Upgrade GLM-5 image to v0.5.10

[chunfangamd]

Upgrade GLM5-FP8-MI355X-SGLang Image to v0.5.10rc0-rocm720-mi35x-20260413 after fixing the Slurm shared cache issue. With this upstream daily image, we use a new aiter 0.1.12.post1 which can improve the performance.

    if [[ "$FRAMEWORK" == "atom" ]] || [[ "$FRAMEWORK" == "sglang" ]]; then
        SLRUM_HOME_MOUNT=""
    else
        SLRUM_HOME_MOUNT=" --container-mount-home "
    fi

Co-authored with: @1am9trash @zhentaocc
Thanks to @cquil11 @billishyahao

[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://cookbook.sglang.io/
• https://docs.vllm.ai/projects/recipes/en/latest/

If it is not, please create a PR first before we can merge your 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/sgl-cookbook

[chunfangamd]

@claude please active the e2e test with test-config --config-files .github/configs/amd-master.yaml --config-keys glm5-fp8-mi355x-sglang

[Klaud-Cold]

Claude finished @chunfangamd's task in 3s —— View job

---

E2E Test Triggered for GLM-5

• Understand the request
• Trigger e2e test workflow with test-config --config-files .github/configs/amd-master.yaml --config-keys glm5-fp8-mi355x-sglang --conc 4
• Monitor workflow run — jobs queued, waiting for MI355X runners
• Analyze benchmark results (pending runner availability)

Workflow triggered: Run #24243787005 — glm5-fp8-mi355x-sglang on branch chun/glm5

Jobs created:

Job	Status
get-jobs	✅ Success
glm5_1k1k fp8 mi355x sglang | tp=8 ep=1 conc-4	⌛ Queued (waiting for mi355x runner)
glm5_8k1k fp8 mi355x sglang | tp=8 ep=1 conc-4	⌛ Queued (waiting for mi355x runner)
glm5_8k1k fp8 mi355x sglang | tp=8 ep=1 conc-4 | eval-only	⌛ Queued (waiting for mi355x runner)
multi-node	⏩ Skipped (not applicable)

The workflow has been successfully dispatched. All 3 MI355X benchmark/eval jobs are queued and waiting for self-hosted MI355X runners to become available (~40min in queue so far). Once a runner picks up the jobs, the benchmarks will run automatically. Please monitor the run directly via the link above. You can tag @claude again once the run completes if you'd like me to analyze the results.

--- | View workflow run | Branch: chun/glm5

[chunfangamd]

@claude, I canceled the job https://github.com/SemiAnalysisAI/InferenceX/actions/runs/24243787005.

[Klaud-Cold]

Claude finished @chunfangamd's task in 15s —— View job

---

Acknowledged Job Cancellation

• Noted cancellation of workflow run #24243787005

The e2e test run for glm5-fp8-mi355x-sglang has been canceled as requested.

If you'd like me to re-trigger the tests or help with anything else on this PR, just let me know!

--- | Branch: chun/glm5

[billishyahao]

Hi @chunfangamd This is new sweep based on sgl-dev nightly build:
https://inferencex.semianalysis.com/inference?unofficialRun=24379639570

[cquil11]

@billishyahao @chunfangamd nightly is fine for now, since this is a new model and the pace of development is rapid. feel free to change this PR to the nightly image

[cquil11]

sweep: https://github.com/SemiAnalysisAI/InferenceX/actions/runs/24420536226