Sglang Diffusion Benchmark Profile
Testing & Qualitysglang-diffusion-benchmark-profileSglang Diffusion Benchmark Profile is an Agent Skill for testing & quality published in the FutureMLS-Lab/OSCAR repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Use when benchmarking denoise latency or profiling a diffusion bottleneck in SGLang.
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Inspect the Agent Skill at https://github.com/FutureMLS-Lab/OSCAR/blob/HEAD/sglang-research/python/sglang/multimodal_gen/.claude/skills/sglang-diffusion-benchmark-profile/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/sglang-diffusion-benchmark-profile, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
SGLang Diffusion Benchmark and Profile
Use this skill when measuring denoise performance, finding the slow op, checking whether an existing fast path can solve it, or verifying that a hotspot is real before any kernel work in sglang.multimodal_gen.
This skill is diagnosis-first. It owns:
- checked-in denoise benchmark presets
- perf dump collection and before/after comparison
torch.profilertrace capture and quick hotspot ranking- mapping hot kernels back to known fast paths and fusion families
- handing confirmed kernel work to a specialized optimization skill such as ../sglang-diffusion-ako4all-kernel/SKILL.md
This skill does not own low-level kernel authoring or standalone Nsight workflows.
Preflight
Before running any benchmark, profiler, or kernel-validation command:
- use
scripts/diffusion_skill_env.pyto derive the repo root fromsglang.__file__ - verify the repo is writable
- export
HF_TOKENbefore using gated Hugging Face models such asblack-forest-labs/FLUX.* - export
FLASHINFER_DISABLE_VERSION_CHECK=1 - choose idle GPU(s) before starting perf work
Main Reference
- benchmark-and-profile.md — canonical denoise benchmark, perf dump, and
torch.profilerworkflow; uses the checked-in nightly-aligned presets, includingLTX-2two-stage - existing-fast-paths.md — map bottlenecks to existing fused kernels, packed QKV paths, fused
QK norm + RoPE, and distributed overlap patterns before proposing new code - scripts/diffusion_skill_env.py — preflight helper: repo root discovery via
sglang.__file__, write-access probe, benchmark/profile output directories, idle GPU selection - scripts/bench_diffusion_denoise.py — end-to-end denoise benchmark preset runner via
sglang generate; use--list-modelsto inspect preset order, then save perf dumps by label and compare them withcompare_perf.py
Opportunity Discovery Rule
Before calling a diffusion hotspot "new", first classify it with existing-fast-paths.md.
Always rule out these existing families first:
- merged Z-Image residual-form modulation
- fused diffusion
QK norm + RoPE - NVFP4 / Nunchaku packed QKV
- Nunchaku fused GELU MLP
- Ulysses / USP attention overlap
- turbo-layer async all-to-all overlap
torch.compilecompute / communication reorder- dual-stream diffusion execution