Skip to content

[PowerX] collect Qwen3.5 GB200 FP4 Dynamo-SGLang AgentX power / [PowerX] 采集 Qwen3.5 GB200 FP4 Dynamo-SGLang AgentX 功耗 - #4414

Open
edwingao28 wants to merge 4 commits into
mainfrom
chore/qwen35-fp8-gb200-sglang-v0522
Open

edwingao28 wants to merge 4 commits into
mainfrom
chore/qwen35-fp8-gb200-sglang-v0522

Conversation

@edwingao28

@edwingao28 edwingao28 commented Oct 23, 2026 •

Copy link
Copy Markdown
Collaborator

Summary

Qwen3.5 FP4 GB200 Dynamo-SGLang AgentX publishes no measured power. Its curve (11 points, 2026-10-22) comes from a run before multi-node power collection became default-on in #3670, so none of its points carries power fields. This PR reruns the unchanged qwen3.5-fp4-gb200-dynamo-sglang-agentic-mtp config on a head that includes #3670, so the curve republishes with GPU power and Grace CPU power.

This PR originally updated the qwen3.5-fp8-gb200-dynamo-sglang 8k1k image to SGLang v0.5.22. #4697 retired every Qwen3.5 8k1k config, so that change no longer applies, and this PR now carries the power rerun instead.

Changes:

  • inferencex-e2e/perf-changelog.yaml: replace the retired 8k1k block with one block that lists qwen3.5-fp4-gb200-dynamo-sglang-agentic-mtp, the only master key of this curve. The sweep reruns its 11 AgentX points and 5 AgentX evals.
  • No config or recipe changes. python -m infx.matrix.changed --base origin/main lists no keys.
  • The recipe runs NATS on the worker node, not a dedicated node, so every variant measures power.

Testing:

  • Planned: full-sweep-fail-fast sweep; the run link and power counts follow once it completes.
PR checklist and change type

Type of Change

  • Bug fix
  • New feature
  • Configuration change
  • Documentation update
  • Other (please describe): rerun of an unchanged config to collect power

Checklist

  • I have completed the AI model disclosure and kept it current
  • I have tested my changes locally
  • I have updated documentation if necessary
  • If python -m infx.matrix.changed --base origin/main lists affected config keys, I have appended one new entry covering them to the physical end of inferencex-e2e/perf-changelog.yaml; if it lists none, I have added none. I have not edited historical entries
  • If this PR appends an inferencex-e2e/perf-changelog.yaml entry, it carries exactly one primary sweep label (a maintainer applies it on fork PRs): full-sweep-fail-fast (recommended), full-sweep-enabled, or non-canary-full-sweep-enabled. Optional modifiers all-evals, evals-only, and agentx-fast require a primary label; the last two block reuse while applied.
  • Before merging via reuse, an authorized maintainer (OWNER/MEMBER/COLLABORATOR) has commented /use <run_id> (or the legacy /reuse-sweep-run) on this PR. Do this only once there is a final full sweep that is all green with evals passing, since after this comment the primary sweep label will no longer automatically kick off new sweeps. Remove and re-add the primary sweep label to force a new sweep.

中文:将 Qwen3.5 FP8 GB200 Dynamo SGLang 镜像更新至 v0.5.22。
@github-actions

Copy link
Copy Markdown
Contributor

Thanks for the contribution!

  • Review: If this PR changes files owned by someone other than a repository admin or @SemiAnalysisAI/core, ask one eligible CODEOWNER to complete the latest PR_REVIEW_CHECKLIST.md before contacting a core maintainer on Slack. Follow the template exactly, including As a PR reviewer and CODEOWNER, I have reviewed this and have, so sign-off verification triggers.
  • PR verification: Sweeps run only when the PR appends an inferencex-e2e/perf-changelog.yaml entry and carries exactly one primary label: full-sweep-fail-fast (strongly recommended; canary plus per-matrix fail-fast), full-sweep-enabled (canary; matrix jobs continue after a failure), or non-canary-full-sweep-enabled (no canary or fail-fast). The modifiers all-evals, evals-only, and agentx-fast require a primary label. On fork PRs, a maintainer applies the label. See sweep labels and reuse.
  • After merging: PR authors must ensure all GitHub Actions jobs pass. Transient failures often pass on rerun; see how to rerun failed jobs.
中文

感谢你的贡献!

  • **审阅:**如果 PR 修改的文件归属于仓库管理员及 @SemiAnalysisAI/core 之外的 CODEOWNER,请先联系一位有资格的 CODEOWNER 填写最新的 PR_REVIEW_CHECKLIST.md,再通过 Slack 联系核心维护者。必须严格遵循模板,并保留 As a PR reviewer and CODEOWNER, I have reviewed this and have,才能触发签核验证。
  • **PR 验证:**只有当 PR 在 inferencex-e2e/perf-changelog.yaml 末尾追加了条目,并且恰好带有一个主标签时,才会运行扫描:full-sweep-fail-fast(强烈推荐;canary 加逐矩阵 fail-fast)、full-sweep-enabled(有 canary;矩阵任务在失败后继续运行)或 non-canary-full-sweep-enabled(无 canary,也无 fail-fast)。修饰标签 all-evals、evals-only 和 agentx-fast 必须与主标签一起使用。fork PR 的标签由维护者添加。参见扫描标签与复用。
  • **合并后:**PR 作者必须确保所有 GitHub Actions 任务通过。临时性失败通常可以通过重新运行恢复;参见重新运行失败任务的说明。

中文:填写 GB200 changelog 的 PR 链接。
@edwingao28 edwingao28 added the full-sweep-fail-fast Full sweep with canary gate; first failure cancels the rest of that matrix (recommended) label Oct 23, 2026

@claude claude Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code review found no issues

No high-confidence issues detected in this change.

This review covers commit 598bd14, which is no longer the latest commit on this pull request; later commits are not covered by it.

@github-actions

github-actions Bot commented Oct 23, 2026 •

Copy link
Copy Markdown
Contributor

@edwingao28 edwingao28 changed the title [Qwen3.5 GB200] update FP8 Dynamo SGLang image to v0.5.22 / [Qwen3.5 GB200] 将 FP8 Dynamo SGLang 镜像更新至 v0.5.22 [PowerX] collect Qwen3.5 GB200 FP4 Dynamo-SGLang AgentX power / [PowerX] 采集 Qwen3.5 GB200 FP4 Dynamo-SGLang AgentX 功耗 Nov 6, 2026

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

full-sweep-fail-fast Full sweep with canary gate; first failure cancels the rest of that matrix (recommended)

Projects

Status: No status

Development

Successfully merging this pull request may close these issues.

1 participant