domada updated this revision to Diff 465332.
domada added a comment.
Patch rebased.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D133578/new/
https://reviews.llvm.org/D133578
Files:
clang/lib/CodeGen/CGStmtOpenMP.cpp
llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h
llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
llvm/unittests/Frontend/OpenMPIRBuilderTest.cpp
mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
Index: mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
===================================================================
--- mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
+++ mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
@@ -996,8 +996,9 @@
if (llvm::Optional<uint64_t> safelenVar = loop.getSafelen())
safelen = builder.getInt64(safelenVar.value());
+ llvm::DenseMap<llvm::Value *, llvm::Value *> alignedVars;
ompBuilder->applySimd(
- loopInfo,
+ loopInfo, alignedVars,
loop.getIfExpr() ? moduleTranslation.lookupValue(loop.getIfExpr())
: nullptr,
llvm::omp::OrderKind::OMP_ORDER_unknown, simdlen, safelen);
Index: llvm/unittests/Frontend/OpenMPIRBuilderTest.cpp
===================================================================
--- llvm/unittests/Frontend/OpenMPIRBuilderTest.cpp
+++ llvm/unittests/Frontend/OpenMPIRBuilderTest.cpp
@@ -1767,11 +1767,12 @@
TEST_F(OpenMPIRBuilderTest, ApplySimd) {
OpenMPIRBuilder OMPBuilder(*M);
-
+ DenseMap<Value *, Value *> AlignedVars;
CanonicalLoopInfo *CLI = buildSingleLoopFunction(DL, OMPBuilder, 32);
// Simd-ize the loop.
- OMPBuilder.applySimd(CLI, /* IfCond */ nullptr, OrderKind::OMP_ORDER_unknown,
+ OMPBuilder.applySimd(CLI, AlignedVars, /* IfCond */ nullptr,
+ OrderKind::OMP_ORDER_unknown,
/* Simdlen */ nullptr,
/* Safelen */ nullptr);
@@ -1798,13 +1799,84 @@
}));
}
-TEST_F(OpenMPIRBuilderTest, ApplySimdlen) {
+TEST_F(OpenMPIRBuilderTest, ApplySimdCustomAligned) {
OpenMPIRBuilder OMPBuilder(*M);
+ IRBuilder<> Builder(BB);
+ const int AlignmentValue = 32;
+ AllocaInst *Alloc1 =
+ Builder.CreateAlloca(Builder.getInt8PtrTy(), Builder.getInt64(1));
+ AllocaInst *Alloc2 = Builder.CreateAlloca(
+ ArrayType::get(Builder.getInt32Ty(), 10), Builder.getInt64(1));
+ DenseMap<Value *, Value *> AlignedVars;
+ auto Int8Ty = Builder.getInt8Ty();
+ Instruction *MallocInstr = CallInst::CreateMalloc(
+ Alloc2, Builder.getInt64Ty(), Int8Ty, ConstantExpr::getSizeOf(Int8Ty),
+ Builder.getInt64(400), nullptr, "");
+ Builder.CreateStore(MallocInstr, Alloc1);
+
+ AlignedVars.insert({Alloc1, Builder.getInt64(AlignmentValue)});
+ AlignedVars.insert({Alloc2, Builder.getInt64(AlignmentValue)});
CanonicalLoopInfo *CLI = buildSingleLoopFunction(DL, OMPBuilder, 32);
// Simd-ize the loop.
- OMPBuilder.applySimd(CLI, /* IfCond */ nullptr, OrderKind::OMP_ORDER_unknown,
+ OMPBuilder.applySimd(CLI, AlignedVars, /* IfCond */ nullptr,
+ OrderKind::OMP_ORDER_unknown,
+ /* Simdlen */ nullptr,
+ /* Safelen */ nullptr);
+
+ OMPBuilder.finalize();
+ EXPECT_FALSE(verifyModule(*M, &errs()));
+
+ PassBuilder PB;
+ FunctionAnalysisManager FAM;
+ PB.registerFunctionAnalyses(FAM);
+ LoopInfo &LI = FAM.getResult<LoopAnalysis>(*F);
+
+ const std::vector<Loop *> &TopLvl = LI.getTopLevelLoops();
+ EXPECT_EQ(TopLvl.size(), 1u);
+
+ Loop *L = TopLvl.front();
+ EXPECT_TRUE(findStringMetadataForLoop(L, "llvm.loop.parallel_accesses"));
+ EXPECT_TRUE(getBooleanLoopAttribute(L, "llvm.loop.vectorize.enable"));
+
+ // Check for llvm.access.group metadata attached to the printf
+ // function in the loop body.
