perf: Add completion queue sharding#8815
Open
geobeau wants to merge 1 commit into
Open
Conversation
Add --grpc-infer-cq-count to control the number of inference completion queues. Default is 1 (single shared CQ, same behavior as before). Set to 0 for one CQ per handler thread, or N>1 for N sharded CQs, to reduce contention at high throughput.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does the PR do?
This PR adds a way to shard and configure more than one completion queue.
Per the commit message, see background for the motivation.
Checklist
Agreement
<commit_type>: <Title>pre-commit install, pre-commit run --all)Commit Type:
Check the conventional commit type
box here and add the label to the github PR.
Test plan:
Caveats:
Increasing the number of queues might make the load a bit unbalanced.
Background
When running above 100k QPS, GRPC threads will be the bottleneck to scale further we need to add more GRPC threads.
Unfortunately, adding more GRPC will have dimishing returns because there is a big contention on the futex required to access the completion queue. By sharding the completion queue into N part, we can reduce the contention drastically.