[AMD][MI35X] 0612 DSV4

[1am9trash]

Successful run:
see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=27406714195
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=27406714195

Change:

• Bump image to lmsysorg/sglang-rocm:v0.5.13-rocm720-mi35x-20260612.
• Fix the intermediate_pad setting in the MoE computation in sglang PR#27858. This avoids the unnecessary overhead of computing useless padding.
• Correct the chunk prefill setting size under tp8/dp8 config.

---

Note

Low Risk
Benchmark and serving-flag tweaks for one AMD DSv4 recipe; no auth, data, or core app logic changes.

Overview
Updates the dsv4-fp4-mi355x-sglang CI/benchmark pin to lmsysorg/sglang-rocm:v0.5.13-rocm720-mi35x-20260612 (from v0.5.12.post1-…-20260610), pairing the new image with the MoE intermediate_pad fix in upstream sglang PR#27858.

In dsv4_fp4_mi355x_sglang.sh, chunked prefill is no longer hard-coded to 8192: it defaults to 8192 and scales to 8192 * TP when DP_ATTENTION is enabled, so TP8/DP8 runs use the intended prefill chunk size. perf-changelog.yaml records the image bump and both fixes.

^{Reviewed by Cursor Bugbot for commit 8d20601. Bugbot is set up for automated code reviews on this repo. Configure here.}

[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.

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=27406714195
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=27406714195

[1am9trash]

@functionstackx could you please review this?

[functionstackx]

yes. lgtm once passes and then i will do /reuse on it https://github.com/SemiAnalysisAI/InferenceX/actions/runs/27475314602/job/81213395277?pr=1715

[functionstackx]

hi @1am9trash it seems like conc

[functionstackx]

hi @1am9trash it seems like conc512 is failing, can u take a look? (will cancel the rest of the conc for now to avoid clogging up the CI queue since conc512 failed already)

[github-actions]

see unofficial run visualizer at https://inferencex.semianalysis.com/inference?unofficialRun=27475314602
see unofficial run visualizer at https://inferencex.semianalysis.com/evaluation?unofficialRun=27475314602