cupy/cupy
NumPy & SciPy for GPU
Python12K stars1.1K forks+352 today
Topics
#cuda#cudnn#cublas#cusolver#nccl#python#numpy#cupy#curand#cusparse#gpu#cutensor#scipy#nvtx#nvrtc#tensor#cusparselt#rocm
What it does
CuPy is a GPU-accelerated library for Python that serves as a drop-in replacement for NumPy and SciPy, enabling high-performance computing on NVIDIA CUDA or AMD ROCm platforms. It matters because it allows developers to leverage the power of GPUs without significant changes to their existing codebase.
Star history
Not enough history yet — 1 day(s) recorded. The daily snapshot builds this up.
Tracking
- Last trending
- 2026-07-20
Creator kit
Hook
🚀 Speed up your Python data processing with CuPy, a GPU-accelerated library for NumPy & SciPy that boosts performance on NVIDIA and AMD GPUs. #GPUComputing
Content angles
- Creating tutorials on how to optimize existing NumPy/SciPy code using CuPy
- Benchmarking the performance gains of CuPy over CPU-based libraries in various data processing tasks
- Exploring advanced features like RawKernels for custom CUDA programming
Who should care
Python developers, data scientists, and machine learning engineers interested in GPU-accelerated computing