad56463..6d939421ea3d8 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,10 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/PostOrderIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
+#include "llvm/IR/Function.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +193,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +219,12 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ ReversePostOrderTraversal RPOT(&F);
+ for (const BasicBlock *BB : RPOT) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
diff --git a/llvm/test/Analysis/IR2Vec/dbg-inst.ll
b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
new file mode 100644
index 0..0f486b0ba6a52
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
@@ -0,0 +1,13 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define void @bar2(ptr %foo) {
+ store i32 0, ptr %foo, align 4
+ tail call void @llvm.dbg.value(metadata !{}, i64 0, metadata !{}, metadata
!{})
+ ret void
+}
+
+declare void @llvm.dbg.value(metadata, i64, metadata, metadata) nounwind
readnone
+
+; CHECK: Instruction vectors:
+; CHECK-NEXT: Instruction: store i32 0, ptr %foo, align 4 [ 7.00 8.00 9.00
]
+; CHECK-NEXT: Instruction: ret void [ 0.00 0.00 0.00 ]
diff --git a/llvm/test/Analysis/IR2Vec/unreachable.ll
b/llvm/test/Analysis/IR2Vec/unreachable.ll
new file mode 100644
index 0..370fe6881d6ce
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/unreachable.ll
@@ -0,0 +1,42 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define dso_local i32 @abc(i32 noundef %a, i32 noundef %b) #0 {
+entry:
+ %retval = alloca i32, align 4
+ %a.addr = alloca i32, align 4
+ %b.addr = alloca i32, align 4
+ store i32 %a, ptr %a.addr, align 4
+ store i32 %b, ptr %b.addr, align 4
+ %0 = load i32, ptr %a.addr, align 4
+ %1 = load i32, ptr %b.addr, align 4
+ %cmp = icmp sgt i32 %0, %1
+ br i1 %cmp, label %if.then, label %if.else
+
+if.then: ; preds = %entry
+ %2 = load i32, ptr %b.addr, align 4
+ store i32 %2, ptr %retval, align 4
+ br label %return
+
+if.else: ; preds = %entry
+ %3 = load i32, ptr %a.addr, align 4
+ store i32 %3, ptr %retval, align 4
+ br label %return
+
+unreachable: ; Unreachable
+ store i32 0, ptr %retval, align 4
+ br label %return
+
+return: ; preds = %if.else, %if.then
+ %4 = load i32, ptr %retval, align 4
+ ret i32 %4
+}
+
+; CHECK: Basic block vectors:
+; CHECK-NEXT: Basic block: entry:
+; CHECK-NEXT: [ 25.00 32.00 39.00 ]
+; CHECK-NEXT: Basic block: if.then:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: if.else:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: return:
+; CHECK-NEXT: [ 4.00 5.00 6.00 ]
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Consider only reachable BBs and non-debug instructions (PR #143476)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143476
>From abee4cd2dd04ce8a99562b79df9ae1eeed10d030 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 04:49:59 +
Subject: [PATCH] reachable BB
---
llvm/lib/Analysis/IR2Vec.cpp | 15 ++---
llvm/test/Analysis/IR2Vec/dbg-inst.ll| 13
llvm/test/Analysis/IR2Vec/unreachable.ll | 42
3 files changed, 66 insertions(+), 4 deletions(-)
create mode 100644 llvm/test/Analysis/IR2Vec/dbg-inst.ll
create mode 100644 llvm/test/Analysis/IR2Vec/unreachable.ll
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 1d17acad56463..6d939421ea3d8 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -13,7 +13,10 @@
#include "llvm/Analysis/IR2Vec.h"
+#include "llvm/ADT/PostOrderIterator.h"
#include "llvm/ADT/Statistic.h"
+#include "llvm/IR/CFG.h"
+#include "llvm/IR/Function.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Debug.h"
