Skip to content

Conversation

@n13
Copy link
Contributor

@n13 n13 commented Jan 5, 2026

Problem

Canceling GPU would have to wait out the entire GPU batch range

This would also block all CPU threads

Solution

  • Canceling GPU asynchronously
  • GPU batch size adjusts to optional GPU time, default 3 seconds

Future improvements

  • Clean up logs - they all need targets and most of them should be debug
  • See if we can initialize GPU more efficiently - we lose a lot compared to the benchmarks

@n13 n13 merged commit 4131b60 into main Jan 5, 2026
4 checks passed
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.

3 participants