optimize-for-gpu
GPU-accelerate Python code using CuPy, Numba CUDA, Warp, cuDF, cuML, cuGraph, KvikIO, cuCIM, cuxfilter, cuVS, cuSpatial, and RAFT. Use whenever the user mentions GPU/CUDA/NVIDIA acceleration, or wants to speed up NumPy, pandas, scikit-learn, scikit-image, NetworkX, GeoPandas, or Faiss workloads. Covers physics simulation, differentiable rendering, mesh ray casting, particle systems (DEM/SPH/fluids), vector/similarity search, GPUDirect Storage file IO, interactive dashboards, geospatial analysis, medical imaging, and sparse eigensolvers. Also use when you see CPU-bound Python code (loops, large
npx skills add BioTender-max/awesome-bio-agent-skills --skill optimize-for-gpu --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
You are an expert GPU optimization engineer. Your job is to help users write new GPU-accelerated code or transform their existing CPU-bound Python code to run on NVIDIA GPUs for dramatic speedups — often 10x to 1000x for suitable workloads.
How it works
The skill outlines decision guidance on when and which GPU libraries to use based on the user’s workload. It describes selecting tools for array/matrix ops (CuPy), custom kernels (Numba CUDA), simulation/geometry and differentiable programming (Warp), dataframe operations (cuDF), ML/ML preprocessing (cuML), graph analytics (cuGraph), high-performance file IO (KvikIO), image processing (cuCIM), vector/search workloads (cuVS), geospatial analytics (cuSpatial), and low-level primitives (RAFT). It provides recommended usage patterns, typical speedups, and interoperability notes, including how to combine libraries via the CUDA Array Interface and common cross-library workflows (e.g., cuDF + cuML, cuGraph + cuML, Warp + PyTorch).
When to use it
Use this skill when the user wants to speed up numerical/scientific Python code, works with large arrays/dataframes, mentions CUDA/GPU/NVIDIA, or aims to accelerate scikit-learn, scikit-image, NetworkX, GeoPandas, Faiss workloads. Also apply when the task involves physics simulation, mesh operations, particle systems, vector search, geospatial or medical imaging workloads, or IO-bound pipelines that would benefit from GPU memory residency.
What it can touch
The skill explicitly references integrating with: CuPy, Numba CUDA, Warp, cuDF, cuML, cuGraph, KvikIO, cuCIM, cuVS, cuSpatial, RAFT, PyTorch, JAX. It notes zero-copy data sharing via the CUDA Array Interface to enable interoperation between these libraries.
Caveats
The text does not specify licensing terms for individual libraries beyond the stated license of the skill as NOASSERTION. It also notes that higher-level RAPIDS libraries are preferred first, with RAFT primitives as fallback for low-level needs; no guarantees about specific performance figures beyond typical ranges.”} } <|> ```json {
# GPU Optimization for Python with NVIDIA You are an expert GPU optimization engineer. Your job is to help users write new GPU-accelerated code or transform their existing CPU-bound Python code to run on NVIDIA GPUs for dramatic speedups — often 10x to 1000x for suitable workloads. ## When This Skill Applies - User wants to speed up numerical/scientific Python code - User is working with large arrays, matrices, or dataframes - User mentions CUDA, GPU, NVIDIA, or parallel computing - User has NumPy, pandas, SciPy, scikit-learn, NetworkX, or scipy.sparse.linalg code that processes large datasets - User needs low-level GPU primitives (sparse eigensolvers, device memory management, multi-GPU communication) - User is doing machine learning (training, inference, hyperparameter tuning, preprocessing) - User is doing graph analytics (centrality, community detection, shortest paths, PageRank, etc.) - User is doing vector search, nearest neighbor search, similarity search, or building a RAG pipeline - User has Faiss, Annoy, ScaNN, or sklearn NearestNeighbors code that could be GPU-accelerated - User wants GPU-accelerated interactive dashboards, cross-filtering, or exploratory data analysis o
- When This Skill Applies
- Decision Framework: Which Library to Use
- CuPy — for array/matrix operations (NumPy replacement)
- Numba CUDA — for custom GPU kernels
- Warp — for simulation, spatial computing, and differentiable programming
- cuDF — for dataframe operations (pandas replacement)
- cuML — for machine learning (scikit-learn replacement)
- cuGraph — for graph analytics (NetworkX replacement)
- KvikIO — for high-performance GPU file IO
- cuxfilter — for GPU-accelerated interactive dashboards
- cuCIM — for image processing (scikit-image replacement)
- cuVS — for vector search (Faiss/Annoy replacement)
- cuSpatial — for geospatial analytics (GeoPandas replacement)
- RAFT (pylibraft) — for low-level GPU primitives and multi-GPU
CuPy (choose the right CUDA version) uv add cupy-cuda12x # For CUDA 12.x (most common) Numba with CUDA support uv add numba numba-cuda # numba-cuda is the actively maintained NVIDIA package Warp (simulation, spatial computing, differentiable programming) uv add warp-lang # CUDA 12 runtime included cuDF (RAPIDS) uv add --extra-index-url=https://pypi.nvidia.com cudf-cu12 # For CUDA 12.x For cudf.pandas accelerator mode, that's all you need Load it with: python -m cudf.pandas your_script.py
What does the optimize-for-gpu skill do?
GPU-accelerate Python code using CuPy, Numba CUDA, Warp, cuDF, cuML, cuGraph, KvikIO, cuCIM, cuxfilter, cuVS, cuSpatial, and RAFT. Use whenever the user mentions GPU/CUDA/NVIDIA acceleration, or wants to speed up NumPy, pandas, scikit-learn, scikit-image, NetworkX, GeoPandas, or Faiss workloads. Covers physics simulation, differentiable rendering, mesh ray casting, particle systems (DEM/SPH/fluids), vector/similarity search, GPUDirect Storage file IO, interactive dashboards, geospatial analysis, medical imaging, and sparse eigensolvers. Also use when you see CPU-bound Python code (loops, large
How do I install it?
Run `npx skills add BioTender-max/awesome-bio-agent-skills --skill optimize-for-gpu --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From BioTender-max/awesome-bio-agent-skills, a repository with 135 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.