@@ -190,7 +193,8 @@ Embedding SymbolicEmbedder::getOperandEmbedding(const Value
*Op) const {
void SymbolicEmbedder::computeEmbeddings(const BasicBlock &BB) const {
Embedding BBVector(Dimension, 0);
- for (const auto &I : BB) {
+ // We consider only the non-debug and non-pseudo instructions
+ for (const auto &I : BB.instructionsWithoutDebug()) {
Embedding InstVector(Dimension, 0);
const auto OpcVec = lookupVocab(I.getOpcodeName());
@@ -215,9 +219,12 @@ void SymbolicEmbedder::computeEmbeddings(const BasicBlock
&BB) const {
void SymbolicEmbedder::computeEmbeddings() const {
if (F.isDeclaration())
return;
- for (const auto &BB : F) {
-computeEmbeddings(BB);
-FuncVector += BBVecMap[&BB];
+
+ // Consider only the basic blocks that are reachable from entry
+ ReversePostOrderTraversal RPOT(&F);
+ for (const BasicBlock *BB : RPOT) {
+computeEmbeddings(*BB);
+FuncVector += BBVecMap[BB];
}
}
diff --git a/llvm/test/Analysis/IR2Vec/dbg-inst.ll
b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
new file mode 100644
index 0..0f486b0ba6a52
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/dbg-inst.ll
@@ -0,0 +1,13 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define void @bar2(ptr %foo) {
+ store i32 0, ptr %foo, align 4
+ tail call void @llvm.dbg.value(metadata !{}, i64 0, metadata !{}, metadata
!{})
+ ret void
+}
+
+declare void @llvm.dbg.value(metadata, i64, metadata, metadata) nounwind
readnone
+
+; CHECK: Instruction vectors:
+; CHECK-NEXT: Instruction: store i32 0, ptr %foo, align 4 [ 7.00 8.00 9.00
]
+; CHECK-NEXT: Instruction: ret void [ 0.00 0.00 0.00 ]
diff --git a/llvm/test/Analysis/IR2Vec/unreachable.ll
b/llvm/test/Analysis/IR2Vec/unreachable.ll
new file mode 100644
index 0..370fe6881d6ce
--- /dev/null
+++ b/llvm/test/Analysis/IR2Vec/unreachable.ll
@@ -0,0 +1,42 @@
+; RUN: opt -passes='print' -o /dev/null
-ir2vec-vocab-path=%S/Inputs/dummy_3D_vocab.json %s 2>&1 | FileCheck %s
+
+define dso_local i32 @abc(i32 noundef %a, i32 noundef %b) #0 {
+entry:
+ %retval = alloca i32, align 4
+ %a.addr = alloca i32, align 4
+ %b.addr = alloca i32, align 4
+ store i32 %a, ptr %a.addr, align 4
+ store i32 %b, ptr %b.addr, align 4
+ %0 = load i32, ptr %a.addr, align 4
+ %1 = load i32, ptr %b.addr, align 4
+ %cmp = icmp sgt i32 %0, %1
+ br i1 %cmp, label %if.then, label %if.else
+
+if.then: ; preds = %entry
+ %2 = load i32, ptr %b.addr, align 4
+ store i32 %2, ptr %retval, align 4
+ br label %return
+
+if.else: ; preds = %entry
+ %3 = load i32, ptr %a.addr, align 4
+ store i32 %3, ptr %retval, align 4
+ br label %return
+
+unreachable: ; Unreachable
+ store i32 0, ptr %retval, align 4
+ br label %return
+
+return: ; preds = %if.else, %if.then
+ %4 = load i32, ptr %retval, align 4
+ ret i32 %4
+}
+
+; CHECK: Basic block vectors:
+; CHECK-NEXT: Basic block: entry:
+; CHECK-NEXT: [ 25.00 32.00 39.00 ]
+; CHECK-NEXT: Basic block: if.then:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: if.else:
+; CHECK-NEXT: [ 11.00 13.00 15.00 ]
+; CHECK-NEXT: Basic block: return:
+; CHECK-NEXT: [ 4.00 5.00 6.00 ]
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [MLGO][IR2Vec] Integrating IR2Vec with MLInliner (PR #143479)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/143479
>From 6085399a3d002b78a093d951c6a9b6c5bb3fb243 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Tue, 10 Jun 2025 05:40:38 +
Subject: [PATCH] [MLIniner][IR2Vec] Integrating IR2Vec with MLInliner
---
.../Analysis/FunctionPropertiesAnalysis.h | 26 ++-
llvm/include/llvm/Analysis/IR2Vec.h | 2 +-
