Skip to content

Conversation

chunhuanMeng
Copy link
Contributor

@chunhuanMeng chunhuanMeng commented Aug 19, 2025

Correct the global range. The global range should be multiples of work group size.
Fix #1465

@Copilot Copilot AI review requested due to automatic review settings August 19, 2025 09:21
Copy link
Contributor

@Copilot Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull Request Overview

Removes the artificial limit on logical threads in grid computation for SYCL upsample bilinear 2D kernels. The change eliminates the use of syclMaxWorkItemsPerTile() to cap the total threads, allowing the grid dimensions to be computed directly based on output dimensions and batch/channel counts.

  • Removes total_threads variable and associated limiting logic in grid computation
  • Simplifies grid dimension calculations to use actual requirements without thread limits
  • Updates both forward and backward kernel launch functions

Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
You can also share your feedback on Copilot code review for a chance to win a $100 gift card. Take the survey.

@CuiYifeng CuiYifeng self-requested a review August 21, 2025 07:25
@fengyuan14
Copy link
Contributor

Please correct the PR comment.

@chunhuanMeng
Copy link
Contributor Author

Please correct the PR comment.

done

@fengyuan14 fengyuan14 changed the title Removing the Limit on Logical Threads in Num_group Computation UpsampleBilinear2Daa: Correct the global range Aug 25, 2025
@chunhuanMeng chunhuanMeng added this pull request to the merge queue Aug 25, 2025
Merged via the queue into main with commit b979b55 Aug 25, 2025
61 of 63 checks passed
@chunhuanMeng chunhuanMeng deleted the meng_fix_upsample branch August 25, 2025 06:24
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.

RuntimeError: Non-uniform work-groups are not supported by the target device reported on BMG
3 participants