This repository was archived by the owner on Apr 20, 2026. It is now read-only.
tune b200 fp8 max throughput config for 8k1k stp 1p1d#210
Open
weireweire wants to merge 4 commits into
Open
Conversation
Adjust memory fraction, prefill tokens, chunked prefill size, and max running requests for both prefill and decode workers to improve max throughput.
Contributor
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
added 3 commits
March 6, 2026 02:05
The sbatch script copies the config YAML to the output directory but
then passes the original file path to do_sweep. If the original file
is modified between job submission and execution (e.g., by submitting
another job with different params), the wrong config gets used.
Now do_sweep reads from ${OUTPUT_DIR}/config.yaml instead.
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 subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
mem-fraction-static,max-prefill-tokens,chunked-prefill-size, andmax-running-requestsfor both prefill and decode workers inmax-tpt-dep8-1p1d.yamlBenchmark Results (job 6983 → 7028)
Concurrency=128, rate=inf, 1280 requests, isl=8192, osl=1024.
Key changes
mem-fraction-static0.85→0.6,max-prefill-tokens32768→16384,chunked-prefill-size32768→131072,max-running-requests1024→16mem-fraction-static0.85→0.6,max-prefill-tokens32768→1048576,chunked-prefill-size32768→8388608The main bottleneck was
max-running-requests: 1024on prefill — too many concurrent prefills caused GPU memory contention and doubled TTFT. Reducing to 16 with largerchunked-prefill-sizeallows prefill to complete efficiently.Test plan