Skip to content

[UR][OpenCL] Save the function pointer for clSetKernelArgMemPointerINTEL per kernel #18103

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

Merged
merged 3 commits into from
Apr 25, 2025

Conversation

bashbaug
Copy link
Contributor

Saves the function pointer for clSetKernelArgMemPointerINTEL when the kernel is created. This eliminates the need to query the context for the kernel and look up the function pointer in the extension function pointer cache each time a pointer is set as a kernel argument.

Additionally:

  • Adds a few missing error conditions.
  • Adds a few missing kernel info queries, for devices that support them.

@bashbaug bashbaug requested a review from a team as a code owner April 19, 2025 00:48
@bashbaug
Copy link
Contributor Author

I looked at the failing checks and I don't think they're caused by my changes. Is any action required on my end? Should I investigate further, or move to a newer tag?

@kbenzie
Copy link
Contributor

kbenzie commented Apr 24, 2025

I looked at the failing checks and I don't think they're caused by my changes. Is any action required on my end? Should I investigate further, or move to a newer tag?

I gave them a retry, if its persistent then pulling in latest sycl branch would be a reasonable next step.

This eliminates the need to query the context for the kernel and
look up the function pointer in the extension function pointer cache
each time a pointer is set as a kernel argument.

Signed-off-by: Ben Ashbaugh <[email protected]>
Because CL_INVALID_VALUE is returned both when the size of the value
being queried is incorrect and when the information being queried is
unrecognized, it is difficult to differentiate between the UR
INVALID_SIZE and UNSUPPORTED_ENUMERATION errors.  For now, we will
check for and return INVALID_SIZE first, which is what the tests are
expecting.
@bashbaug
Copy link
Contributor Author

Great, I updated to a newer tag and it looks like all tests are passing now 👍

@kbenzie kbenzie merged commit 7d063d2 into intel:sycl Apr 25, 2025
56 of 63 checks passed
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