Draft
Conversation
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR adds a reproducible benchmark flow for ATOM vLLM Plugin. The default environment is rocm/atom-dev:vllm-latest with the benchmarking recipe aligned with InferenceMax and ATOM itself.
It also provides an option to compare ATOM vLLM plugin against upstream vLLM under a fixed A/B setup. It standardizes benchmark execution and reports key serving metrics (throughput + latency) for direct runtime comparison.
Here is an example to compare performance between ATOM vLLM Plugin and Upstream vllm (with vllm/vllm-openai-rocm:nightly) for Kimi-K2-Thinking-MXFP4 (ISL/OSL=1K/1K, concurrency=8, TP=4).
Key Results (1K/1K, C8, TP4)
1k/1k c8 主结果(TP4)
ATOM plugin
Output TPUT: 578.85 tok/s
Total TPUT: 1152.17 tok/s
mean TTFT: 135.47 ms
mean TPOT: 12.97 ms
mean E2EL: 13898.61 ms
Upstream nightly
Output TPUT: 459.33 tok/s
Total TPUT: 914.28 tok/s
mean TTFT: 225.03 ms
mean TPOT: 16.38 ms
mean E2EL: 17570.20 ms
对比结论(ATOM 相对 upstream)
吞吐:
Output TPUT +26.02%
Total TPUT +26.02%
时延(越低越好):
TTFT -39.80%(更低)
TPOT -20.82%
E2EL -20.90%