Numba documentation

This is the Numba documentation. Unless you are already acquainted with Numba, we suggest you start with the User manual.

• 1. User Manual
  • 1.1. Overview
  • 1.2. Getting started
    • 1.2.1. Installing using Conda
    • 1.2.2. Installing from source
    • 1.2.3. Checking your installation
  • 1.3. Compiling Python code with @jit
    • 1.3.1. Basic usage
    • 1.3.2. Calling and inlining other functions
    • 1.3.3. Signature specifications
    • 1.3.4. Compilation options
  • 1.4. Creating Numpy universal functions
    • 1.4.1. The @vectorize decorator
    • 1.4.2. The @guvectorize decorator
  • 1.5. Compiling code ahead of time
  • 1.6. Troubleshooting and tips
    • 1.6.1. What to compile
    • 1.6.2. My code doesn’t compile
    • 1.6.3. The compiled code is too slow
  • 1.7. Examples
    • 1.7.1. Mandelbrot
    • 1.7.2. Moving average
    • 1.7.3. Multi-threading
• 2. Reference Manual
  • 2.1. Numba Types
    • 2.1.1. Basic Types
    • 2.1.2. Advanced Types
  • 2.2. Compilation
    • 2.2.1. JIT functions
    • 2.2.2. Vectorized functions (ufuncs)
  • 2.3. Environment variables
    • 2.3.1. Errors and warnings display
    • 2.3.2. Debugging
    • 2.3.3. Compilation options
    • 2.3.4. GPU support
  • 2.4. Supported Python features
    • 2.4.1. Language
    • 2.4.2. Built-in types
    • 2.4.3. Built-in functions
    • 2.4.4. Standard library modules
    • 2.4.5. Third-party modules
  • 2.5. Supported Numpy features
    • 2.5.1. Scalar types
    • 2.5.2. Array types
    • 2.5.3. Functions
    • 2.5.4. Modules
    • 2.5.5. Standard ufuncs
  • 2.6. Deviations from Python semantics
    • 2.6.1. Integer width
    • 2.6.2. Global and closure variables
• 3. Numba for CUDA GPUs
  • 3.1. Overview
    • 3.1.1. Terminology
    • 3.1.2. Programming model
    • 3.1.3. Requirements
    • 3.1.4. Missing CUDA Features
  • 3.2. Writing CUDA Kernels
    • 3.2.1. Introduction
    • 3.2.2. Kernel declaration
    • 3.2.3. Kernel invocation
    • 3.2.4. Thread positioning
  • 3.3. Memory management
    • 3.3.1. Data transfer
    • 3.3.2. Streams
    • 3.3.3. Shared memory and thread synchronization
    • 3.3.4. Local memory
  • 3.4. Writing Device Functions
  • 3.5. Supported Atomic Operations
    • 3.5.1. Example
  • 3.6. Device management
    • 3.6.1. Device Selection
  • 3.7. Examples
    • 3.7.1. Matrix multiplication
• 4. Developer Manual
  • 4.1. Contributing to Numba
    • 4.1.1. Communication
    • 4.1.2. Getting set up
    • 4.1.3. Development rules
    • 4.1.4. Documentation
  • 4.2. Numba Architecture
    • 4.2.1. Introduction
    • 4.2.2. Contexts
    • 4.2.3. Compiler Stages
  • 4.3. Notes on generators
    • 4.3.1. Terminology
    • 4.3.2. Function analysis
    • 4.3.3. The generator structure
    • 4.3.4. Compiling to native code
  • 4.4. Extending the Numba Frontend
    • 4.4.1. Overview
    • 4.4.2. Tutorial
    • 4.4.3. Conclusion
  • 4.5. Extending the Numba backend
    • 4.5.1. Helper Lib
    • 4.5.2. Python API
    • 4.5.3. Target Interval Objects
    • 4.5.4. Base Target
• 5. Glossary
• 6. Release Notes
  • 6.1. Version 0.18.2
  • 6.2. Version 0.18.1
  • 6.3. Version 0.17.0
  • 6.4. Version 0.16.0
  • 6.5. Version 0.15.1
  • 6.6. Version 0.15
  • 6.7. Version 0.14
  • 6.8. Version 0.13.4
  • 6.9. Version 0.13.3
  • 6.10. Version 0.13.2
  • 6.11. Version 0.13.1
  • 6.12. Version 0.13
  • 6.13. Version 0.12.2
  • 6.14. Version 0.12.1
  • 6.15. Version 0.12
  • 6.16. Version 0.11
  • 6.17. Version 0.10
  • 6.18. Version 0.9
  • 6.19. Version 0.8
  • 6.20. Version 0.7.2
  • 6.21. Version 0.7.1
  • 6.22. Version 0.7
  • 6.23. Version 0.6.1
  • 6.24. Version 0.6
  • 6.25. Version 0.5
  • 6.26. Version 0.4
  • 6.27. Version 0.3.2
  • 6.28. Version 0.3
  • 6.29. Version 0.2