llvm/include/llvm/Analysis/InlineAdvisor.h| 3 +
.../llvm/Analysis/InlineModelFeatureMaps.h| 6 +-
llvm/include/llvm/Analysis/MLInlineAdvisor.h | 1 +
.../Analysis/FunctionPropertiesAnalysis.cpp | 115 ++-
llvm/lib/Analysis/IR2Vec.cpp | 4 +-
llvm/lib/Analysis/InlineAdvisor.cpp | 29 +++
llvm/lib/Analysis/MLInlineAdvisor.cpp | 34 +++-
.../FunctionPropertiesAnalysisTest.cpp| 179 +++---
10 files changed, 361 insertions(+), 38 deletions(-)
diff --git a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
index babb6d9d6cf0c..06dbfc35a5294 100644
--- a/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
+++ b/llvm/include/llvm/Analysis/FunctionPropertiesAnalysis.h
@@ -15,6 +15,7 @@
#define LLVM_ANALYSIS_FUNCTIONPROPERTIESANALYSIS_H
#include "llvm/ADT/DenseSet.h"
+#include "llvm/Analysis/IR2Vec.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/PassManager.h"
#include "llvm/Support/Compiler.h"
@@ -32,17 +33,19 @@ class FunctionPropertiesInfo {
void updateAggregateStats(const Function &F, const LoopInfo &LI);
void reIncludeBB(const BasicBlock &BB);
+ ir2vec::Embedding FunctionEmbedding = ir2vec::Embedding(0.0);
+ std::optional IR2VecVocab;
+
public:
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(const Function &F, const DominatorTree &DT,
-const LoopInfo &LI);
+const LoopInfo &LI,
+const IR2VecVocabResult *VocabResult);
LLVM_ABI static FunctionPropertiesInfo
getFunctionPropertiesInfo(Function &F, FunctionAnalysisManager &FAM);
- bool operator==(const FunctionPropertiesInfo &FPI) const {
-return std::memcmp(this, &FPI, sizeof(FunctionPropertiesInfo)) == 0;
- }
+ bool operator==(const FunctionPropertiesInfo &FPI) const;
bool operator!=(const FunctionPropertiesInfo &FPI) const {
return !(*this == FPI);
@@ -137,6 +140,19 @@ class FunctionPropertiesInfo {
int64_t CallReturnsVectorPointerCount = 0;
int64_t CallWithManyArgumentsCount = 0;
int64_t CallWithPointerArgumentCount = 0;
+
+ const ir2vec::Embedding &getFunctionEmbedding() const {
+return FunctionEmbedding;
+ }
+
+ const std::optional &getIR2VecVocab() const {
+return IR2VecVocab;
+ }
+
+ // Helper intended to be useful for unittests
+ void setFunctionEmbeddingForTest(const ir2vec::Embedding &Embedding) {
+FunctionEmbedding = Embedding;
+ }
};
// Analysis pass
@@ -192,7 +208,7 @@ class FunctionPropertiesUpdater {
DominatorTree &getUpdatedDominatorTree(FunctionAnalysisManager &FAM) const;
- DenseSet Successors;
+ DenseSet Successors, CallUsers;
// Edges we might potentially need to remove from the dominator tree.
SmallVector DomTreeUpdates;
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 3a6f47ded8ca4..9acffb996283c 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -239,7 +239,7 @@ class IR2VecVocabAnalysis : public
AnalysisInfoMixin {
public:
static AnalysisKey Key;
IR2VecVocabAnalysis() = default;
- explicit IR2VecVocabAnalysis(ir2vec::Vocab &&Vocab);
+ explicit IR2VecVocabAnalysis(ir2vec::Vocab Vocab);
using Result = IR2VecVocabResult;
Result run(Module &M, ModuleAnalysisManager &MAM);
};
diff --git a/llvm/include/llvm/Analysis/InlineAdvisor.h
b/llvm/include/llvm/Analysis/InlineAdvisor.h
index 9d15136e81d10..d2cad4717cbdb 100644
--- a/llvm/include/llvm/Analysis/InlineAdvisor.h
+++ b/llvm/include/llvm/Analysis/InlineAdvisor.h
@@ -331,6 +331,9 @@ class InlineAdvisorAnalysis : public
AnalysisInfoMixin {
};
Result run(Module &M, ModuleAnalysisManager &MAM) { return Result(M, MAM); }
+
+private:
+ static bool initializeIR2VecVocab(Module &M, ModuleAnalysisManager &MAM);
};
/// Printer pass for the InlineAdvisorAnalysis results.
