Skip to content

Triton rope #5

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
Merged
merged 6 commits into from
Apr 18, 2025
Merged

Triton rope #5

merged 6 commits into from
Apr 18, 2025

Conversation

Godofnothing
Copy link
Collaborator

This PR adds triton implementation for RoPE operations.

Triton implementation turns out to be slightly faster than original compiled version:

end-2-end generation latency

  • baseline (torch) - 7.59it/s
  • triton - 8.65it/s

Triton implementation can be turned on via APPLY_ROTARY_COS_SIN_TRITON env variable set to 1.

@justheuristic justheuristic self-requested a review April 18, 2025 18:04
@Godofnothing Godofnothing merged commit 389aca0 into main Apr 18, 2025
@Godofnothing Godofnothing deleted the triton_rope branch April 18, 2025 18:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@justheuristic justheuristic justheuristic approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@Godofnothing @justheuristic