optimize-for-gpu
GPU-accelerates scientific Python on NVIDIA hardware and verifies
by k-dense-aiFree
Description
optimize-for-gpu is an agent skill for speeding up scientific Python on NVIDIA hardware with GPU acceleration, and for checking that the result is both correct and faster. It applies to CUDA and GPU optimisation of CPU-bound NumPy, SciPy and similar code. The skill treats GPU acceleration as an evidence-driven optimisation, not an automatic rewrite. It preserves the numerical and algorithmic behaviour the user depends on, measures with representative data, and keeps the GPU version only when synchronised end-to-end benchmarks show a useful improvement. The skill was reviewed against RAPIDS 26.08, CuPy 14.2, Numba-CUDA 0.30.4 and Warp 1.17. Its reference files contain illustrative GPU examples, and the review checked the source and the API but did not run them on CUDA hardware. The skill tells you to validate the examples on your own target GPU before you report correctness or a speed-up. It also warns against using a moving, latest documentation page to infer compatibility with a pinned release. Use it when a Python workload is slow on the CPU and you want to know whether a GPU version is worth the effort.…
Highlights
- GPU acceleration for NumPy and SciPy code
- Evidence-driven, measured optimisation
- End-to-end benchmarks before keeping a GPU version
- Covers RAPIDS, CuPy, Numba-CUDA and Warp
- Warns that examples need validation on your GPU
Ratings & Reviews
No ratings yet
Be the first to rate optimize-for-gpu.
People also visit
optimize-for-gpu FAQ
What does the skill do?
It considers GPU acceleration for CPU-bound scientific Python, and keeps the GPU version only if benchmarks show a useful improvement.
Does it rewrite code automatically?
No. The skill says GPU acceleration is an evidence-driven optimisation, not an automatic rewrite.
Which libraries does it cover?
RAPIDS, CuPy, Numba-CUDA and Warp, reviewed against RAPIDS 26.08, CuPy 14.2, Numba-CUDA 0.30.4 and Warp 1.17.
Were the GPU examples run?
The review checked the source and API, but did not run the examples on CUDA hardware. The skill says to validate them on your own GPU.
How do I install it?
Run npx skills add https://github.com/k-dense-ai/scientific-agent-skills --skill optimize-for-gpu.
Is it checked for security?
Its security audits show passes from Gen Agent Trust Hub, Socket and Snyk.
What are the best optimize-for-gpu alternatives?
Popular alternatives to optimize-for-gpu include diagnosing-bugs, benchmark-optimization-loop and content-hash-cache-pattern. See all optimize-for-gpu alternatives