diff --git a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
index 961d5091bf9f3..91d3378565fc5 100644
--- a/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
+++ b/llvm/include/llvm/Analysis/InlineModelFeatureMaps.h
@@ -142,6 +142,10 @@ enum class FeatureIndex : size_t {
INLINE_FEATURE_ITERATOR(POPULATE_INDICES)
#undef POPULATE_INDICES
+// IR2Vec embeddings
+ callee_embedding,
+ caller_embedding,
+
NumberOfFeatures
};
// clang-format on
@@ -154,7 +158,7 @@ inlineCostFeatureToMlFeature(InlineCostFeatu
[llvm-branch-commits] [llvm] [IR2Vec] Overloading `operator+` for `Embeddings` (PR #145118)
@@ -106,6 +106,7 @@ struct Embedding {
const std::vector &getData() const { return Data; }
/// Arithmetic operators
+ Embedding operator+(const Embedding &RHS) const;
svkeerthy wrote:
Sure. Will add them too!
https://github.com/llvm/llvm-project/pull/145118
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Add out-of-place arithmetic operators to Embedding class (PR #145118)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/145118
>From 10019cae162bb53e147797b655da75aac33b0a20 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 20 Jun 2025 23:00:40 +
Subject: [PATCH] Overloading operator+ for Embeddngs
---
llvm/include/llvm/Analysis/IR2Vec.h| 9 --
llvm/lib/Analysis/IR2Vec.cpp | 19 -
llvm/unittests/Analysis/IR2VecTest.cpp | 39 ++
3 files changed, 63 insertions(+), 4 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 040cb84ff27a1..ef8f630d7feb1 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -107,9 +107,12 @@ struct Embedding {
const std::vector &getData() const { return Data; }
/// Arithmetic operators
- Embedding &operator+=(const Embedding &RHS);
- Embedding &operator-=(const Embedding &RHS);
- Embedding &operator*=(double Factor);
+ LLVM_ABI Embedding &operator+=(const Embedding &RHS);
+ LLVM_ABI Embedding operator+(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator-=(const Embedding &RHS);
+ LLVM_ABI Embedding operator-(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator*=(double Factor);
+ LLVM_ABI Embedding operator*(double Factor) const;
/// Adds Src Embedding scaled by Factor with the called Embedding.
/// Called_Embedding += Src * Factor
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 895b3de58a54e..bf456102bb618 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -70,7 +70,6 @@ inline bool fromJSON(const llvm::json::Value &E, Embedding
&Out,
//
==--===//
// Embedding
//===--===//
-
Embedding &Embedding::operator+=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -78,6 +77,12 @@ Embedding &Embedding::operator+=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator+(const Embedding &RHS) const {
+ Embedding Result(*this);
+ Result += RHS;
+ return Result;
+}
+
Embedding &Embedding::operator-=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -85,12 +90,24 @@ Embedding &Embedding::operator-=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator-(const Embedding &RHS) const {
+ Embedding Result(*this);
+ Result -= RHS;
+ return Result;
+}
+
Embedding &Embedding::operator*=(double Factor) {
std::transform(this->begin(), this->end(), this->begin(),
[Factor](double Elem) { return Elem * Factor; });
return *this;
}
+Embedding Embedding::operator*(double Factor) const {
+ Embedding Result(*this);
+ Result *= Factor;
+ return Result;
+}
+
Embedding &Embedding::scaleAndAdd(const Embedding &Src, float Factor) {
assert(this->size() == Src.size() && "Vectors must have the same dimension");
for (size_t Itr = 0; Itr < this->size(); ++Itr)
diff --git a/llvm/unittests/Analysis/IR2VecTest.cpp
b/llvm/unittests/Analysis/IR2VecTest.cpp
index 3c97c20ae72d5..70d4808dc6d54 100644
--- a/llvm/unittests/Analysis/IR2VecTest.cpp
+++ b/llvm/unittests/Analysis/IR2VecTest.cpp
@@ -109,6 +109,18 @@ TEST(EmbeddingTest, ConstructorsAndAccessors) {
}
}
+TEST(EmbeddingTest, AddVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 + E2;
+ EXPECT_THAT(E3, ElementsAre(1.5, 3.5, 2.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, AddVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -120,6 +132,18 @@ TEST(EmbeddingTest, AddVectors) {
EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
}
+TEST(EmbeddingTest, SubtractVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 - E2;
+ EXPECT_THAT(E3, ElementsAre(0.5, 0.5, 4.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, SubtractVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -137,6 +161,15 @@ TEST(EmbeddingTest, ScaleVector) {
EXPECT_THAT(E1, ElementsAre(0.5, 1.0, 1.5));
