Skip to content

Enable deterministic support for oneDNN #127277

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
Closed
wants to merge 7 commits into from
Closed

Enable deterministic support for oneDNN #127277

wants to merge 7 commits into from

Conversation

weishi-deng
Copy link
Contributor

@weishi-deng weishi-deng commented May 28, 2024 •
edited
Loading

This PR is a part of RFC #114848.
For the request for Torchbenchmark models, this PR enables the deterministic attribute for the oneDNN operators for XPU backends, like convolution, deconvolution and matmult.

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @voznesenskym @penguinwu @EikanWang @Guobing-Chen @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @chenyang78 @kadeng @chauhang

@pytorch-bot PyTorch Bot
Copy link

pytorch-bot bot commented May 28, 2024 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/127277

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

✅ No Failures

As of commit d010a4d with merge base 9d06e37 (image):
💚 Looks good so far! There are no failures yet. 💚

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

@pytorch-bot pytorch-bot bot added module: cpu CPU specific problem (e.g., perf, algorithm) module: dynamo labels May 28, 2024
@weishi-deng weishi-deng marked this pull request as draft May 28, 2024 08:15
@PenghuiCheng PenghuiCheng mentioned this pull request Jun 7, 2024
Closed
EikanWang
EikanWang reviewed Jun 11, 2024
View reviewed changes
@EikanWang EikanWang requested review from gujinghui and jgong5 June 11, 2024 06:48
@EikanWang
Copy link
Collaborator

@weishi-deng , please fix lint issue.

@gujinghui
Copy link
Collaborator

Will we have PRs to enable the TF32, BF32 in oneDNN? @EikanWang

jgong5
jgong5 requested changes Jun 11, 2024
View reviewed changes
Copy link
Collaborator

@jgong5 jgong5 left a comment

Choose a reason for hiding this comment

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

Besides the naming (use mkldnn instead of onednn for consistency), do we need to update the documentation somewhere since it is about API update?

@EikanWang
Copy link
Collaborator

Will we have PRs to enable the TF32, BF32 in oneDNN? @EikanWang

Yes, we will.

@EikanWang
Copy link
Collaborator

Will we have PRs to enable the TF32, BF32 in oneDNN? @EikanWang

Yes, we will.

@gujinghui , @weishi-deng , https://github.com/pytorch/pytorch/pull/127294/files#diff-40b3d6cc2026cae8f139b15c4b0b05fd2e69c2715be0ce15648de79b0e15a4ebR365 FYI

@EikanWang EikanWang requested a review from jgong5 June 12, 2024 09:17
jgong5
jgong5 reviewed Jun 13, 2024
View reviewed changes
@weishi-deng weishi-deng requested a review from jgong5 June 14, 2024 04:36
@EikanWang EikanWang marked this pull request as ready for review June 15, 2024 03:56
@EikanWang EikanWang requested a review from desertfire June 15, 2024 03:59
@EikanWang
Copy link
Collaborator

@desertfire , we're trying to enable deterministic for XPU and enable it in dynamo benchmarks for accuracy tests.

@EikanWang EikanWang added the topic: not user facing topic category label Jun 20, 2024
@EikanWang
Copy link
Collaborator

@pytorchbot merge

@pytorch-bot pytorch-bot bot added the ciflow/trunk Trigger trunk jobs on your pull request label Jun 20, 2024
@pytorchmergebot
Copy link
Collaborator

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging
Check the merge workflow status
here

@pytorchmergebot
Copy link
Collaborator

Merge failed

Reason: 6 mandatory check(s) failed. The first few are:

Dig deeper by viewing the failures on hud

Details for Dev Infra team Raised by workflow job

Failing merge rule: Core Maintainers

@gujinghui
Copy link
Collaborator

@pytorchbot rebase

@pytorchmergebot
Copy link
Collaborator

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

@pytorchmergebot
Copy link
Collaborator

Successfully rebased onednn-deter onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout onednn-deter && git pull --rebase)

@gujinghui
Copy link
Collaborator

@pytorchbot merge

@pytorchmergebot
Copy link
Collaborator

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging
Check the merge workflow status
here

facebook-github-bot pushed a commit to pytorch/benchmark that referenced this pull request Jun 24, 2024
@weishi-deng @facebook-github-bot
Enable deterministic support for oneDNN (#127277)
a4ad760 
Summary:
This PR is a part of RFC pytorch/pytorch#114848.
For the request for Torchbenchmark models, this PR enables the deterministic attribute for the oneDNN operators for XPU backends, like convolution, deconvolution and matmult.

X-link: pytorch/pytorch#127277
Approved by: https://github.com/jgong5, https://github.com/EikanWang, https://github.com/desertfire, https://github.com/gujinghui

Reviewed By: fbgheith

Differential Revision: D58918395

fbshipit-source-id: 333cde99143d7e20ca838c82f5ffc0eb63f33a7c
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@desertfire desertfire desertfire approved these changes

@jgong5 jgong5 jgong5 approved these changes

@gujinghui gujinghui gujinghui approved these changes

@EikanWang EikanWang EikanWang approved these changes

Assignees

No one assigned

Labels

ciflow/trunk Trigger trunk jobs on your pull request Merged module: cpu CPU specific problem (e.g., perf, algorithm) module: dynamo open source topic: not user facing topic category

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

7 participants

@weishi-deng @EikanWang @gujinghui @pytorchmergebot @desertfire @jgong5 @pytorchbot