perf: retune serving yield batch to 64 - #6
Merged
Merged
Conversation
Why: a same-runner sweep showed batch 64 improves tell throughput by about 16% over 32, while batch 128 violates the ask-cancellation timing contract. What: use 64 for normal serving and keep stranded-message cleanup at 32. Result: higher steady-backlog throughput without changing public APIs or weakening lifecycle tests.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
The current 32-message serving batch leaves measurable scheduler overhead on sustained backlogs.
A same-runner sweep of 32/64/128/256/512 showed that 64 improves tell throughput materially while keeping the existing cancellation/lifecycle tests intact. Batch 128 was rejected because it caused T10 (asker cancellation behind backlog) to complete the ask before cancellation was observed.
Measured result
On AMD EPYC 7763 with the same MoonBit toolchain:
What changed
The cleanup cadence is split deliberately so throughput tuning does not make abnormal cleanup less responsive.