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

Integrate Signed Bucket Indices method for Efficient Bucket Accumulation for Later Phase #8

Open
moven0831 opened this issue Oct 25, 2024 · 0 comments
Assignees

Comments

@moven0831
Copy link
Collaborator

Problem

The scalar decomposition and bucket aggregation stages are not optimized, leading to inefficient parallel processing and increased computation time in the MSM process.

Details

Adopt the signed bucket indices technique to optimize scalar decomposition and bucket aggregation. This method reduces the number of buckets by half, enhancing parallel processing efficiency and balancing the workload across Metal's GPU threads.

Acceptance criteria

  • Modify the scalar decomposition shader to convert scalars into signed index form within Metal.
  • Adjust the bucket aggregation logic to handle signed indices, including point negation based on index signs.
  • Ensure compatibility with Metal's buffer handling and memory access patterns.
  • Validate the correctness of the implementation through comprehensive testing with various scalar distributions.

Reference

@moven0831 moven0831 self-assigned this Nov 1, 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

1 participant