[flang][cuda] Make launch configuration optional for cuf kernel #115947
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
llvmbot
added
flang
|
@llvm/pr-subscribers-flang-parser @llvm/pr-subscribers-flang-fir-hlfir Author: Valentin Clement (バレンタイン クレメン) (clementval) ChangesLaunch configuration on This patch updates the parse tree, parser and lowering to take this into account. Full diff: https://github.com/llvm/llvm-project/pull/115947.diff 6 Files Affected:
diff --git a/flang/include/flang/Parser/dump-parse-tree.h b/flang/include/flang/Parser/dump-parse-tree.h
index 4bbf9777a54ccb..2fb863738d62d0 100644
--- a/flang/include/flang/Parser/dump-parse-tree.h
+++ b/flang/include/flang/Parser/dump-parse-tree.h
@@ -236,6 +236,7 @@ class ParseTreeDumper {
NODE(parser, CUFKernelDoConstruct)
NODE(CUFKernelDoConstruct, StarOrExpr)
NODE(CUFKernelDoConstruct, Directive)
+ NODE(CUFKernelDoConstruct, LaunchConfiguration)
NODE(parser, CUFReduction)
NODE(parser, CycleStmt)
NODE(parser, DataComponentDefStmt)
diff --git a/flang/include/flang/Parser/parse-tree.h b/flang/include/flang/Parser/parse-tree.h
index 5f5650304f9987..ce0b6167de9fc8 100644
--- a/flang/include/flang/Parser/parse-tree.h
+++ b/flang/include/flang/Parser/parse-tree.h
@@ -4527,12 +4527,17 @@ struct CUFReduction {
struct CUFKernelDoConstruct {
TUPLE_CLASS_BOILERPLATE(CUFKernelDoConstruct);
WRAPPER_CLASS(StarOrExpr, std::optional<ScalarIntExpr>);
+ struct LaunchConfiguration {
+ TUPLE_CLASS_BOILERPLATE(LaunchConfiguration);
+ std::tuple<std::list<StarOrExpr>, std::list<StarOrExpr>,
+ std::optional<ScalarIntExpr>>
+ t;
+ };
struct Directive {
TUPLE_CLASS_BOILERPLATE(Directive);
CharBlock source;
- std::tuple<std::optional<ScalarIntConstantExpr>, std::list<StarOrExpr>,
- std::list<StarOrExpr>, std::optional<ScalarIntExpr>,
- std::list<CUFReduction>>
+ std::tuple<std::optional<ScalarIntConstantExpr>,
+ std::optional<LaunchConfiguration>, std::list<CUFReduction>>
t;
};
std::tuple<Directive, std::optional<DoConstruct>> t;
diff --git a/flang/lib/Lower/Bridge.cpp b/flang/lib/Lower/Bridge.cpp
index 0e3011e73902da..da53edf7e734b0 100644
--- a/flang/lib/Lower/Bridge.cpp
+++ b/flang/lib/Lower/Bridge.cpp
@@ -2862,14 +2862,11 @@ class FirConverter : public Fortran::lower::AbstractConverter {
if (nestedLoops > 1)
n = builder->getIntegerAttr(builder->getI64Type(), nestedLoops);
- const std::list<Fortran::parser::CUFKernelDoConstruct::StarOrExpr> &grid =
- std::get<1>(dir.t);
- const std::list<Fortran::parser::CUFKernelDoConstruct::StarOrExpr> &block =
- std::get<2>(dir.t);
- const std::optional<Fortran::parser::ScalarIntExpr> &stream =
- std::get<3>(dir.t);
+ const auto &launchConfig = std::get<std::optional<
+ Fortran::parser::CUFKernelDoConstruct::LaunchConfiguration>>(dir.t);
+
const std::list<Fortran::parser::CUFReduction> &cufreds =
- std::get<4>(dir.t);
