-
Notifications
You must be signed in to change notification settings - Fork 10.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[flang][cuda] Handle lowering of stars in cuf kernel launch parameters #85695
Conversation
@llvm/pr-subscribers-flang-fir-hlfir Author: Valentin Clement (バレンタイン クレメン) (clementval) ChangesParsing of the cuf kernel loop directive has been updated to handle variants with the * syntax. This patch updates the lowering to make use of them.
Full diff: https://github.com/llvm/llvm-project/pull/85695.diff 2 Files Affected:
diff --git a/flang/lib/Lower/Bridge.cpp b/flang/lib/Lower/Bridge.cpp
index 650ec5db2d0ccb..1b9a8a867b0804 100644
--- a/flang/lib/Lower/Bridge.cpp
+++ b/flang/lib/Lower/Bridge.cpp
@@ -2529,23 +2529,42 @@ class FirConverter : public Fortran::lower::AbstractConverter {
const std::optional<Fortran::parser::ScalarIntExpr> &stream =
std::get<3>(dir.t);
+ auto isOnlyStars =
+ [&](const std::list<Fortran::parser::CUFKernelDoConstruct::StarOrExpr>
+ &list) -> bool {
+ for (const Fortran::parser::CUFKernelDoConstruct::StarOrExpr &expr :
+ list) {
+ if (expr.v)
+ return false;
+ }
+ return true;
+ };
+
+ mlir::Value zero =
+ builder->createIntegerConstant(loc, builder->getI32Type(), 0);
+
llvm::SmallVector<mlir::Value> gridValues;
- for (const Fortran::parser::CUFKernelDoConstruct::StarOrExpr &expr : grid) {
- if (expr.v) {
- gridValues.push_back(fir::getBase(
- genExprValue(*Fortran::semantics::GetExpr(*expr.v), stmtCtx)));
- } else {
- // TODO: '*'
+ 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;
- for (const Fortran::parser::CUFKernelDoConstruct::StarOrExpr &expr :
- block) {
- if (expr.v) {
- blockValues.push_back(fir::getBase(
- genExprValue(*Fortran::semantics::GetExpr(*expr.v), stmtCtx)));
- } else {
- // TODO: '*'
+ 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);
+ }
}
}
mlir::Value streamValue;
diff --git a/flang/test/Lower/CUDA/cuda-kernel-loop-directive.cuf b/flang/test/Lower/CUDA/cuda-kernel-loop-directive.cuf
index c017561447f85d..6179e609db383c 100644
--- a/flang/test/Lower/CUDA/cuda-kernel-loop-directive.cuf
+++ b/flang/test/Lower/CUDA/cuda-kernel-loop-directive.cuf
@@ -42,7 +42,20 @@ subroutine sub1()
! CHECK: fir.cuda_kernel<<<%c1{{.*}}, (%c256{{.*}}, %c1{{.*}})>>> (%{{.*}} : index, %{{.*}} : index) = (%{{.*}}, %{{.*}} : index, index) to (%{{.*}}, %{{.*}} : index, index) step (%{{.*}}, %{{.*}} : index, index)
! CHECK: {n = 2 : i64}
-! TODO: lowering for these cases
-! !$cuf kernel do(2) <<< (1,*), (256,1) >>>
-! !$cuf kernel do(2) <<< (*,*), (32,4) >>>
+ !$cuf kernel do(2) <<< (1,*), (256,1) >>>
+ do i = 1, n
+ do j = 1, n
+ c(i,j) = c(i,j) * d(i,j)
+ end do
+ end do
+! CHECK: fir.cuda_kernel<<<(%c1{{.*}}, %c0{{.*}}), (%c256{{.*}}, %c1{{.*}})>>> (%{{.*}} : index, %{{.*}} : index) = (%{{.*}}, %{{.*}} : index, index) to (%{{.*}}, %{{.*}} : index, index) step (%{{.*}}, %{{.*}} : index, index)
+
+!$cuf kernel do(2) <<< (*,*), (32,4) >>>
+ do i = 1, n
+ do j = 1, n
+ c(i,j) = c(i,j) * d(i,j)
+ end do
+ end do
+
+! CHECK: fir.cuda_kernel<<<*, (%c32{{.*}}, %c4{{.*}})>>> (%{{.*}} : index, %{{.*}} : index) = (%{{.*}}, %{{.*}} : index, index) to (%{{.*}}, %{{.*}} : index, index) step (%{{.*}}, %{{.*}} : index, index)
end
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM!
Adding a comment about special meaning of 0
into the operation description might be useful.
Thanks! |
llvm#85695) Parsing of the cuf kernel loop directive has been updated to handle variants with the * syntax. This patch updates the lowering to make use of them. - If the grid or block syntax uses only stars then the operation variadic operand remains empty. - If there is values and stars, then stars are represented as a zero constant value.
Parsing of the cuf kernel loop directive has been updated to handle variants with the * syntax. This patch updates the lowering to make use of them.