rocm-kernels Skill
Provides guidance for writing and benchmarking optimized Triton kernels for AMD GPUs (MI355X, R9700) on ROCm, targeting HuggingFace diffusers (LTX-Video, SD3, FLUX) and transformers. Core kernels: RMSNorm, RoPE 3D, GEGLU, AdaLN. Includes XCD swizzle, autotune, diffusers integration patterns, and LTX-Video pipeline inje Published by huggingface in kernels.
What is rocm-kernels Skill?
Provides guidance for writing and benchmarking optimized Triton kernels for AMD GPUs (MI355X, R9700) on ROCm, targeting HuggingFace diffusers (LTX-Video, SD3, FLUX) and transformers. Core kernels: RMSNorm, RoPE 3D, GEGLU, AdaLN. Includes XCD swizzle, autotune, diffusers integration patterns, and LTX-Video pipeline inje Published by huggingface in kernels. This profile combines repository metadata with install, compatibility, and usage signals so developers can quickly decide whether it fits their agent workflow before opening the source repository.
Automated repository signals based on public metadata such as recency, license, installation evidence, and adoption. These are not a security audit or endorsement.
Key capabilities
- Includes SKILL.md support
- Reusable instructions support
- Design and media
- Writing
- Design and media use cases
- Writing use cases
Technical details
- Install or run with Copy skill directory
When to use rocm-kernels Skill
- Use it for design and media.
- Use it for writing.
Built with
Editorial notes
Source
- Creator: huggingface
- Repository: huggingface/kernels
- Skill file: kernel-builder/skills/rocm-kernels/SKILL.md
What it does
Provides guidance for writing and benchmarking optimized Triton kernels for AMD GPUs (MI355X, R9700) on ROCm, targeting HuggingFace diffusers (LTX-Video, SD3, FLUX) and transformers. Core kernels: RMSNorm, RoPE 3D, GEGLU, AdaLN. Includes XCD swizzle, autotune, diffusers integration patterns, and LTX-Video pipeline inje
Skill instructions
ROCm Triton Kernels for Diffusers & Transformers This skill provides patterns and guidance for developing optimized Triton kernels targeting AMD GPUs (MI355X, R9700) on ROCm, for use with HuggingFace diffusers (LTX-Video, SD3, FLUX) and transformers libraries. Quick Start Diffusers (LTX-Video) Inject optimized kernels into LTX-Video pipeline: python import os os.environ['TRITONHIPUSEBLOCKPINGPONG'] = '1' os.environ['TRITONHIPUSEASYNCCOPY'] = '1' from diffusers import LTXPipeline pipe = LTXPipeline.frompretrained("Lightricks/LTX-Video", torchdtype=torch.bfloat16) pipe.to("cuda") ROCm uses same API via HIP injectoptimizedkernels(pipe) BEFORE CPU offloading pipe.enablemodelcpuoffload() For a minimal integration example (~150 lines): bash python scripts/ltxkernelinjectionexample.py Isolated Kernel Micro-benchmarks bash All 4 kernels: correctness + performance + bandwidth python scripts/benchmarkkernels.py Single kernel python scripts/benchmarkkernels.py --kernel rmsnorm python scripts/benc
Explore related resources
Frequently asked questions
What is rocm-kernels?
rocm-kernels is a open-source AI agent skill with Copy skill directory. Provides guidance for writing and benchmarking optimized Triton kernels for AMD GPUs (MI355X, R9700) on ROCm, targeting HuggingFace diffusers (LTX-Video, SD3, FLUX) and transformers.
Who is rocm-kernels best for?
rocm-kernels is best for reusing agent instructions, scripts, and references, design and media workflows, writing workflows.
How do I install rocm-kernels?
Install or run rocm-kernels using Copy skill directory. Check rocm-kernels for the latest setup command.
Is rocm-kernels actively maintained?
rocm-kernels may need a closer maintenance check before production use.
Auto-fetched from GitHub.
