openmoss Diffusion Kernel Skills
Directory Layout
python/sglang/multimodal_gen/.claude/skills/diffusion-kernel/
├── SKILL.md
├── add-triton-kernel.md
├── diffusion-benchmark-and-profile.md
├── nsight-profiler.md
└── use-efficient-diffusion-kernels.md
Index
-
Step-by-step guide for adding a new Triton kernel to SGLang Diffusion's
jit_kernelmodule, including authoring, autotune,torch.compilecompatibility, integration, and tests. use-efficient-diffusion-kernels.md
Practical guidance for using SGLang Diffusion fused kernels and fast CUDA paths, including constraints, fallbacks, and where the fused ops are wired into the runtime.
diffusion-benchmark-and-profile.md
End-to-end benchmarking and profiling guide for SGLang Diffusion models, including denoise latency measurement, per-layer breakdown, and regression tracking.
-
Advanced profiling skill for NVIDIA Nsight Systems / Nsight Compute: collecting traces, reading reports, and interpreting kernel-level performance metrics.