Skip to content

[flang][cuda] Add interface and lowering for fence_proxy_async #163014

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.

Sign up for GitHub

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

Merged
merged 3 commits into from
Oct 11, 2025
Merged

[flang][cuda] Add interface and lowering for fence_proxy_async #163014

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d400fe5
[flang][cuda] Add interface and lowering for fence_proxy_async
clementval Oct 11, 2025
d87528f
Merge branch 'main' into cuf_fence_proxy
clementval Oct 11, 2025
070d11e
Update flang/test/Lower/CUDA/cuda-device-proc.cuf
clementval Oct 11, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions flang/include/flang/Optimizer/Builder/IntrinsicCall.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -274,6 +274,7 @@ struct IntrinsicLibrary {
llvm::ArrayRef<fir::ExtendedValue>);
template <Extremum, ExtremumBehavior>
mlir::Value genExtremum(mlir::Type, llvm::ArrayRef<mlir::Value>);
void genFenceProxyAsync(llvm::ArrayRef<fir::ExtendedValue>);
mlir::Value genFloor(mlir::Type, llvm::ArrayRef<mlir::Value>);
mlir::Value genFraction(mlir::Type resultType,
mlir::ArrayRef<mlir::Value> args);
Expand Down
15 changes: 15 additions & 0 deletions flang/lib/Optimizer/Builder/IntrinsicCall.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -502,6 +502,10 @@ static constexpr IntrinsicHandler handlers[]{
&I::genExtendsTypeOf,
{{{"a", asBox}, {"mold", asBox}}},
/*isElemental=*/false},
{"fence_proxy_async",
&I::genFenceProxyAsync,
{},
/*isElemental=*/false},
{"findloc",
&I::genFindloc,
{{{"array", asBox},
Expand Down Expand Up @@ -4354,6 +4358,17 @@ IntrinsicLibrary::genExtendsTypeOf(mlir::Type resultType,
fir::getBase(args[1])));
}

// FENCE_PROXY_ASYNC (CUDA)
void IntrinsicLibrary::genFenceProxyAsync(
llvm::ArrayRef<fir::ExtendedValue> args) {
assert(args.size() == 0);
auto kind = mlir::NVVM::ProxyKindAttr::get(
builder.getContext(), mlir::NVVM::ProxyKind::async_shared);
auto space = mlir::NVVM::SharedSpaceAttr::get(
builder.getContext(), mlir::NVVM::SharedSpace::shared_cta);
mlir::NVVM::FenceProxyOp::create(builder, loc, kind, space);
}

// FINDLOC
fir::ExtendedValue
IntrinsicLibrary::genFindloc(mlir::Type resultType,
Expand Down
5 changes: 5 additions & 0 deletions flang/module/cudadevice.f90
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2008,6 +2008,11 @@ attributes(device) function barrier_arrive_cnt(barrier, count) result(token)
end function
end interface

interface
attributes(device) subroutine fence_proxy_async()
end subroutine
end interface

contains

attributes(device) subroutine syncthreads()
Expand Down
7 changes: 7 additions & 0 deletions flang/test/Lower/CUDA/cuda-device-proc.cuf
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -419,3 +419,10 @@ end subroutine
! CHECK: %[[LLVM_PTR:.*]] = fir.convert %[[DECL_SHARED]]#0 : (!fir.ref<i64>) -> !llvm.ptr
! CHECK: %[[SHARED_PTR:.*]] = llvm.addrspacecast %[[LLVM_PTR]] : !llvm.ptr to !llvm.ptr<3>
! CHECK: nvvm.mbarrier.arrive.expect_tx %[[SHARED_PTR]], %{{.*}} : !llvm.ptr<3>, i32

attributes(global) subroutine test_fence()
call fence_proxy_async()
end subroutine

! CHECK-LABEL: func.func @_QPtest_fence()
! CHECK: nvvm.fence.proxy {kind = #nvvm.proxy_kind<async.shared>, space = #nvvm.shared_space<cta>}
Loading