+ BasicBlock *LoopBody = CLI->getBody();
+ EXPECT_TRUE(any_of(*LoopBody, [](Instruction &I) {
+ return I.getMetadata("llvm.access.group") != nullptr;
+ }));
+
+ // Check if number of assumption instructions is equal to number of aligned
+ // variables
+ BasicBlock *LoopPreheader = CLI->getPreheader();
+ size_t NumAssummptionCallsInPreheader = count_if(
+ *LoopPreheader, [](Instruction &I) { return isa<AssumeInst>(I); });
+ EXPECT_EQ(NumAssummptionCallsInPreheader, AlignedVars.size());
+
+ // Check if variables are correctly aligned
+ for (auto &Instr : LoopPreheader->getInstList()) {
+ if (isa<AssumeInst>(Instr)) {
+ auto AssumeInstruction = dyn_cast<AssumeInst>(&Instr);
+ if (AssumeInstruction->getNumTotalBundleOperands()) {
+ auto Bundle = AssumeInstruction->getOperandBundleAt(0);
+ if (Bundle.getTagName() == "align") {
+ EXPECT_TRUE(isa<ConstantInt>(Bundle.Inputs[1]));
+ auto ConstIntVal = dyn_cast<ConstantInt>(Bundle.Inputs[1]);
+ EXPECT_EQ(ConstIntVal->getSExtValue(), AlignmentValue);
+ }
+ }
+ }
+ }
+}
+TEST_F(OpenMPIRBuilderTest, ApplySimdlen) {
+ OpenMPIRBuilder OMPBuilder(*M);
+ DenseMap<Value *, Value *> AlignedVars;
+ CanonicalLoopInfo *CLI = buildSingleLoopFunction(DL, OMPBuilder, 32);
+
+ // Simd-ize the loop.
+ OMPBuilder.applySimd(CLI, AlignedVars,
+ /* IfCond */ nullptr, OrderKind::OMP_ORDER_unknown,
ConstantInt::get(Type::getInt32Ty(Ctx), 3),
/* Safelen */ nullptr);
@@ -1834,12 +1906,13 @@
TEST_F(OpenMPIRBuilderTest, ApplySafelenOrderConcurrent) {
OpenMPIRBuilder OMPBuilder(*M);
+ DenseMap<Value *, Value *> AlignedVars;
CanonicalLoopInfo *CLI = buildSingleLoopFunction(DL, OMPBuilder, 32);
// Simd-ize the loop.
OMPBuilder.applySimd(
- CLI, /* IfCond */ nullptr, OrderKind::OMP_ORDER_concurrent,
+ CLI, AlignedVars, /* IfCond */ nullptr, OrderKind::OMP_ORDER_concurrent,
/* Simdlen */ nullptr, ConstantInt::get(Type::getInt32Ty(Ctx), 3));
OMPBuilder.finalize();
@@ -1870,13 +1943,13 @@
TEST_F(OpenMPIRBuilderTest, ApplySafelen) {
OpenMPIRBuilder OMPBuilder(*M);
+ DenseMap<Value *, Value *> AlignedVars;
CanonicalLoopInfo *CLI = buildSingleLoopFunction(DL, OMPBuilder, 32);
- // Simd-ize the loop.
- OMPBuilder.applySimd(CLI, /* IfCond */ nullptr, OrderKind::OMP_ORDER_unknown,
- /* Simdlen */ nullptr,
- ConstantInt::get(Type::getInt32Ty(Ctx), 3));
+ OMPBuilder.applySimd(
+ CLI, AlignedVars, /* IfCond */ nullptr, OrderKind::OMP_ORDER_unknown,
+ /* Simdlen */ nullptr, ConstantInt::get(Type::getInt32Ty(Ctx), 3));
OMPBuilder.finalize();
EXPECT_FALSE(verifyModule(*M, &errs()));
@@ -1904,11 +1977,12 @@
TEST_F(OpenMPIRBuilderTest, ApplySimdlenSafelen) {
OpenMPIRBuilder OMPBuilder(*M);
+ DenseMap<Value *, Value *> AlignedVars;
CanonicalLoopInfo *CLI = buildSingleLoopFunction(DL, OMPBuilder, 32);
- // Simd-ize the loop.
