OpenSource-Hub
C

cupy

SHA-256
11.6k stars·AI Productivity·SHA-256 checksum verified

CuPy is a NumPy/SciPy-compatible array library for GPU-accelerated computing, offering drop-in replacement to run existing code on NVIDIA CUDA or AMD ROCm platforms.

Smart Download

Download Download Version

v14.1.1 · 138.4 MB

CuPy enables NumPy/SciPy code to run on GPU with minimal changes, providing massive speedups for array operations.

Core Features

  • Drop-in replacement for NumPy/SciPy with identical API
  • Supports both NVIDIA CUDA and AMD ROCm platforms
  • Low-level CUDA features: RawKernels, Streams, Runtime APIs

What It Can't Do

  • The CuPy version must exactly match your CUDA/ROCm version; ROCm support is experimental; CuPy does not fall back to CPU if no GPU is available.

Use Cases

  • Accelerating large-scale scientific computing and data analysis
  • Fast array preprocessing in deep learning pipelines

Detailed Introduction

CuPy provides a high-performance, GPU-accelerated implementation of the NumPy and SciPy APIs. It allows data scientists and engineers to accelerate array operations, linear algebra, Fourier transforms, and more with minimal code changes. CuPy supports both NVIDIA CUDA and AMD ROCm, and includes low-level CUDA features for custom kernels, streams, and runtime APIs. Compared to PyTorch or TensorFlow, CuPy offers a more direct drop-in for traditional NumPy/SciPy workflows, reducing migration effort while delivering significant speedups on GPUs.

Tags

gpunumpyscipypythoncudarocm

Getting Started

1

Download installer

Click the button above to download the installer for your system

2

Install the software

Install the appropriate package for your distro (dpkg / rpm / AppImage)

3

Choose the matching wheel for your CUDA/ROCm version, e.g., pip install cupy-cuda12x

4

Import cupy as cp and use it like numpy

5

Run your code; operations automatically execute on GPU

Install Guide
  1. Choose the matching wheel for your CUDA/ROCm version, e.g., pip install cupy-cuda12x
  2. Import cupy as cp and use it like numpy
  3. Run your code; operations automatically execute on GPU
File Integrity

SHA-256 checksum verified

Checksum extracted from GitHub official Release page

SHA256 Checksum

8889cb83dbb7dbea593e60c85fcc91e21b0ccd10cd5380dfdfaac70b6bd9390a

This checksum is extracted from the GitHub Release page. Verify file integrity after download.

All SHA-256 checksums on this platform are extracted from the project's official GitHub Release page, without any modification. You can independently verify them on the GitHub Releases page.

Open Source Transparency

View GitHub Source
Environment Guide

Uninstall Info

Run pip uninstall cupy-cuda12x (replace with your package name) or conda remove cupy if installed via conda.

No Extra Dependencies

Ready to use after download. No additional runtime required.

Project Info
LicenseMIT
Last Updated2026-06-29T06:45:35Z
GitHub RepositoryOfficial Website

Similar Projects