skills.rest
Ritabrata-Chakraborty
Ritabrata-ChakrabortyCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

optimize-for-gpu

Accelerate CPU-bound Python code on NVIDIA GPUs with RAPIDS libraries.

Updated Jun 6, 2026
Search Tags:#python#gpu#cuda#rapids#cuML#cuDF#cuGraph
One-click install
npx skills add https://github.com/Ritabrata-Chakraborty/Claude-Setup --skill optimize-for-gpu-ritabrata-chakraborty
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: optimize-for-gpu
Source: https://github.com/Ritabrata-Chakraborty/Claude-Setup/tree/main/skills/optimize-for-gpu
Command: npx skills add https://github.com/Ritabrata-Chakraborty/Claude-Setup --skill optimize-for-gpu-ritabrata-chakraborty

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires cupy, numba, warp, cudf, cuml, cugraph, cuvs, cuspatial, cucim, cuxfilter, kvikio, raft, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill accelerates CPU-bound Python code using GPU/CUDA/NVIDIA technologies, dramatically improving performance for suitable workloads.

Core Features & Use Cases

  • GPU Array Operations: Utilizes CuPy for array/matrix operations, replacing NumPy for performance improvements.
  • Custom GPU Kernels: Compiles Python code to CUDA kernels with Numba for fine-grained control.
  • Physics Simulations: Uses Warp for efficient physics simulations and differentiable programming.
  • Data Wrangling: cuDF provides pandas-like DataFrame operations on GPU for data manipulation.
  • Machine Learning: cuML offers scikit-learn replacement for GPU-accelerated machine learning.
  • Graph Analytics: cuGraph replaces NetworkX for GPU-accelerated graph algorithms.
  • Vector Search: cuVS performs GPU-accelerated vector search and similarity search.
  • Geospatial Analysis: cuSpatial offers GPU-accelerated geospatial analytics.
  • Image Processing: cuCIM provides GPU-accelerated image processing.
  • File I/O: KvikIO accelerates file reading and writing with GPUDirect Storage.
  • Interactive Dashboards: cuxfilter allows building GPU-accelerated interactive dashboards.
  • Use Case: For example, a machine learning engineer wants to accelerate the training of a large-scale neural network model. They can use this Skill to leverage cuML for GPU-accelerated training.

Quick Start

Use the optimize-for-gpu skill to accelerate your Python code. For example, to accelerate a NumPy array operation, replace import numpy as np with import cupy as cp and change your operations accordingly.

Frequently Asked Questions about optimize-for-gpu

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I accelerate Python code using NVIDIA GPUs?▼

To accelerate Python code using NVIDIA GPUs, you replace CPU-bound libraries with RAPIDS equivalents, such as swapping NumPy for CuPy or pandas for cuDF. This enables GPU-accelerated array operations, machine learning, and graph analytics.

Can I use scikit-learn and pandas syntax for GPU-accelerated machine learning?▼

Yes, you can use scikit-learn and pandas syntax for GPU-accelerated machine learning. The cuML library serves as a scikit-learn replacement and cuDF provides pandas-like DataFrame operations directly on NVIDIA GPUs.

What is the best way to run NetworkX graph algorithms on a GPU?▼

The best way to run NetworkX graph algorithms on a GPU is by using the cuGraph library. It provides GPU-accelerated graph analytics to significantly speed up complex network computations compared to CPU-based approaches.

Do I need a CUDA-capable GPU to use RAPIDS libraries like cuDF and cuML?▼

Yes, you need a CUDA-capable GPU to use RAPIDS libraries like cuDF and cuML. These libraries rely on NVIDIA GPU hardware and CUDA technology to perform numerical, scientific, and machine learning computations.

How do I compile custom Python code to CUDA kernels?▼

To compile custom Python code to CUDA kernels, you can use Numba. It allows you to write custom GPU kernels for fine-grained control over numerical computations and hardware-level optimizations.

Does GPU-accelerated Python support geospatial and image processing workloads?▼

Yes, GPU-accelerated Python supports geospatial and image processing workloads. The cuSpatial library handles geospatial analytics, while cuCIM provides GPU-accelerated medical image processing capabilities.

Related Skills

Explore More Software Engineering Skills→
amdamd
154

quark-env-preflight

Collect OS, Python, GPU, and accelerator facts for Quark installation readiness.

Official
Intermediate
pilotspacepilotspace
2

gpu-bench

Benchmark GPU-accelerated Moon workloads against CPU baselines in CUDA environments.

Official
Intermediate
mkurmanmkurman
321

optimize-for-gpu

Enable GPU acceleration for Python workloads across CUDA-enabled libraries.

Community
Advanced
ovachieverovachiever
52

pytorch-fsdp

Apply PyTorch FSDP for sharded, mixed-precision distributed training of large models.

Community
Advanced