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 RDNA Config #640

Merged
merged 4 commits into from
Oct 1, 2024
Merged

add RDNA Config #640

merged 4 commits into from
Oct 1, 2024

Conversation

micmelesse
Copy link
Collaborator

@micmelesse micmelesse commented Sep 13, 2024

code to use custom navi configs. The People working on navi perf want their own configs and keys for running the kernels. They just need to provide that

This is a combination of 3 commits.

fix nit

rdna
@micmelesse micmelesse marked this pull request as ready for review September 30, 2024 23:20
python/perf-kernels/flash-attention.py Outdated Show resolved Hide resolved
This is a combination of 2 commits.

clean up

clean up more
@micmelesse micmelesse merged commit b36e072 into main_perf Oct 1, 2024
4 checks passed
hyoon1 added a commit to hyoon1/vllm that referenced this pull request Oct 25, 2024
- added Navi configurations (Related PR: ROCm/triton#640)
- resolved cache miss issue during flash attention calls by fixing max_seqlen_q/k to 0
micmelesse added a commit that referenced this pull request Oct 28, 2024
hyoon1 added a commit to hyoon1/vllm that referenced this pull request Nov 12, 2024
- added Navi configurations (Related PR: ROCm/triton#640)
- resolved cache miss issue during flash attention calls by fixing max_seqlen_q/k to 0
hyoon1 added a commit to hyoon1/vllm that referenced this pull request Nov 12, 2024
- added Navi configurations (Related PR: ROCm/triton#640)
- resolved cache miss issue during flash attention calls by fixing max_seqlen_q/k to 0
hyoon1 added a commit to hyoon1/vllm that referenced this pull request Nov 13, 2024
- added Navi configurations (Related PR: ROCm/triton#640)
- resolved cache miss issue during flash attention calls by fixing max_seqlen_q/k to 0
hyoon1 added a commit to hyoon1/vllm that referenced this pull request Nov 15, 2024
- added Navi configurations (Related PR: ROCm/triton#640)
- resolved cache miss issue during flash attention calls by fixing max_seqlen_q/k to 0
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