Skip to content

Reduce bulk init time and fix OOM #3828

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.

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

Closed
wants to merge 1 commit into from

Conversation

peterfu0
Copy link
Contributor

@peterfu0 peterfu0 commented Mar 15, 2025

Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/911

Disable compaction when bulk initialize TBE in SSD, this reduce the initialization time from over 5mins to 2-3 mins. Also use bytes as the chunk size rather than row count, as each row might have different dimensions in different TBE, to avoid OOM issue.

Differential Revision: D70921864

@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D70921864

Copy link

netlify bot commented Mar 15, 2025

Deploy Preview for pytorch-fbgemm-docs ready!

Name Link
🔨 Latest commit a637c89
🔍 Latest deploy log https://app.netlify.com/sites/pytorch-fbgemm-docs/deploys/67d8f8460e32910008668e98
😎 Deploy Preview https://deploy-preview-3828--pytorch-fbgemm-docs.netlify.app
📱 Preview on mobile
Toggle QR Code...

QR Code

Use your smartphone camera to open QR code link.

To edit notification comments on pull requests, go to your Netlify site configuration.

@peterfu0 peterfu0 closed this Mar 15, 2025
@peterfu0 peterfu0 reopened this Mar 15, 2025
peterfu0 added a commit to peterfu0/FBGEMM-1 that referenced this pull request Mar 15, 2025
Summary:

X-link: facebookresearch/FBGEMM#911


X-link: facebookresearch/FBGEMM#909


X-link: facebookresearch/FBGEMM#908

Disable compaction when bulk initialize TBE in SSD, this reduce the initialization time from over 5mins to 2-3 mins. Also use bytes as the chunk size rather than row count, as each row might have different dimensions in different TBE, to avoid OOM issue.

Differential Revision: D70921864
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D70921864

@peterfu0 peterfu0 changed the title Reduce bulk init time and fix OOM (#3826) Reduce bulk init time and fix OOM Mar 15, 2025
peterfu0 added a commit to peterfu0/FBGEMM-1 that referenced this pull request Mar 17, 2025
Summary:

X-link: facebookresearch/FBGEMM#911


X-link: facebookresearch/FBGEMM#909


X-link: facebookresearch/FBGEMM#908

Disable compaction when bulk initialize TBE in SSD, this reduce the initialization time from over 5mins to 2-3 mins. Also use bytes as the chunk size rather than row count, as each row might have different dimensions in different TBE, to avoid OOM issue.

Differential Revision: D70921864
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D70921864

Summary:

X-link: facebookresearch/FBGEMM#911


X-link: facebookresearch/FBGEMM#909


X-link: facebookresearch/FBGEMM#908

Disable compaction when bulk initialize TBE in SSD, this reduce the initialization time from over 5mins to 2-3 mins. Also use bytes as the chunk size rather than row count, as each row might have different dimensions in different TBE, to avoid OOM issue.

Differential Revision: D70921864
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D70921864

@facebook-github-bot
Copy link
Contributor

This pull request has been merged in 63e378a.

liligwu pushed a commit to ROCm/FBGEMM that referenced this pull request Mar 19, 2025
Summary:
Pull Request resolved: pytorch#3828

X-link: https://github.com/facebookresearch/FBGEMM/pull/911

Disable compaction when bulk initialize TBE in SSD, this reduce the initialization time from over 5mins to 2-3 mins. Also use bytes as the chunk size rather than row count, as each row might have different dimensions in different TBE, to avoid OOM issue.

Reviewed By: duduyi2013

Differential Revision: D70921864

fbshipit-source-id: bb3c5103761a8f641d4ed8be93757902fcef91e0
q10 pushed a commit to q10/FBGEMM that referenced this pull request Apr 10, 2025
Summary:
X-link: pytorch#3828

Pull Request resolved: facebookresearch/FBGEMM#911

Disable compaction when bulk initialize TBE in SSD, this reduce the initialization time from over 5mins to 2-3 mins. Also use bytes as the chunk size rather than row count, as each row might have different dimensions in different TBE, to avoid OOM issue.

Reviewed By: duduyi2013

Differential Revision: D70921864

fbshipit-source-id: bb3c5103761a8f641d4ed8be93757902fcef91e0
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants