菜单
Use when benchmarking denoise latency or profiling a diffusion bottleneck in SGLang.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Step-by-step tutorial for adding a new lightweight JIT CUDA kernel to sglang's jit_kernel module
Step-by-step tutorial for adding a heavyweight AOT CUDA/C++ kernel to sgl-kernel (including tests & benchmarks)
Guide to SGLang CI workflow orchestration — stage ordering, fast-fail, gating, partitioning, execution modes, and debugging CI failures. Use when modifying CI workflows, adding stages, debugging CI pipeline issues, or understanding how tests are dispatched and gated across stages.
Generate an e2e profiling trace of an SGLang server run. Launches a server, validates accuracy, captures a Chrome-compatible trace, and returns the profile path.
Run SGLang auto benchmark searches with tiered server-flag sweeps, canonical dataset preparation, ShareGPT auto-download, custom-data conversion/validation, SLA or fixed-QPS benchmarking, CSV export, and optional second-stage speculative/EAGLE tuning. Use when the user wants an AI-operated benchmark workflow rather than a one-off bench_serving command.
Compact SGLang torch-profiler triage skill. Use when Codex should inspect an existing `trace.json(.gz)` or profile directory, trigger `sglang.profiler` against a live server, and return one compact report with kernel, overlap-opportunity, and fuse-pattern tables. Single-trace triage is enough for quick diagnosis; mapping+formal two-trace triage gives stronger overlap conclusions.
| name | sglang-diffusion-benchmark-profile |
| description | Use when benchmarking denoise latency or profiling a diffusion bottleneck in SGLang. |
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:
torch.profiler trace capture and quick hotspot rankingThis skill does not own low-level kernel authoring or standalone Nsight workflows.
Before running any benchmark, profiler, or kernel-validation command:
scripts/diffusion_skill_env.py to derive the repo root from sglang.__file__HF_TOKEN before using gated Hugging Face models such as black-forest-labs/FLUX.*FLASHINFER_DISABLE_VERSION_CHECK=1torch.profiler workflow; uses the checked-in nightly-aligned presets, including LTX-2 two-stageQK norm + RoPE, and distributed overlap patterns before proposing new codesglang.__file__, write-access probe, benchmark/profile output directories, idle GPU selectionsglang generate; use --list-models to inspect preset order, then save perf dumps by label and compare them with compare_perf.pyBefore calling a diffusion hotspot "new", first classify it with existing-fast-paths.md.
Always rule out these existing families first:
QK norm + RoPEtorch.compile compute / communication reorder