cuda.core: Pythonic access to CUDA core functionality

Welcome to the documentation for cuda.core.

Contents:

• Overview
  • What is cuda.core?
  • Example: Compiling and Launching a CUDA kernel
  • Examples and Recipes
• Installation
  • Runtime Requirements
  • Installing from PyPI
  • Installing from Conda (conda-forge)
  • Development environment
• 10 minutes to cuda.core
  • Why cuda.core?
  • Before you begin
  • Selecting a device
  • Compiling a kernel
  • Streams
  • Allocating memory
  • Copying and launching
  • Timing with events
  • Working with multiple streams
  • Capturing work in a CUDA graph
  • Working with CuPy and PyTorch
  • Cleaning up
  • Where to go next
• Examples
  • Compilation and kernel launch
  • Multi-device and advanced launch configuration
  • Linking and graphs
  • Interoperability and memory access
  • System inspection
• Interoperability
  • Current device/context
  • __cuda_stream__ protocol
  • Memory view utilities for CPU/GPU buffers
• cuda.core API Reference
  • Devices and execution
  • Memory management
  • CUDA graphs
  • Graphics interoperability
  • Tensor Memory Accelerator (TMA)
  • Textures and surfaces
  • CUDA compilation toolchain
  • CUDA process checkpointing
  • Utility functions
• cuda.core.system API Reference
  • Basic functions
  • Events
  • Types
• Environment Variables
  • Runtime Environment Variables
• Contributing

• cuda.core Support Policy
• Code of Conduct
• Software License Agreement

• Release Notes
  • 1.1.0
  • 1.0.1
  • 1.0.0
  • 0.7.0
  • 0.6.0
  • 0.5.1
  • 0.5.0
  • 0.4.2
  • 0.4.1
  • 0.4.0
  • 0.3.2
  • 0.3.1
  • 0.3.0
  • 0.2.0
  • 0.1.1
  • 0.1.0

Indices and tables

• Index

• Module Index

• Search Page