CuPy – NumPy-like API accelerated with CUDA

This is the CuPy documentation.

• Overview
• Tutorial
  • Basics of CuPy
  • User-Defined Kernels
• Reference Manual
  • Multi-Dimensional Array (ndarray)
  • Universal Functions (ufunc)
  • Routines
  • SciPy-compatible Routines
  • Sparse matrices
  • Multi-dimensional image processing
  • NumPy-CuPy Generic Code Support
  • Memory Management
  • Low-Level CUDA Support
  • Kernel binary memoization
  • Custom kernels
  • Interoperability
  • Testing Modules
  • Profiling
  • Environment variables
  • Difference between CuPy and NumPy
  • Comparison Table

Development

• API Compatibility Policy
  • Versioning and Backward Compatibilities
  • Processes to Break Backward Compatibilities
  • Supported Backward Compatibility
  • Installation Compatibility
• Contribution Guide
  • Classification of Contributions
  • Development Cycle
  • Issues and Pull Requests
  • Coding Guidelines
  • Unit Testing
  • Documentation

Misc Notes

• Installation Guide
  • Recommended Environments
  • Requirements
  • Install CuPy
  • Install CuPy from conda-forge
  • Install CuPy from Source
  • Uninstall CuPy
  • Upgrade CuPy
  • Reinstall CuPy
  • Run CuPy with Docker
  • FAQ
• [Experimental] Installation Guide for ROCm environemt
  • Recommended Environments
  • Requirements
  • Install CuPy from Source
  • Uninstall CuPy
  • Upgrade CuPy
  • Reinstall CuPy
  • FAQ
• Upgrade Guide
  • CuPy v7
  • CuPy v6
  • CuPy v5
  • CuPy v4
  • CuPy v2
• License
  • NumPy
  • SciPy