+ std::get<2>(dir.t);
llvm::SmallVector<mlir::Value> reduceOperands;
llvm::SmallVector<mlir::Attribute> reduceAttrs;
@@ -2913,35 +2910,45 @@ class FirConverter : public Fortran::lower::AbstractConverter {
builder->createIntegerConstant(loc, builder->getI32Type(), 0);
llvm::SmallVector<mlir::Value> gridValues;
- if (!isOnlyStars(grid)) {
- for (const Fortran::parser::CUFKernelDoConstruct::StarOrExpr &expr :
- grid) {
- if (expr.v) {
- gridValues.push_back(fir::getBase(
- genExprValue(*Fortran::semantics::GetExpr(*expr.v), stmtCtx)));
- } else {
- gridValues.push_back(zero);
+ llvm::SmallVector<mlir::Value> blockValues;
+ mlir::Value streamValue;
+
+ if (launchConfig) {
+ const std::list<Fortran::parser::CUFKernelDoConstruct::StarOrExpr> &grid =
+ std::get<0>(launchConfig->t);
+ const std::list<Fortran::parser::CUFKernelDoConstruct::StarOrExpr>
+ &block = std::get<1>(launchConfig->t);
+ const std::optional<Fortran::parser::ScalarIntExpr> &stream =
+ std::get<2>(launchConfig->t);
+ if (!isOnlyStars(grid)) {
+ for (const Fortran::parser::CUFKernelDoConstruct::StarOrExpr &expr :
+ grid) {
+ if (expr.v) {
+ gridValues.push_back(fir::getBase(
+ genExprValue(*Fortran::semantics::GetExpr(*expr.v), stmtCtx)));
+ } else {
+ gridValues.push_back(zero);
+ }
}
}
- }
- llvm::SmallVector<mlir::Value> blockValues;
- if (!isOnlyStars(block)) {
- for (const Fortran::parser::CUFKernelDoConstruct::StarOrExpr &expr :
- block) {
- if (expr.v) {
- blockValues.push_back(fir::getBase(
- genExprValue(*Fortran::semantics::GetExpr(*expr.v), stmtCtx)));
- } else {
- blockValues.push_back(zero);
+ if (!isOnlyStars(block)) {
+ for (const Fortran::parser::CUFKernelDoConstruct::StarOrExpr &expr :
+ block) {
+ if (expr.v) {
+ blockValues.push_back(fir::getBase(
+ genExprValue(*Fortran::semantics::GetExpr(*expr.v), stmtCtx)));
+ } else {
+ blockValues.push_back(zero);
+ }
}
}
+
+ if (stream)
+ streamValue = builder->createConvert(
+ loc, builder->getI32Type(),
+ fir::getBase(
+ genExprValue(*Fortran::semantics::GetExpr(*stream), stmtCtx)));
}
- mlir::Value streamValue;
- if (stream)
- streamValue = builder->createConvert(
- loc, builder->getI32Type(),
- fir::getBase(
- genExprValue(*Fortran::semantics::GetExpr(*stream), stmtCtx)));
const auto &outerDoConstruct =
std::get<std::optional<Fortran::parser::DoConstruct>>(kernel.t);
diff --git a/flang/lib/Parser/executable-parsers.cpp b/flang/lib/Parser/executable-parsers.cpp
index 5057e89164c9f2..730165613d91db 100644
--- a/flang/lib/Parser/executable-parsers.cpp
+++ b/flang/lib/Parser/executable-parsers.cpp
@@ -563,11 +563,15 @@ TYPE_PARSER(("REDUCTION"_tok || "REDUCE"_tok) >>
parenthesized(construct<CUFReduction>(Parser<CUFReduction::Operator>{},
":" >> nonemptyList(scalar(variable)))))
+TYPE_PARSER("<<<" >>
+ construct<CUFKernelDoConstruct::LaunchConfiguration>(gridOrBlock,
