Skip to content
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

Add fp8 kv cache for ROCm #2856

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Add fp8 kv cache for ROCm #2856

wants to merge 3 commits into from

Conversation

mht-sharma
Copy link
Collaborator

What does this PR do?

As per title!

Fixes # (issue)

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you read the contributor guideline,
    Pull Request section?
  • Was this discussed/approved via a Github issue or the forum? Please add a link
    to it if that's the case.
  • Did you make sure to update the documentation with your changes? Here are the
    documentation guidelines, and
    here are tips on formatting docstrings.
  • Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

@mht-sharma mht-sharma changed the title Add fp8 kv cache for rocm Add fp8 kv cache for ROCm Dec 18, 2024
@mht-sharma mht-sharma marked this pull request as ready for review January 3, 2025 12:07
@mht-sharma mht-sharma requested a review from danieldk January 3, 2025 12:07
Comment on lines +68 to +71
self.kv_cache_dtype = "auto"
if SYSTEM == "rocm" and dtype == torch.float8_e4m3fn:
self.kv_cache_dtype = "fp8"
dtype = torch.uint8
Copy link
Member

Choose a reason for hiding this comment

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

I think it would be nicer to allocate the cache in the requested type (torch.float8_e4m3fn) and then cast to uint8 where/when needed. Then we don't need another bookkeeping field (kv_cache_dtype) and the cache carries the type information.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants