diff --git a/perf-changelog.yaml b/perf-changelog.yaml index a10f2a425c..a90f4cf806 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -1015,3 +1015,13 @@ - "Config concurrency: 32-256" - "update image to vllm/vllm-openai-rocm:v0.18.0" pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/927 + +- config-keys: + - gptoss-fp4-mi325x-vllm + - minimaxm2.5-fp8-mi325x-vllm + description: + - "Document --exclusive flag on MI325X salloc prevents node sharing during benchmarks" + - "Only non-TP8 configs are affected; TP8 already uses all GPUs on the node" + - "May yield performance improvements" + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/931 +