Skip to content

STBE GPU coalescing kernel#2275

Closed
YazhiGao wants to merge 1 commit into
pytorch:mainfrom
YazhiGao:export-D52903658
Closed

STBE GPU coalescing kernel#2275
YazhiGao wants to merge 1 commit into
pytorch:mainfrom
YazhiGao:export-D52903658

Conversation

@YazhiGao

Copy link
Copy Markdown

Summary:

  • this is a follow up on GPU coalescing of stbe output, we follow the old idea of sparse feature rebatching to enable this. sequence embedding is afterall jagged tensor in dim D and stbe output is well aligned with input as [sum_l(T, B), D].
  • we explictly fork code for avoiding spaming all function into one reorder kernel suite.

Differential Revision: D52903658

@netlify

netlify Bot commented Jan 20, 2024

Copy link
Copy Markdown

Deploy Preview for pytorch-fbgemm-docs ready!

Name Link
🔨 Latest commit af9e290
🔍 Latest deploy log https://app.netlify.com/sites/pytorch-fbgemm-docs/deploys/65b20df08ce3d60008ab10a5
😎 Deploy Preview https://deploy-preview-2275--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.

@facebook-github-bot

Copy link
Copy Markdown
Contributor

This pull request was exported from Phabricator. Differential Revision: D52903658

YazhiGao pushed a commit to YazhiGao/FBGEMM that referenced this pull request Jan 20, 2024
Summary:

* this is a follow up on GPU coalescing of stbe output, we follow the old idea of sparse feature rebatching to enable this. sequence embedding is afterall jagged tensor in dim D and stbe output is well aligned with input as [sum_l(T, B), D].
* we explictly fork code for avoiding spaming all function into one reorder kernel suite.

Differential Revision: D52903658
@facebook-github-bot

Copy link
Copy Markdown
Contributor

This pull request was exported from Phabricator. Differential Revision: D52903658

Summary:

* this is a follow up on GPU coalescing of stbe output, we follow the old idea of sparse feature rebatching to enable this. sequence embedding is afterall jagged tensor in dim D and stbe output is well aligned with input as [sum_l(T, B), D].
* we explictly fork code for avoiding spaming all function into one reorder kernel suite.

Reviewed By: jspark1105, 842974287

Differential Revision: D52903658
@facebook-github-bot

Copy link
Copy Markdown
Contributor

This pull request was exported from Phabricator. Differential Revision: D52903658

@facebook-github-bot

Copy link
Copy Markdown
Contributor

This pull request has been merged in 17a0604.

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.

2 participants