Skip to content

Conversation

aarongreig
Copy link
Contributor

@aarongreig aarongreig commented Jun 2, 2025

Introduces urEnqueueKernelLaunchWithArgsExp, a new UR entry point which allows combining KernelSetArg and EnqueueKernelLaunch calls into one. Replaces urEnqueueKernelLaunch as the default kernel launch path in sycl RT.

@aarongreig aarongreig force-pushed the aaron/enqueueKernelWithArgs branch from 76612c5 to 1554457 Compare June 4, 2025 15:52
@aarongreig aarongreig force-pushed the aaron/enqueueKernelWithArgs branch from 7a2f91c to 6123fa6 Compare June 6, 2025 11:16
@aarongreig aarongreig force-pushed the aaron/enqueueKernelWithArgs branch from 6123fa6 to cebad02 Compare June 6, 2025 16:22
@aarongreig
Copy link
Contributor Author

CI fails are unrelated, same set seen in another PR here https://github.com/intel/llvm/actions/runs/16532939566/job/46762455916

@aarongreig
Copy link
Contributor Author

@intel/llvm-gatekeepers fails are unrelated and there's a change open to disable to tests in question #19606 I think this is ready to merge

@ldrumm
Copy link
Contributor

ldrumm commented Jul 28, 2025

Happy to merge, but please add a description first

@aarongreig
Copy link
Contributor Author

Happy to merge, but please add a description first

done

@ldrumm ldrumm merged commit 24f54ea into intel:sycl Jul 28, 2025
30 of 34 checks passed
sarnex pushed a commit that referenced this pull request Aug 4, 2025
…e call." (#19661)

Reverts #18764

Looks like the naive level zero implementation now results in a
performance regression, I think it's best to revert until l0 can better
match the new interface.
github-actions bot pushed a commit to oneapi-src/unified-runtime that referenced this pull request Aug 5, 2025
…e call." (#19661)

Reverts intel/llvm#18764

Looks like the naive level zero implementation now results in a
performance regression, I think it's best to revert until l0 can better
match the new interface.
aarongreig added a commit to oneapi-src/unified-runtime that referenced this pull request Aug 5, 2025
…e call." (#19661)

Reverts intel/llvm#18764

Looks like the naive level zero implementation now results in a
performance regression, I think it's best to revert until l0 can better
match the new interface.
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.