}
+TEST(EmbeddingTest, ScaleVectorOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = E1 * 0.5f;
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.0, 1.5));
+
+ // Check that E1 is unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+}
+
TEST(EmbeddingTest, AddS
[llvm-branch-commits] [llvm] [IR2Vec] Overloading `operator+` for `Embeddings` (PR #145118)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/145118
>From f1976fa2454846d80822761f7a095b29c2062652 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 20 Jun 2025 23:00:40 +
Subject: [PATCH] Overloading operator+ for Embeddngs
---
llvm/include/llvm/Analysis/IR2Vec.h| 9 --
llvm/lib/Analysis/IR2Vec.cpp | 23 +++
llvm/unittests/Analysis/IR2VecTest.cpp | 39 ++
3 files changed, 68 insertions(+), 3 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 040cb84ff27a1..d63be227b1849 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -107,9 +107,12 @@ struct Embedding {
const std::vector &getData() const { return Data; }
/// Arithmetic operators
- Embedding &operator+=(const Embedding &RHS);
- Embedding &operator-=(const Embedding &RHS);
- Embedding &operator*=(double Factor);
+ LLVM_ABI Embedding operator+(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator+=(const Embedding &RHS);
+ LLVM_ABI Embedding operator-(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator-=(const Embedding &RHS);
+ LLVM_ABI Embedding operator*(double Factor) const;
+ LLVM_ABI Embedding &operator*=(double Factor);
/// Adds Src Embedding scaled by Factor with the called Embedding.
/// Called_Embedding += Src * Factor
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 895b3de58a54e..e499ebdd5ed3c 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -71,6 +71,14 @@ inline bool fromJSON(const llvm::json::Value &E, Embedding
&Out,
// Embedding
//===--===//
+Embedding Embedding::operator+(const Embedding &RHS) const {
+ assert(this->size() == RHS.size() && "Vectors must have the same dimension");
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), RHS.begin(), Result.begin(),
+ std::plus());
+ return Result;
+}
+
Embedding &Embedding::operator+=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -78,6 +86,14 @@ Embedding &Embedding::operator+=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator-(const Embedding &RHS) const {
+ assert(this->size() == RHS.size() && "Vectors must have the same dimension");
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), RHS.begin(), Result.begin(),
+ std::minus());
+ return Result;
+}
+
Embedding &Embedding::operator-=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -85,6 +101,13 @@ Embedding &Embedding::operator-=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator*(double Factor) const {
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), Result.begin(),
+ [Factor](double Elem) { return Elem * Factor; });
+ return Result;
+}
+
Embedding &Embedding::operator*=(double Factor) {
std::transform(this->begin(), this->end(), this->begin(),
[Factor](double Elem) { return Elem * Factor; });
diff --git a/llvm/unittests/Analysis/IR2VecTest.cpp
b/llvm/unittests/Analysis/IR2VecTest.cpp
index 3c97c20ae72d5..70d4808dc6d54 100644
--- a/llvm/unittests/Analysis/IR2VecTest.cpp
+++ b/llvm/unittests/Analysis/IR2VecTest.cpp
@@ -109,6 +109,18 @@ TEST(EmbeddingTest, ConstructorsAndAccessors) {
}
}
+TEST(EmbeddingTest, AddVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 + E2;
+ EXPECT_THAT(E3, ElementsAre(1.5, 3.5, 2.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, AddVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -120,6 +132,18 @@ TEST(EmbeddingTest, AddVectors) {
EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
}
+TEST(EmbeddingTest, SubtractVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 - E2;
+ EXPECT_THAT(E3, ElementsAre(0.5, 0.5, 4.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, SubtractVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -137,6 +161,15 @@ TEST(EmbeddingTest, ScaleVector) {
EXPECT_THAT(E1, ElementsAre(0.5, 1.0, 1.5));
}
+TEST(EmbeddingTest, ScaleVectorOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = E1 * 0.5f;
+ EXPECT_THAT(E2,
[llvm-branch-commits] [llvm] [IR2Vec] Add out-of-place arithmetic operators to Embedding class (PR #145118)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/145118 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Overloading `operator+` for `Embeddings` (PR #145118)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/145118
>From f1976fa2454846d80822761f7a095b29c2062652 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 20 Jun 2025 23:00:40 +
Subject: [PATCH] Overloading operator+ for Embeddngs
---
llvm/include/llvm/Analysis/IR2Vec.h| 9 --
llvm/lib/Analysis/IR2Vec.cpp | 23 +++
llvm/unittests/Analysis/IR2VecTest.cpp | 39 ++
3 files changed, 68 insertions(+), 3 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 040cb84ff27a1..d63be227b1849 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -107,9 +107,12 @@ struct Embedding {
const std::vector &getData() const { return Data; }
/// Arithmetic operators
- Embedding &operator+=(const Embedding &RHS);
- Embedding &operator-=(const Embedding &RHS);
- Embedding &operator*=(double Factor);
+ LLVM_ABI Embedding operator+(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator+=(const Embedding &RHS);
+ LLVM_ABI Embedding operator-(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator-=(const Embedding &RHS);
+ LLVM_ABI Embedding operator*(double Factor) const;
+ LLVM_ABI Embedding &operator*=(double Factor);
/// Adds Src Embedding scaled by Factor with the called Embedding.
/// Called_Embedding += Src * Factor
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 895b3de58a54e..e499ebdd5ed3c 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -71,6 +71,14 @@ inline bool fromJSON(const llvm::json::Value &E, Embedding
&Out,
// Embedding
//===--===//
+Embedding Embedding::operator+(const Embedding &RHS) const {
+ assert(this->size() == RHS.size() && "Vectors must have the same dimension");
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), RHS.begin(), Result.begin(),
+ std::plus());
+ return Result;
+}
+
Embedding &Embedding::operator+=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -78,6 +86,14 @@ Embedding &Embedding::operator+=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator-(const Embedding &RHS) const {
+ assert(this->size() == RHS.size() && "Vectors must have the same dimension");
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), RHS.begin(), Result.begin(),
+ std::minus());
+ return Result;
+}
+
Embedding &Embedding::operator-=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -85,6 +101,13 @@ Embedding &Embedding::operator-=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator*(double Factor) const {
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), Result.begin(),
+ [Factor](double Elem) { return Elem * Factor; });
+ return Result;
+}
+
Embedding &Embedding::operator*=(double Factor) {
std::transform(this->begin(), this->end(), this->begin(),
[Factor](double Elem) { return Elem * Factor; });
diff --git a/llvm/unittests/Analysis/IR2VecTest.cpp
b/llvm/unittests/Analysis/IR2VecTest.cpp
index 3c97c20ae72d5..70d4808dc6d54 100644
--- a/llvm/unittests/Analysis/IR2VecTest.cpp
+++ b/llvm/unittests/Analysis/IR2VecTest.cpp
@@ -109,6 +109,18 @@ TEST(EmbeddingTest, ConstructorsAndAccessors) {
}
}
+TEST(EmbeddingTest, AddVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 + E2;
+ EXPECT_THAT(E3, ElementsAre(1.5, 3.5, 2.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, AddVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -120,6 +132,18 @@ TEST(EmbeddingTest, AddVectors) {
EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
}
+TEST(EmbeddingTest, SubtractVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 - E2;
+ EXPECT_THAT(E3, ElementsAre(0.5, 0.5, 4.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, SubtractVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -137,6 +161,15 @@ TEST(EmbeddingTest, ScaleVector) {
EXPECT_THAT(E1, ElementsAre(0.5, 1.0, 1.5));
}
+TEST(EmbeddingTest, ScaleVectorOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = E1 * 0.5f;
+ EXPECT_THAT(E2,
[llvm-branch-commits] [llvm] [IR2Vec] Overloading `operator+` for `Embeddings` (PR #145118)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/145118 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Add out-of-place arithmetic operators to Embedding class (PR #145118)
@@ -71,20 +71,43 @@ inline bool fromJSON(const llvm::json::Value &E, Embedding
&Out,
// Embedding
//===--===//
+Embedding Embedding::operator+(const Embedding &RHS) const {
+ assert(this->size() == RHS.size() && "Vectors must have the same dimension");
+ Embedding Result(*this);
+ std::transform(this->begin(), this->end(), RHS.begin(), Result.begin(),
+ std::plus());
+ return Result;
+}
+
Embedding &Embedding::operator+=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
svkeerthy wrote:
Implemented in terms of = as it would avoid copies in
=.
https://github.com/llvm/llvm-project/pull/145118
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Add out-of-place arithmetic operators to Embedding class (PR #145118)
https://github.com/svkeerthy updated
https://github.com/llvm/llvm-project/pull/145118
>From 10019cae162bb53e147797b655da75aac33b0a20 Mon Sep 17 00:00:00 2001
From: svkeerthy
Date: Fri, 20 Jun 2025 23:00:40 +
Subject: [PATCH] Overloading operator+ for Embeddngs
---
llvm/include/llvm/Analysis/IR2Vec.h| 9 --
llvm/lib/Analysis/IR2Vec.cpp | 19 -
llvm/unittests/Analysis/IR2VecTest.cpp | 39 ++
3 files changed, 63 insertions(+), 4 deletions(-)
diff --git a/llvm/include/llvm/Analysis/IR2Vec.h
b/llvm/include/llvm/Analysis/IR2Vec.h
index 040cb84ff27a1..ef8f630d7feb1 100644
--- a/llvm/include/llvm/Analysis/IR2Vec.h
+++ b/llvm/include/llvm/Analysis/IR2Vec.h
@@ -107,9 +107,12 @@ struct Embedding {
const std::vector &getData() const { return Data; }
/// Arithmetic operators
- Embedding &operator+=(const Embedding &RHS);
- Embedding &operator-=(const Embedding &RHS);
- Embedding &operator*=(double Factor);
+ LLVM_ABI Embedding &operator+=(const Embedding &RHS);
+ LLVM_ABI Embedding operator+(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator-=(const Embedding &RHS);
+ LLVM_ABI Embedding operator-(const Embedding &RHS) const;
+ LLVM_ABI Embedding &operator*=(double Factor);
+ LLVM_ABI Embedding operator*(double Factor) const;
/// Adds Src Embedding scaled by Factor with the called Embedding.
/// Called_Embedding += Src * Factor
diff --git a/llvm/lib/Analysis/IR2Vec.cpp b/llvm/lib/Analysis/IR2Vec.cpp
index 895b3de58a54e..bf456102bb618 100644
--- a/llvm/lib/Analysis/IR2Vec.cpp
+++ b/llvm/lib/Analysis/IR2Vec.cpp
@@ -70,7 +70,6 @@ inline bool fromJSON(const llvm::json::Value &E, Embedding
&Out,
//
==--===//
// Embedding
//===--===//
-
Embedding &Embedding::operator+=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -78,6 +77,12 @@ Embedding &Embedding::operator+=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator+(const Embedding &RHS) const {
+ Embedding Result(*this);
+ Result += RHS;
+ return Result;
+}
+
Embedding &Embedding::operator-=(const Embedding &RHS) {
assert(this->size() == RHS.size() && "Vectors must have the same dimension");
std::transform(this->begin(), this->end(), RHS.begin(), this->begin(),
@@ -85,12 +90,24 @@ Embedding &Embedding::operator-=(const Embedding &RHS) {
return *this;
}
+Embedding Embedding::operator-(const Embedding &RHS) const {
+ Embedding Result(*this);
+ Result -= RHS;
+ return Result;
+}
+
Embedding &Embedding::operator*=(double Factor) {
std::transform(this->begin(), this->end(), this->begin(),
[Factor](double Elem) { return Elem * Factor; });
return *this;
}
+Embedding Embedding::operator*(double Factor) const {
+ Embedding Result(*this);
+ Result *= Factor;
+ return Result;
+}
+
Embedding &Embedding::scaleAndAdd(const Embedding &Src, float Factor) {
assert(this->size() == Src.size() && "Vectors must have the same dimension");
for (size_t Itr = 0; Itr < this->size(); ++Itr)
diff --git a/llvm/unittests/Analysis/IR2VecTest.cpp
b/llvm/unittests/Analysis/IR2VecTest.cpp
index 3c97c20ae72d5..70d4808dc6d54 100644
--- a/llvm/unittests/Analysis/IR2VecTest.cpp
