Skip to content

[ExecuTorch][WebGPU] Per-pass compute dispatch ordering for fused multi-dispatch ops#20162

Merged
JulianCloudNTH merged 2 commits into
mainfrom
gh/JulianCloudNTH/12/orig
Jun 9, 2026
Merged

[ExecuTorch][WebGPU] Per-pass compute dispatch ordering for fused multi-dispatch ops#20162
JulianCloudNTH merged 2 commits into
mainfrom
gh/JulianCloudNTH/12/orig

Conversation

@pytorchbot

Copy link
Copy Markdown
Collaborator

This PR was created by the merge bot to help merge the original PR into the main branch.
ghstack PR number: #20072 by @JulianCloudNTH
^ Please use this as the source of truth for the PR details, comments, and reviews
ghstack PR base: https://github.com/pytorch/executorch/tree/gh/JulianCloudNTH/12/base
ghstack PR head: https://github.com/pytorch/executorch/tree/gh/JulianCloudNTH/12/head
Merge bot PR base: https://github.com/pytorch/executorch/tree/main
Merge bot PR head: https://github.com/pytorch/executorch/tree/gh/JulianCloudNTH/12/orig

@diff-train-skip-merge

…ti-dispatch ops

Pull Request resolved: #20072

WebGPU has no write->read ordering between dispatches in a single compute pass, so a fused multi-dispatch op (SDPA) can read stale writes. Record one compute pass per dispatch in `execute()` (both the full and ranged paths) -- the pass boundary is WebGPU's implicit barrier (there is no `vkCmdPipelineBarrier`). Single-dispatch ops are unchanged. Also flips this file to the C++17 nested namespace. Consumed by the fused SDPA op above.
ghstack-source-id: 391378799
@exported-using-ghexport

Differential Revision: [D107543258](https://our.internmc.facebook.com/intern/diff/D107543258/)
@pytorch-bot

pytorch-bot Bot commented Jun 9, 2026

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20162

Note: Links to docs will display an error until the docs builds have been completed.

⏳ No Failures, 24 Pending

As of commit 90eb21d with merge base 189ffaa (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@meta-cla meta-cla Bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jun 9, 2026
@github-actions

github-actions Bot commented Jun 9, 2026

Copy link
Copy Markdown

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

@JulianCloudNTH JulianCloudNTH self-requested a review June 9, 2026 16:49
@JulianCloudNTH JulianCloudNTH merged commit 6fa26d4 into main Jun 9, 2026
182 checks passed
@JulianCloudNTH JulianCloudNTH deleted the gh/JulianCloudNTH/12/orig branch June 9, 2026 17:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants