Skip to content

[SYCL-MLIR][KernelDisjointSpecialization] Allow functions called indirectly from GPU kernel #9194

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

Jump to bottom
Merged
merged 1 commit into from
Apr 25, 2023
Merged

[SYCL-MLIR][KernelDisjointSpecialization] Allow functions called indirectly from GPU kernel #9194

merged 1 commit into from
Apr 25, 2023

Conversation

whitneywhtsang
Copy link
Contributor

@whitneywhtsang whitneywhtsang commented Apr 25, 2023 •
edited by etiotto
Loading

The SYCL kernel body function could be called indirectly from GPU kernel, GPUKernel -> Function -> KernelBody.
This PR modifies the pass to allow versioning either in GPUKernel (when KernelBody is called directly from it) or Function.

@whitneywhtsang
[SYCL-MLIR] Allow specializing kernel body function called from GPU k…
570c566 
…erenl with depth 2

Signed-off-by: Tsang, Whitney <whitney.tsang@intel.com>
@whitneywhtsang whitneywhtsang added the sycl-mlir Pull requests or issues for sycl-mlir branch label Apr 25, 2023
@whitneywhtsang whitneywhtsang requested a review from etiotto as a code owner April 25, 2023 02:30
@whitneywhtsang whitneywhtsang self-assigned this Apr 25, 2023
@victor-eds
Copy link
Contributor

The SYCL kernel body function could be called indirectly from GPU kernel, GPUKernel -> Function -> KernelBody. Actually there is not much needed to be changed in the pass, as the middle function is not a candidate for KernelDisjointSpecialization.

Do we know when can this happen?

@whitneywhtsang
Copy link
Contributor Author

The SYCL kernel body function could be called indirectly from GPU kernel, GPUKernel -> Function -> KernelBody. Actually there is not much needed to be changed in the pass, as the middle function is not a candidate for KernelDisjointSpecialization.

Do we know when can this happen?

By default, when -D__SYCL_DISABLE_PARALLEL_FOR_RANGE_ROUNDING__ is not passed.

@whitneywhtsang whitneywhtsang merged commit 786dc32 into intel:sycl-mlir Apr 25, 2023
@whitneywhtsang whitneywhtsang deleted the wrapper-disjoint branch April 25, 2023 14:34
@whitneywhtsang whitneywhtsang mentioned this pull request Apr 26, 2023
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@victor-eds victor-eds victor-eds approved these changes

@etiotto etiotto etiotto approved these changes

@sommerlukas sommerlukas Awaiting requested review from sommerlukas

Assignees

@whitneywhtsang whitneywhtsang

Labels
sycl-mlir Pull requests or issues for sycl-mlir branch
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@whitneywhtsang @victor-eds @etiotto