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

Metal MSM GPU Acceleration Support #19

Conversation

ElusAegis
Copy link

This PR introduces support for Metal-based MSM GPU acceleration, leveraging the mopro GPU acceleration fork.

Key Highlights:

  • Performance Gains:

    • Expected performance improvements of approximately 25% for log 17 instance sizes.
    • Gains increase to 35% or more for log 20+ instance sizes.
    • Detailed benchmarking results are available in the mopro GPU acceleration fork.
  • Selective Activation:

    • The GPU acceleration is feature gated and further will only be enabled only for MSM operations with an instance size of log 17 or greater.
    • A user warning is displayed to inform about the activation threshold and expected benefits.

Notes:

This enhancement provides a significant boost for large-scale MSM computations, by collectively utilizing Metal's GPU resources with CPU resources on supported devices.

@ElusAegis ElusAegis force-pushed the ac/msm-gpu-acceleration branch from c1c2c10 to e88c505 Compare January 8, 2025 13:40
@ElusAegis ElusAegis closed this Jan 17, 2025
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.

1 participant