+ "," >> gridOrBlock,
+ maybe((", 0 ,"_tok || ", STREAM ="_tok) >> scalarIntExpr) / ">>>"))
+
TYPE_PARSER(sourced(beginDirective >> "$CUF KERNEL DO"_tok >>
construct<CUFKernelDoConstruct::Directive>(
- maybe(parenthesized(scalarIntConstantExpr)), "<<<" >> gridOrBlock,
- "," >> gridOrBlock,
- maybe((", 0 ,"_tok || ", STREAM ="_tok) >> scalarIntExpr) / ">>>",
+ maybe(parenthesized(scalarIntConstantExpr)),
+ maybe(Parser<CUFKernelDoConstruct::LaunchConfiguration>{}),
many(Parser<CUFReduction>{}) / endDirective)))
TYPE_CONTEXT_PARSER("!$CUF KERNEL DO construct"_en_US,
extension<LanguageFeature::CUDA>(construct<CUFKernelDoConstruct>(
diff --git a/flang/lib/Parser/unparse.cpp b/flang/lib/Parser/unparse.cpp
index 20022f8fa984ce..4b511da69832c5 100644
--- a/flang/lib/Parser/unparse.cpp
+++ b/flang/lib/Parser/unparse.cpp
@@ -2932,11 +2932,9 @@ class UnparseVisitor {
Word("*");
}
}
- void Unparse(const CUFKernelDoConstruct::Directive &x) {
- Word("!$CUF KERNEL DO");
- Walk(" (", std::get<std::optional<ScalarIntConstantExpr>>(x.t), ")");
+ void Unparse(const CUFKernelDoConstruct::LaunchConfiguration &x) {
Word(" <<<");
- const auto &grid{std::get<1>(x.t)};
+ const auto &grid{std::get<0>(x.t)};
if (grid.empty()) {
Word("*");
} else if (grid.size() == 1) {
@@ -2945,7 +2943,7 @@ class UnparseVisitor {
Walk("(", grid, ",", ")");
}
Word(",");
- const auto &block{std::get<2>(x.t)};
+ const auto &block{std::get<1>(x.t)};
if (block.empty()) {
Word("*");
} else if (block.size() == 1) {
@@ -2953,10 +2951,16 @@ class UnparseVisitor {
} else {
Walk("(", block, ",", ")");
}
- if (const auto &stream{std::get<3>(x.t)}) {
+ if (const auto &stream{std::get<2>(x.t)}) {
Word(",STREAM="), Walk(*stream);
}
Word(">>>");
+ }
+ void Unparse(const CUFKernelDoConstruct::Directive &x) {
+ Word("!$CUF KERNEL DO");
+ Walk(" (", std::get<std::optional<ScalarIntConstantExpr>>(x.t), ")");
+ Walk(std::get<std::optional<CUFKernelDoConstruct::LaunchConfiguration>>(
+ x.t));
Walk(" ", std::get<std::list<CUFReduction>>(x.t), " ");
Word("\n");
}
diff --git a/flang/test/Parser/cuf-sanity-common b/flang/test/Parser/cuf-sanity-common
index 7005ef07b22650..816e03bed7220a 100644
--- a/flang/test/Parser/cuf-sanity-common
+++ b/flang/test/Parser/cuf-sanity-common
@@ -31,6 +31,9 @@ module m
!$cuf kernel do <<<1, (2, 3), stream = 1>>>
do j = 1, 10
end do
+ !$cuf kernel do
+ do j = 1, 10
+ end do
!$cuf kernel do <<<*, *>>> reduce(+:x,y) reduce(*:z)
do j = 1, 10
x = x + a(j)
|
klausler
approved these changes
Nov 12, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Launch configuration on
CUF KERNEL DOdirective is optional as shown in example code from https://docs.nvidia.com/hpc-sdk/compilers/cuda-fortran-prog-guide/index.html#cfpg-cuda-fort-host-dev-codeThis patch updates the parse tree, parser and lowering to take this into account.