[HunyuanImage][Perf] adapt to deploy config changes

[Bounty-hunter]

PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.

Purpose

（1）Remove it from test-nightly.yaml. Actually, the current nightly tests never run it.
（2）Adapt it to deploy_config
（3）Not use fp8

Before:
run_diffusion_benchmark.py starts vLLM Omni via stage_config + CLI.

After:
run_diffusion_benchmark.py starts vLLM Omni via deploy_config + CLI.

These performance tests will run daily on H200 in local testing.

Known issue: further investigation is needed into why TP2_SP2 performance on H200 is below expectations.

Test Plan

pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuan_image_tp4.json

pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuan_image_tp2_sp2.json

pytest -s -v tests/dfx/perf/scripts/run_diffusion_benchmark.py --test-config-file tests/dfx/perf/tests/test_hunyuan_image_tp2_cfgp2.json

Test Result

tp4

tp2 + sp2

tp2 + cfgp2

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan. Please provide the test scripts & test commands. Please state the reasons if your codes don't require additional test scripts. For test file guidelines, please check the test style doc
• The test results. Please paste the results comparison before and after, or the e2e results.
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model. Please run mkdocs serve to sync the documentation editions to ./docs.
• (Optional) Release notes update. If your change is user-facing, please update the release notes draft.

BEFORE SUBMITTING, PLEASE READ https://github.com/vllm-project/vllm-omni/blob/main/CONTRIBUTING.md (anything written below this line will be removed by GitHub Actions)

[chatgpt-codex-connector]

Codex usage limits have been reached for code reviews. Please check with the admins of this repo to increase the limits by adding credits.
Credits must be used to enable repository wide code reviews.

[gcanlin]

Maybe use absolute path for UX.

[Bounty-hunter]

This is equivalent to using an absolute path, because the benchmark sets the working directory to tests after launching the Omni process (

vllm-omni/tests/dfx/perf/scripts/run_diffusion_benchmark.py

Line 635 in f43bb12

cwd=str(Path(__file__).parent.parent.parent.parent),

). Therefore, regardless of the directory from which pytest is executed, the deploy config can always be accessed correctly via the relative path.

[lishunyang12]

Can you leave a issue to keep track of the issue you found for undesirable performance under TP2+SP2?

[Bounty-hunter]

Can you leave a issue to keep track of the issue you found for undesirable performance under TP2+SP2?

track with #4000