Update qwen3.5-fp8-mi300x-sglang SGLang image to v0.5.11-rocm720-mi30x

[Klaud-Cold]

Summary

• Update qwen3.5-fp8-mi300x-sglang image from lmsysorg/sglang:v0.5.10-rocm720-mi30x to lmsysorg/sglang:v0.5.11-rocm720-mi30x

Ref #1154

Generated with Claude Code

[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.

[claude]

LGTM — straightforward image tag bump with matching changelog entry; only nit (pr-link placeholder XXX) is flagged inline.

Extended reasoning...

Overview

This PR makes two trivial changes: it bumps the qwen3.5-fp8-mi300x-sglang image tag in .github/configs/amd-master.yaml from lmsysorg/sglang:v0.5.10-rocm720-mi30x to v0.5.11-rocm720-mi30x, and appends a corresponding entry to perf-changelog.yaml. No code paths, scripts, workflows, or test logic are touched.

Security risks

None. Both files are static YAML benchmark configuration: a model/runner image pin and a human-readable changelog. The new image tag follows the same registry/repo (lmsysorg/sglang) and naming scheme as adjacent entries in the file, so this is not a registry/provenance change.

Level of scrutiny

Minimal. This is a mechanical config bump of a single image tag, following the exact pattern used in surrounding entries (e.g., other v0.5.10-rocm720-mi30x pins and prior changelog rows referencing similar bumps). The change is self-contained and trivially reversible.

Other factors

The bug hunting system found a single nit: pr-link: ...pull/XXX in the new changelog entry was never substituted with 1353. That's a one-character provenance-link fix and is already flagged as an inline comment, so I don't need to restate it here. It doesn't affect runtime/build behavior, so it's not a blocker for approval.

[claude]

🟡 The newly added perf-changelog entry uses an unfilled placeholder pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXX instead of the actual PR number (1353). Replace XXX with 1353 so the changelog navigation link resolves rather than 404s.

Extended reasoning...

What the bug is

At perf-changelog.yaml:2351, the new entry added by this PR contains a literal placeholder:

- config-keys:
    - qwen3.5-fp8-mi300x-sglang
  description:
    - "Update SGLang image from v0.5.10-rocm720-mi30x to v0.5.11-rocm720-mi30x"
  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXX

The XXX is a placeholder that was never substituted with the real PR number. Every surrounding entry in this file (e.g. lines around 2318/2325/2332/2338/2345 referencing /pull/1304, /pull/1305, /pull/1308, /pull/1310) uses a real numeric PR ID, so this entry is clearly malformed relative to the rest of the file.

Why this matters

perf-changelog.yaml is the project's authoritative provenance log mapping config-key changes back to the PR that introduced them. Once this PR merges as #1353, https://github.com/SemiAnalysisAI/InferenceX/pull/XXX will resolve to a non-existent PR (GitHub returns 404 for non-numeric pull paths), so anyone using the changelog to trace why qwen3.5-fp8-mi300x-sglang jumped from v0.5.10 to v0.5.11 will hit a dead end on this row specifically.

Step-by-step proof

1. Open perf-changelog.yaml at the bottom — the diff appends exactly one entry.
2. Read line 2351: pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/XXX.
3. Inspect the immediately preceding entry (the FP8 MI355X ATOM block): pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1310 — a real PR number.
4. PR metadata for this change identifies it as PR Update qwen3.5-fp8-mi300x-sglang SGLang image to v0.5.11-rocm720-mi30x #1353.
5. Therefore the correct value on line 2351 is https://github.com/SemiAnalysisAI/InferenceX/pull/1353; XXX is an unfilled template marker.

Fix

Change line 2351 to:

  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1353

Addressing the refutation

A refuting verifier argued this should be dropped because it duplicates bug_002/bug_003 (which were refuted) and is purely cosmetic. The duplication concern is resolved — synthesis merged the three duplicate reports into this single canonical entry, so this is the surviving (not duplicate) record of the issue. On severity: the refuter is correct that there is no runtime, build, or behavioral impact — only a broken provenance link in a documentation file — so this is filed as nit rather than normal. It's still worth flagging because it's a single one-character fix the author can make before merge, and the surrounding entries set the convention that pr-links resolve.

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25763483436
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25763483436

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=25763483436
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=25763483436