fix: recover PR 1863 ingest via sweep reuse

[Oseltamivir]

Recover the missing official ingest from source run 27951287552.

---

Note

Low Risk
Documentation-only changelog update with no runtime or benchmark code changes in the diff.

Overview
Recovers the official performance changelog link for minimaxm3-fp8-b300-dynamo-vllm after the missing ingest from source run 27951287552 (work originally associated with PR #1863).

The only file change is a new perf-changelog.yaml block keyed to PR #1893. It documents the MiniMax-M3 MXFP8 B300 disaggregated Dynamo vLLM STP coverage for 1k1k and 8k1k, including local srt-slurm recipes, TP4/Marlin low-concurrency decode shapes, single-node colocation with CUDA IPC for NIXL KV transfer, the 0618 MSA prefill top-k patch, and backports for Slurm node-IP discovery and heterogeneous-TP NIXL block validation.

^{Reviewed by Cursor Bugbot for commit b691c61. Bugbot is set up for automated code reviews on this repo. Configure here.}

[Oseltamivir]

/reuse-sweep-run 27951287552

[github-actions]

Thanks for the contribution! For vLLM & SGLang, please ensure that your recipes is similar to the official vLLM recipes and/or the SGLang cookbook

• https://docs.sglang.io/cookbook/intro
• https://recipes.vllm.ai/

If it is not, please create a PR first before we can merge your single node PR into the master branch. Let's ensure that the documentation is first class such that the entire ML community can benefit from your hard work! Thank you

• https://github.com/vllm-project/recipes
• https://github.com/sgl-project/sglang/tree/main/docs_new

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. If re-running failed jobs is attempted, PR authors are responsible for ensuring it passes. See GitHub's docs on re-running failed jobs: https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow

As a rule of thumb, generally, PR authors should request a review & get a PR approval from the respective companies' CODEOWNERS before requesting a review from core maintainers.

If additional help is needed, PR authors can reach out to core maintainers over Slack.

---

感谢你的贡献！对于 vLLM 与 SGLang，请确保你的 recipe 与官方 vLLM recipes 和/或 SGLang cookbook 保持一致

• https://docs.sglang.io/cookbook/intro
• https://recipes.vllm.ai/

如果不一致，请先创建一个 PR，之后我们才能将你的单节点 PR 合并到 master 分支。让我们确保文档保持一流水准，使整个 ML 社区都能从你的辛勤工作中受益！谢谢

• https://github.com/vllm-project/recipes
• https://github.com/sgl-project/sglang/tree/main/docs_new

PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动（flake），重新运行失败的任务即可解决。如果选择重新运行失败的任务，PR 作者有责任确保其最终通过。参见 GitHub 关于重新运行失败任务的文档：https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow

一般而言，PR 作者应先向相应公司的 CODEOWNERS 请求审阅并获得 PR 批准，然后再请求核心维护者审阅。

如需更多帮助，PR 作者可通过 Slack 联系核心维护者。

[Oseltamivir]

Recovery audit:

• Original failed run: https://github.com/SemiAnalysisAI/InferenceX/actions/runs/28001011122
• Failed job: https://github.com/SemiAnalysisAI/InferenceX/actions/runs/28001011122/job/82873180990
• Root cause: the reuse selector rejected the all-evals label before artifact ingestion.
• Reused source: run 27951287552, attempt 2, SHA aec850f26b4453f69f6fc29a409ab1b89722c182: https://github.com/SemiAnalysisAI/InferenceX/actions/runs/27951287552/attempts/2
• Recovery merge run: https://github.com/SemiAnalysisAI/InferenceX/actions/runs/28001851974
• Downstream ingest: https://github.com/SemiAnalysisAI/InferenceX-app/actions/runs/28001895287
• Recovered: 29 benchmark rows, 9 run-stat rows, 14 eval rows, 7,914 eval samples across 6 files, and 1 changelog entry. The ingest also updated 58 duplicate benchmark conflicts and upserted 87 availability rows.
• Verification: source artifacts validated; recovery GPU jobs were skipped; database ingest, overrides, database verification, cache invalidation, and the unmapped-entity check all succeeded; database benchmark errors: 0.