8000 `topk` on CUDA supports `bfloat16` by eqy · Pull Request #59977 · pytorch/pytorch · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

topk on CUDA supports bfloat16 #59977

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 5 commits into from
Closed

topk on CUDA supports bfloat16 #59977

wants to merge 5 commits into from

Conversation

eqy
Copy link
Collaborator
@eqy eqy commented Jun 14, 2021

@facebook-github-bot
Copy link
Contributor
facebook-github-bot commented Jun 14, 2021

💊 CI failures summary and remediations

As of commit 804c3a8 (more details on the Dr. CI page and at hud.pytorch.org/pr/59977):


  • 1/1 failures possibly* introduced in this PR
    • 1/1 non-scanned failure(s)

This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

@eqy
Copy link
Collaborator Author
eqy commented Jun 14, 2021

running CI to see what happens on ROCM, probably need to add skips for the new tests

@soulitzer soulitzer requested a review from ngimel June 14, 2021 22:38
@soulitzer soulitzer added the triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module label Jun 14, 2021
@ngimel
Copy link
Collaborator
ngimel commented Jun 15, 2021

ROCm is failing as expected, you need to adjust skips.

@facebook-github-bot
Copy link
Contributor

@ngimel has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

small = 10
large = 8192
for curr_size in (small, large):
self._test_topk_dtype(device, dtype, False, curr_size)

@dtypesIfCUDA(*torch.testing.get_all_fp_dtypes())
Copy link
Collaborator

Choose a reason for hiding this comment

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

you probably should include_bfloat16=True here, otherwise it's not tested

@facebook-github-bot
Copy link
Contributor

@ngimel has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

@facebook-github-bot
Copy link
Contributor

@ngimel merged this pull request in 40de03f.

pytorchmergebot pushed a commit that referenced this pull request Mar 28, 2022
Support for PR #[59977](#59977) in ROCm
Fixes #[56176](#56176) in ROCm since ROCm 5.0 supports dtype bfloat16 for sorting.

Dependency - PR #[72854](#72854) to be merged before

CC Please review @jithunnair-amd @jeffdaily
Pull Request resolved: #71913
Approved by: https://github.com/jithunnair-amd, https://github.com/osalpekar
facebook-github-bot pushed a commit that referenced this pull request Mar 30, 2022
Summary:
Support for PR #[59977](#59977) in ROCm
Fixes #[56176](#56176) in ROCm since ROCm 5.0 supports dtype bfloat16 for sorting.

Dependency - PR #[72854](#72854) to be merged before

CC Please review jithunnair-amd jeffdaily
Pull Request resolved: #71913
Approved by: https://github.com/jithunnair-amd, https://github.com/osalpekar

Test Plan: contbuild & OSS CI, see https://hud.pytorch.org/commit/pytorch/pytorch/6675f1e6970087cfc5ba67e7cf7690768e326464

Reviewed By: malfet

Differential Revision: D35205235

fbshipit-source-id: 906575b43aa57d8674a774d812ebfcd0037f4617
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
cla signed Merged open source triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module
Projects
None yet
Development

Successfully merging this pull request may close these issues.

TopK only implements partial support for BFloat16
5 participants
0