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.

Sign up for GitHub

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

Jump to bottom

Issue with block_size = 64 #5

Open
PiotrNawrot opened this issue Mar 10, 2025 · 1 comment
Open

Issue with block_size = 64 #5

PiotrNawrot opened this issue Mar 10, 2025 · 1 comment

Comments

@PiotrNawrot
Copy link

Thanks for the amazing research and releasing the code!
I get the following error when I change the block size to 64 on H100, any ideas?

SharedEncodingAttr builder when the MMAEncodingAttr is Hopper has not been implemented yet
UNREACHABLE executed at /project/python/build/cmake.linux-x86_64-cpython-3.10/include/triton/Dialect/TritonGPU/IR/TritonGPUAttrDefs.cpp.inc:347!
Aborted
@XunhaoLai
Copy link
Collaborator

@PiotrNawrot Thank you for your feedback! FlexPrefill was developed and tested on A100 GPUs and has not yet been tested on Hopper GPUs. I’ll work on adding support for this in a future update. Apologies for the inconvenience!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@PiotrNawrot @XunhaoLai