- OMPBuilder.applySimd(CLI, /* IfCond */ nullptr, OrderKind::OMP_ORDER_unknown,
+ OMPBuilder.applySimd(CLI, AlignedVars, /* IfCond */ nullptr,
+ OrderKind::OMP_ORDER_unknown,
ConstantInt::get(Type::getInt32Ty(Ctx), 2),
ConstantInt::get(Type::getInt32Ty(Ctx), 3));
@@ -1939,6 +2013,7 @@
TEST_F(OpenMPIRBuilderTest, ApplySimdLoopIf) {
OpenMPIRBuilder OMPBuilder(*M);
IRBuilder<> Builder(BB);
+ DenseMap<Value *, Value *> AlignedVars;
AllocaInst *Alloc1 = Builder.CreateAlloca(Builder.getInt32Ty());
AllocaInst *Alloc2 = Builder.CreateAlloca(Builder.getInt32Ty());
@@ -1953,7 +2028,7 @@
CanonicalLoopInfo *CLI = buildSingleLoopFunction(DL, OMPBuilder, 32);
// Simd-ize the loop with if condition
- OMPBuilder.applySimd(CLI, IfCmp, OrderKind::OMP_ORDER_unknown,
+ OMPBuilder.applySimd(CLI, AlignedVars, IfCmp, OrderKind::OMP_ORDER_unknown,
ConstantInt::get(Type::getInt32Ty(Ctx), 3),
/* Safelen */ nullptr);
Index: llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
===================================================================
--- llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
+++ llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
@@ -3005,9 +3005,10 @@
Builder.CreateBr(NewBlocks.front());
}
-void OpenMPIRBuilder::applySimd(CanonicalLoopInfo *CanonicalLoop, Value *IfCond,
- OrderKind Order, ConstantInt *Simdlen,
- ConstantInt *Safelen) {
+void OpenMPIRBuilder::applySimd(CanonicalLoopInfo *CanonicalLoop,
+ DenseMap<Value *, Value *> AlignedVars,
+ Value *IfCond, OrderKind Order,
+ ConstantInt *Simdlen, ConstantInt *Safelen) {
LLVMContext &Ctx = Builder.getContext();
Function *F = CanonicalLoop->getFunction();
@@ -3026,6 +3027,31 @@
Loop *L = LI.getLoopFor(CanonicalLoop->getHeader());
+ for (auto &AlignedItem : AlignedVars) {
+ assert(isa<AllocaInst>(AlignedItem.first) &&
+ "Value which needs to be aligned must represented by alloca "
+ "instruction");
+ assert(AlignedItem.second &&
+ "Alignment should be specified for all aligned variables");
+ Value *AlignPtrInstruction = nullptr;
+ AllocaInst *AllocaInstruction = dyn_cast<AllocaInst>(AlignedItem.first);
+ Builder.SetInsertPoint(CanonicalLoop->getPreheader()->getTerminator());
+ Type *AllocatedVarType = AllocaInstruction->getAllocatedType();
+ if (isa<ArrayType>(AllocatedVarType))
+ AlignPtrInstruction = Builder.CreateInBoundsGEP(
+ AllocaInstruction->getAllocatedType(), AllocaInstruction,
+ SmallVector<Value *>{Builder.getInt64(0), Builder.getInt64(0)},
+ "arraydecay");
+ else if (isa<PointerType>(AllocatedVarType))
+ AlignPtrInstruction = Builder.CreateLoad(
+ AllocaInstruction->getAllocatedType(), AllocaInstruction);
+ assert(AlignPtrInstruction &&
+ "Aligned variables must be either pointer or array type");
+
+ Builder.CreateAlignmentAssumption(F->getParent()->getDataLayout(),
+ AlignPtrInstruction, AlignedItem.second);
+ }
+
if (IfCond) {
ValueToValueMapTy VMap;
createIfVersion(CanonicalLoop, IfCond, VMap, "simd");
Index: llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h
===================================================================
--- llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h
+++ llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h
@@ -617,13 +617,17 @@
/// to the cloned loop. The cloned loop is executed when ifCond is evaluated
/// to false.
///
- /// \param Loop The loop to simd-ize.
- /// \param IfCond The value which corresponds to the if clause condition.
- /// \param Order The enum to map order clause
- /// \param Simdlen The Simdlen length to apply to the simd loop.
- /// \param Safelen The Safelen length to apply to the simd loop.
- void applySimd(CanonicalLoopInfo *Loop, Value *IfCond, omp::OrderKind Order,
- ConstantInt *Simdlen, ConstantInt *Safelen);
+ /// \param Loop The loop to simd-ize.
+ /// \param AlignedVars The map of the variables which need to aligned with.
+ /// \param IfCond The value which corresponds to the if clause
+ /// condition.
+ /// \param Order The enum to map order clause.
+ /// \param Simdlen The Simdlen length to apply to the simd loop.
+ /// \param Safelen The Safelen length to apply to the simd loop.
+ void applySimd(CanonicalLoopInfo *Loop,
+ llvm::DenseMap<llvm::Value *, llvm::Value *> AlignedVars,
+ Value *IfCond, omp::OrderKind Order, ConstantInt *Simdlen,
+ ConstantInt *Safelen);
/// Generator for '#omp flush'
///
Index: clang/lib/CodeGen/CGStmtOpenMP.cpp
===================================================================
--- clang/lib/CodeGen/CGStmtOpenMP.cpp
+++ clang/lib/CodeGen/CGStmtOpenMP.cpp
@@ -2637,6 +2637,7 @@
PrePostActionTy &) {
// Use the OpenMPIRBuilder if enabled.
if (UseOMPIRBuilder) {
+ llvm::DenseMap<llvm::Value *, llvm::Value *> AlignedVars;
// Emit the associated statement and get its loop representation.
const Stmt *Inner = S.getRawStmt();
llvm::CanonicalLoopInfo *CLI =
@@ -2669,7 +2670,8 @@
}
// Add simd metadata to the collapsed loop. Do not generate
// another loop for if clause. Support for if clause is done earlier.
- OMPBuilder.applySimd(CLI, /*IfCond*/ nullptr, Order, Simdlen, Safelen);
+ OMPBuilder.applySimd(CLI, AlignedVars,
+ /*IfCond*/ nullptr, Order, Simdlen, Safelen);
return;
}
};
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
