Skip to content

[SYCL][E2E] Enable max_work_group_query test for OpenCL - #22920

Open
VerenaBeckham wants to merge 3 commits into
intel:syclfrom
VerenaBeckham:verena/enable_test_upstream
Open

[SYCL][E2E] Enable max_work_group_query test for OpenCL#22920
VerenaBeckham wants to merge 3 commits into
intel:syclfrom
VerenaBeckham:verena/enable_test_upstream

Conversation

@VerenaBeckham

Copy link
Copy Markdown
Contributor

This test was marked as not supported on OpenCL in error in commit f19914a.

This test was marked as not supported on OpenCL in error in commit f19914a.
@VerenaBeckham
VerenaBeckham requested a review from a team as a code owner August 11, 2026 13:02
@VerenaBeckham

Copy link
Copy Markdown
Contributor Author

Tagging @abagusetty as the author of the commit referenced.

Comment thread sycl/test-e2e/Basic/max_work_group_query.cpp
@abagusetty

Copy link
Copy Markdown
Contributor

Tagging @abagusetty as the author of the commit referenced.

@VerenaBeckham IIRC back then, OpenCL didnt have the capability. Thanks for checking!

@VerenaBeckham

Copy link
Copy Markdown
Contributor Author

Tagging @abagusetty as the author of the commit referenced.

@VerenaBeckham IIRC back then, OpenCL didnt have the capability. Thanks for checking!

Actually, I was working under the assumption that there was no OpenCL capability, but the UR adapter just returns MAX_INT anyway, see

case UR_DEVICE_INFO_MAX_WORK_GROUPS_3D: {
. It's possible this wasn't implemented yet back then.
I see now that they have added CL_DEVICE_MAX_WORK_GROUP_SIZES since, so the adapter could be modified to get the correct value from OpenCL.
Either way, the test passes, so we should enable it, even if the query is not returning meaningful values (yet).

VerenaBeckham and others added 2 commits August 12, 2026 12:21
Enable test for all platforms.

Co-authored-by: Nikita Kornev <nikita.kornev@intel.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants