[Klaud Cold] Update DSR1 FP8 MI300X SGLang image to v0.5.22-rocm720-mi30x / [Klaud Cold] 将 DSR1 FP8 MI300X SGLang 镜像更新至 v0.5.22-rocm720-mi30x by Klaud-Cold · Pull Request #4363 · SemiAnalysisAI/InferenceX · GitHub
Skip to content

[Klaud Cold] Update DSR1 FP8 MI300X SGLang image to v0.5.22-rocm720-mi30x / [Klaud Cold] 将 DSR1 FP8 MI300X SGLang 镜像更新至 v0.5.22-rocm720-mi30x - #4363

Open
Klaud-Cold wants to merge 2 commits into
mainfrom
klaud/auto-4bfcce0abcfcb357-908c81de52c49bbd
Open

[Klaud Cold] Update DSR1 FP8 MI300X SGLang image to v0.5.22-rocm720-mi30x / [Klaud Cold] 将 DSR1 FP8 MI300X SGLang 镜像更新至 v0.5.22-rocm720-mi30x#4363
Klaud-Cold wants to merge 2 commits into
mainfrom
klaud/auto-4bfcce0abcfcb357-908c81de52c49bbd

Conversation

@Klaud-Cold

@Klaud-Cold Klaud-Cold commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator

Goal: Update SGLang image from lmsysorg/sglang:v0.5.21-rocm720-mi30x to lmsysorg/sglang:v0.5.22-rocm720-mi30x.
Baseline: 2026-10-20 · lmsysorg/sglang:v0.5.21-rocm720-mi30x@sha256:0534eb382626ef55ace8be24f7d81cc6eb7451acac8a5727848397a5e9309d91
8k/1k · TP8/EP1 · Mean latency · Sources: API 1, API 2, API 3

Concurrency Total tok/s/GPU ↑ Output tok/s/GPU ↑ TTFT ms ↓ TPOT ms ↓
4 195.06 21.7 1,561.01 20.9
8 326.91 36.75 1,428.4 25.22
16 502.8 55.71 1,849.88 33.01
32 663.63 74.21 2,114.03 50.57
64 802.33 89.01 2,948.56 85.51
Eval Score ↑ Samples
gsm8k/em_strict · c32 95.45% 1,319
gsm8k/em_strict · c64 95.53% 1,319

Stop: Add the klaud-abort label at any time to stop this candidate now; Klaud cancels its runs, closes this PR and does not retry it. Remove the label to let the planner pick it up again.

中文

**目标：**将 SGLang 镜像从 lmsysorg/sglang:v0.5.21-rocm720-mi30x 更新至 lmsysorg/sglang:v0.5.22-rocm720-mi30x。
**基线：**2026-10-20 · lmsysorg/sglang:v0.5.21-rocm720-mi30x@sha256:0534eb382626ef55ace8be24f7d81cc6eb7451acac8a5727848397a5e9309d91
8k/1k · TP8/EP1 · 平均延迟 · 来源： API 1, API 2, API 3；数值及异常说明见上表。
**停止：**随时添加 klaud-abort 标签即可立即停止该候选：Klaud 会取消其运行、关闭此 PR 且不再重试。移除该标签后，规划器会在下次扫描时重新考虑它。

@github-actions @claude
chore(dsr1): update MI300X FP8 SGLang image to v0.5.22-rocm720-mi30x
b297ca4 
Pin lmsysorg/sglang:v0.5.22-rocm720-mi30x by digest for
dsr1-fp8-mi300x-sglang. All recipe flags and environment variables
remain supported by SGLang v0.5.22.

将 dsr1-fp8-mi300x-sglang 的 SGLang 镜像更新至 v0.5.22-rocm720-mi30x
并按摘要固定；配方中的所有参数和环境变量在 SGLang v0.5.22 中仍受支持。

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Oct 21, 2026

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。

@Klaud-Cold

Klaud-Cold commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator Author

Initial attempt · Passed · Run 39557708209 / attempt 1 · 2026-10-21 07:29 UTC
lmsysorg/sglang:v0.5.22-rocm720-mi30x@sha256:58b6bc333f527c88fe5ff1b4bd8d17e82214eabcb3044549117159878967c930 · 6046414d2d1a · 8k/1k · TP8/EP1 · Mean latency
Change: Measured ref b297ca4: pin lmsysorg/sglang:v0.5.22-rocm720-mi30x by digest (bundles SGLang v0.5.22 baa9befa, AITER e7d2453f, gfx942/ROCm 7.2); source diff of sgl-project/sglang v0.5.21..v0.5.22 shows every recipe flag and SGLANG_USE_AITER/SGLANG_AITER_MLA_PERSIST still supported, so the recipe is unchanged.

Concurrency Output tok/s/GPU ↑ TTFT ms ↓ TPOT ms ↓
4 25.54 (+17.7%) 720.5 (-53.8%) 18.37 (-12.1%)

Note: All rows: request errors unavailable.

Eval Score ↑ Samples
gsm8k/em_strict · c32 95.83% (+0.38 pp) 1,319 each

Next: Smoke benchmark and gsm8k eval passed; append the perf-changelog entry and run the full fail-fast sweep.

中文

初次尝试 · 已通过 · Run 39557708209 / attempt 1 · 2026-10-21 07:29 UTC
lmsysorg/sglang:v0.5.22-rocm720-mi30x@sha256:58b6bc333f527c88fe5ff1b4bd8d17e82214eabcb3044549117159878967c930 · 6046414d2d1a · 8k/1k · TP8/EP1 · 平均延迟
**变更：**测量提交 b297ca4：按摘要固定 lmsysorg/sglang:v0.5.22-rocm720-mi30x（内含 SGLang v0.5.22 baa9befa、AITER e7d2453f、gfx942/ROCm 7.2）；对比 sgl-project/sglang v0.5.21..v0.5.22 源码，配方中所有参数及 SGLANG_USE_AITER/SGLANG_AITER_MLA_PERSIST 仍受支持，因此配方保持不变。实测数值及异常说明见上表。
**下一步：**冒烟基准测试和 gsm8k 评测均已通过；追加 perf-changelog 条目并运行完整的 fail-fast 扫描。

@github-actions @claude
chore(dsr1): add perf-changelog entry for MI300X SGLang v0.5.22
c8fa96f 
为 MI300X SGLang v0.5.22 添加 perf-changelog 条目。

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@Klaud-Cold Klaud-Cold added the full-sweep-fail-fast Full sweep with canary gate; first failure cancels the rest of that matrix (recommended) label Oct 21, 2026
@Klaud-Cold

Klaud-Cold commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator Author

Final full sweep · Passed · Run 39562635990 / attempt 1 · 2026-10-21 08:20 UTC
lmsysorg/sglang:v0.5.22-rocm720-mi30x@sha256:58b6bc333f527c88fe5ff1b4bd8d17e82214eabcb3044549117159878967c930 · c8fa96f46514 · 8k/1k · TP8/EP1 · Mean latency
Change: Validate the complete updated-image family at this head.

Concurrency Output tok/s/GPU ↑ TTFT ms ↓ TPOT ms ↓
4 25.62 (+18.1%) 710.27 (-54.5%) 18.32 (-12.3%)
8 43.21 (+17.6%) 843.55 (-40.9%) 21.86 (-13.3%)
16 62.87 (+12.9%) 1,096.56 (-40.7%) 29.77 (-9.8%)
32 80.97 (+9.1%) 1,641.47 (-22.4%) 46.64 (-7.8%)
64 96.88 (+8.8%) 2,500.75 (-15.2%) 78.78 (-7.9%)

Note: All rows: request errors unavailable.

Eval Score ↑ Samples
gsm8k/em_strict · c32 95.15% (-0.30 pp) 1,319 each
gsm8k/em_strict · c64 95.68% (+0.15 pp) 1,319 each

Next: Mark ready for maintainer review.

中文

最终完整 sweep · 已通过 · Run 39562635990 / attempt 1 · 2026-10-21 08:20 UTC
lmsysorg/sglang:v0.5.22-rocm720-mi30x@sha256:58b6bc333f527c88fe5ff1b4bd8d17e82214eabcb3044549117159878967c930 · c8fa96f46514 · 8k/1k · TP8/EP1 · 平均延迟
**变更：**验证此提交更新镜像后的完整配置族。实测数值及异常说明见上表。
**下一步：**标记为就绪，等待维护者审查。

@github-actions

github-actions Bot commented Oct 21, 2026

Copy link
Copy Markdown
Contributor

View unofficial run (performance): https://inferencex.semianalysis.com/inference?unofficialRun=39562635990

View unofficial run (accuracy): https://inferencex.semianalysis.com/evaluation?unofficialRun=39562635990

@Klaud-Cold
Klaud-Cold marked this pull request as ready for review October 21, 2026 08:20
@Klaud-Cold
Klaud-Cold requested a review from a team October 21, 2026 08:20
@Klaud-Cold

Klaud-Cold commented Oct 21, 2026

Copy link
Copy Markdown
Collaborator Author

validated · Repairs: 0 · Runs: 39557708209, 39562584770, 39562635990
All owned runs ended. Full sweep verified; ready for review.

中文

validated · 修复次数：0 · 运行：39557708209, 39562584770, 39562635990
所有自有运行均已结束。完整 sweep 已验证；已就绪，等待审查。

@Klaud-Cold

Klaud-Cold commented Oct 21, 2026

Copy link
Copy Markdown
Collaborator Author

/use 39562635990

@claude

claude Bot commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

Claude finished @Klaud-Cold's task in 10s —— View job

LGTM - no blocking issues found

@github-actions

github-actions Bot commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

claude[bot]
claude Bot reviewed Oct 21, 2026
View reviewed changes

@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 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

Reviewers

@claude claude[bot] claude[bot] left review comments
@billishyahao billishyahao Awaiting requested review from billishyahao billishyahao is a code owner
@chunfangamd chunfangamd Awaiting requested review from chunfangamd chunfangamd is a code owner
@seungrokj seungrokj Awaiting requested review from seungrokj seungrokj is a code owner
@1am9trash 1am9trash Awaiting requested review from 1am9trash 1am9trash is a code owner
@zijiecode zijiecode Awaiting requested review from zijiecode zijiecode is a code owner

At least 1 approving review is required to merge this pull request.

Assignees

No one assigned

Labels

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

Projects

InferenceX Board
Status: No status

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@Klaud-Cold