Skip to content

[NativeCPU] Handle local args. #2512

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 1 commit into from
Jan 8, 2025
Merged

[NativeCPU] Handle local args. #2512

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
8 changes: 4 additions & 4 deletions source/adapters/native_cpu/enqueue.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -138,12 +138,12 @@ UR_APIEXPORT ur_result_t UR_APICALL urEnqueueKernelLaunch(
#else
bool isLocalSizeOne =
ndr.LocalSize[0] == 1 && ndr.LocalSize[1] == 1 && ndr.LocalSize[2] == 1;
if (isLocalSizeOne && ndr.GlobalSize[0] > numParallelThreads) {
if (isLocalSizeOne && ndr.GlobalSize[0] > numParallelThreads &&
!hKernel->hasLocalArgs()) {
// If the local size is one, we make the assumption that we are running a
// parallel_for over a sycl::range.
// Todo: we could add compiler checks and
// kernel properties for this (e.g. check that no barriers are called, no
// local memory args).
// Todo: we could add more compiler checks and
// kernel properties for this (e.g. check that no barriers are called).

// Todo: this assumes that dim 0 is the best dimension over which we want to
// parallelize
Expand Down
4 changes: 3 additions & 1 deletion source/adapters/native_cpu/kernel.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -142,7 +142,9 @@ struct ur_kernel_handle_t_ : RefCounted {
_localMemPoolSize = reqSize;
}

// To be called before executing a work group
bool hasLocalArgs() const { return !_localArgInfo.empty(); }

// To be called before executing a work group if local args are present
void handleLocalArgs(size_t numParallelThread, size_t threadId) {
// For each local argument we have size*numthreads
size_t offset = 0;
Expand Down
Loading