Perf Baseline Generation Script

python/sglang/multimodal_gen/test/scripts/gen_perf_baselines.py starts a local diffusion server, issues requests for selected test cases, aggregates stage/denoise-step/E2E timings from the perf log, and writes the results back to the scenarios section of perf_baselines.json.

Usage

Update a single case:

python python/sglang/multimodal_gen/test/scripts/gen_perf_baselines.py --case qwen_image_t2i

Select by regex:

python python/sglang/multimodal_gen/test/scripts/gen_perf_baselines.py --match 'qwen_image_.*'

Run all keys from the baseline file scenarios:

python python/sglang/multimodal_gen/test/scripts/gen_perf_baselines.py --all-from-baseline

Specify input/output paths and timeout:

python python/sglang/multimodal_gen/test/scripts/gen_perf_baselines.py --baseline python/sglang/multimodal_gen/test/server/perf_baselines.json --out /tmp/perf_baselines.json --timeout 600