+++ b/llvm/unittests/Analysis/IR2VecTest.cpp
@@ -109,6 +109,18 @@ TEST(EmbeddingTest, ConstructorsAndAccessors) {
}
}
+TEST(EmbeddingTest, AddVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 + E2;
+ EXPECT_THAT(E3, ElementsAre(1.5, 3.5, 2.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, AddVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -120,6 +132,18 @@ TEST(EmbeddingTest, AddVectors) {
EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
}
+TEST(EmbeddingTest, SubtractVectorsOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = {0.5, 1.5, -1.0};
+
+ Embedding E3 = E1 - E2;
+ EXPECT_THAT(E3, ElementsAre(0.5, 0.5, 4.0));
+
+ // Check that E1 and E2 are unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.5, -1.0));
+}
+
TEST(EmbeddingTest, SubtractVectors) {
Embedding E1 = {1.0, 2.0, 3.0};
Embedding E2 = {0.5, 1.5, -1.0};
@@ -137,6 +161,15 @@ TEST(EmbeddingTest, ScaleVector) {
EXPECT_THAT(E1, ElementsAre(0.5, 1.0, 1.5));
}
+TEST(EmbeddingTest, ScaleVectorOutOfPlace) {
+ Embedding E1 = {1.0, 2.0, 3.0};
+ Embedding E2 = E1 * 0.5f;
+ EXPECT_THAT(E2, ElementsAre(0.5, 1.0, 1.5));
+
+ // Check that E1 is unchanged
+ EXPECT_THAT(E1, ElementsAre(1.0, 2.0, 3.0));
+}
+
TEST(EmbeddingTest, AddS
[llvm-branch-commits] [llvm] [IR2Vec] Add out-of-place arithmetic operators to Embedding class (PR #145118)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/145118 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][IR2Vec] Minor refactoring of opcode access in vocabulary (PR #147585)
https://github.com/svkeerthy ready_for_review https://github.com/llvm/llvm-project/pull/147585 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][IR2Vec] Minor refactoring of opcode access in vocabulary (PR #147585)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/147585 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][IR2Vec] Exposing helpers in IR2Vec Vocabulary (PR #147841)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/147841 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][IR2Vec] Exposing helpers in IR2Vec Vocabulary (PR #147841)
https://github.com/svkeerthy ready_for_review https://github.com/llvm/llvm-project/pull/147841 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [IR2Vec] Add llvm-ir2vec tool for generating triplet embeddings (PR #147842)
https://github.com/svkeerthy edited https://github.com/llvm/llvm-project/pull/147842 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [llvm] [NFC][IR2Vec] Minor refactoring of opcode access in vocabulary (PR #147585)
@@ -447,21 +453,18 @@ void IR2VecVocabAnalysis::generateNumMappedVocab() {
// Handle Opcodes
std::vector NumericOpcodeEmbeddings(Vocabulary::MaxOpcodes,
Embedding(Dim, 0));
-#define HANDLE_INST(NUM, OPCODE, CLASS)
\
- {
\
-auto It = OpcVocab.find(#OPCODE);
\
-if (It != OpcVocab.end())
\
- NumericOpcodeEmbeddings[NUM - 1] = It->second;
\
-else
\
- handleMissingEntity(#OPCODE);
\
+ for (unsigned Opcode : seq(0u, Vocabulary::MaxOpcodes)) {
svkeerthy wrote:
Without `u` it would be considered as `int` and will throw compilation errors.
https://github.com/llvm/llvm-project/pull/147585
___
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
