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

Support for Pascal architecture GPUs #32

Closed
DwayneX77 opened this issue Aug 7, 2022 · 4 comments
Closed

Support for Pascal architecture GPUs #32

DwayneX77 opened this issue Aug 7, 2022 · 4 comments

Comments

@DwayneX77
Copy link

Wonderful work presented! May I ask if there is any plan to expand support list to include Pascal architecture GPUs? Take Tesla P-Series as examples.

@tridao
Copy link
Contributor

tridao commented Aug 7, 2022

We currently support Turing (e.g. RTX 2080) and Ampere (e.g. RTX 3080) GPUs. We rely on tensor cores for matrix multiplication, which older GPUs lack.
You might be interested in the memory-efficient attention implemented by the xformers team (targeting fp32 instead of fp16), which I believe supports Pascal cards:
facebookresearch/xformers#267
facebookresearch/xformers#281

@DwayneX77
Copy link
Author

Thanks! Your advice helps a lot.

We currently support Turing (e.g. RTX 2080) and Ampere (e.g. RTX 3080) GPUs. We rely on tensor cores for matrix multiplication, which older GPUs lack. You might be interested in the memory-efficient attention implemented by the xformers team (targeting fp32 instead of fp16), which I believe supports Pascal cards: facebookresearch/xformers#267 facebookresearch/xformers#281

@fmassa
Copy link

fmassa commented Aug 10, 2022

Hi,

We have just pushed a PR in facebookresearch/xformers#362 which contains V100 and P100 support as well, and dispatches to FlashAttention for the cases where it is supported.

@ifffrt
Copy link

ifffrt commented Sep 26, 2022

Hi,

We have just pushed a PR in facebookresearch/xformers#362 which contains V100 and P100 support as well, and dispatches to FlashAttention for the cases where it is supported.

Would this PR make support for Maxwell GPUs possible too?

@tridao tridao closed this as completed Nov 10, 2022
wwbitejotunn pushed a commit to wwbitejotunn/flash-attention that referenced this issue Mar 28, 2024
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

No branches or pull requests

4 participants