Add Qwen3.5 FP8 GB300 Dynamo-SGLang MTP recipes / 新增 Qwen3.5 FP8 GB300 Dynamo-SGLang MTP 配方#2316
Add Qwen3.5 FP8 GB300 Dynamo-SGLang MTP recipes / 新增 Qwen3.5 FP8 GB300 Dynamo-SGLang MTP 配方#2316RohitNagraj wants to merge 3 commits into
Conversation
Add the 8k/1k multi-node configuration and seven recipes with Mooncake KV transfer and the pinned SGLang image. 中文：新增 Qwen3.5 FP8 GB300 Dynamo-SGLang MTP 的 8k/1k 多节点配置及七个配方，使用 Mooncake KV 传输和固定版本的 SGLang 镜像。
|
Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase For PR verification, add the 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. See GitHub's docs on re-running failed jobs 感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 如需进行 PR 验证，请为此 PR 添加 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。参见 GitHub 关于重新运行失败任务的文档 |
3 similar comments
|
Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase For PR verification, add the 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. See GitHub's docs on re-running failed jobs 感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 如需进行 PR 验证，请为此 PR 添加 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。参见 GitHub 关于重新运行失败任务的文档 |
|
Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase For PR verification, add the 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. See GitHub's docs on re-running failed jobs 感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 如需进行 PR 验证，请为此 PR 添加 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。参见 GitHub 关于重新运行失败任务的文档 |
|
Thanks for the contribution! Please reach out to respective companies' CODEOWNER to fill in the latest PR_REVIEW_CHECKLIST.md before pinging core maintainer on Slack for review. In order for the signoff PR check bot to trigger, you must follow the PR_REVIEW_CHECKLIST.md template correctly, including the phrase For PR verification, add the 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. See GitHub's docs on re-running failed jobs 感谢你的贡献！请联系相应公司的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md，然后再在 Slack 上联系核心维护者进行审阅。为了触发 signoff PR 检查机器人，你必须正确遵循 PR_REVIEW_CHECKLIST.md 模板，包括保留英文语句 如需进行 PR 验证，请为此 PR 添加 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。参见 GitHub 关于重新运行失败任务的文档 |
Record the public pull request URL in the appended changelog entry. 中文：在新增的性能变更日志条目中记录公开拉取请求链接。
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
| - qwen3.5-fp8-gb300-dynamo-sglang-mtp | ||
| description: | ||
| - "Add Qwen3.5-397B-A17B-FP8 GB300 disaggregated Dynamo-SGLang MTP configurations for 8k/1k." | ||
| - "Cover 1P1D TP4/TP4, 1P1D TEP8/TEP8, and 3P/4P/6P/7P/8P DEP4/DEP16 topologies." |
There was a problem hiding this comment.
🟡 The new perf-changelog.yaml entry for qwen3.5-fp8-gb300-dynamo-sglang-mtp uses a placeholder pr-link (pull/XXX) instead of the actual PR number. Since this is PR #2316, it should read pull/2316.
Extended reasoning...
The last entry appended to perf-changelog.yaml for this PR (line 5068-5070) ends with:
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXXXXX is a literal template placeholder that was never substituted with the real PR number.
Where it comes from: Contributors adding a new recipe append a changelog entry to perf-changelog.yaml following the existing format, then fill in pr-link once the PR number is known (typically after opening the PR, since GitHub assigns the number at creation time). In this case the author appended the new config-keys: qwen3.5-fp8-gb300-dynamo-sglang-mtp entry but never went back to replace the XXX placeholder with the assigned PR number, 2316.
Why nothing catches it: perf-changelog.yaml is a free-form documentation/metadata file — nothing in CI parses or validates the pr-link field's contents (it isn't used to drive benchmark execution, recipe selection, or config parsing), so a bad or placeholder URL will not fail any check and will silently sit in the file post-merge.
Proof this is the only offending entry: Grepping the file for pull/XXX returns exactly one hit — this new entry at line 5070. Every other entry in the file, including the immediately preceding one (pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2301 at line 5062), has a real, resolvable PR number. Cross-referencing the PR metadata confirms this PR is #2316, so the correct value is:
pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2316Impact: Purely cosmetic/documentation — the changelog is a historical record for humans browsing perf-changelog.yaml, and clicking the current link 404s (there's no PR numbered "XXX"). It has no effect on benchmark execution, sweep behavior, or config parsing, so it doesn't block functionality. It should still be fixed before merge since it's trivial and the file is otherwise consistently maintained with real PR links for every entry.
Fix: Replace pull/XXX with pull/2316 on line 5068-5070.
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=30037509371 |
Enable chat-formatted benchmark inputs in all seven Qwen3.5 FP8 GB300 MTP recipes and record the change in the existing changelog entry. 中文：为 Qwen3.5 GB300 MTP 启用聊天模板 在全部七个 Qwen3.5 FP8 GB300 MTP 配方中启用聊天格式的基准测试输入，并在现有变更日志条目中记录该变更。
|
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=30115636016 |
|
/reuse-sweep-run |
1 participant
Summary
中文说明