compiling Python code using LLVM
Numba is an Open Source NumPy-aware optimizing compiler for Python sponsored by Continuum Analytics, Inc. It uses the remarkable LLVM compiler infrastructure to compile Python syntax to machine code.
It is aware of NumPy arrays as typed memory regions and so can speed-up code using NumPy arrays. Other, less well-typed code will be translated to Python C-API calls effectively removing the “interpreter” but not removing the dynamic indirection.
Numba is also not a tracing JIT. It compiles your code before it gets run either using run-time type information or type information you provide in the decorator.
Numba is a mechanism for producing machine code from Python syntax and typed data structures such as those that exist in NumPy.
- numpy (version 1.6 or higher)
- argparse (for pycc in python2.6)
- funcsigs (for Python 2)
The easiest way to install numba and get updates is by using the Anaconda Distribution: https://store.continuum.io/cshop/anaconda/
$ conda install numba
If you wanted to compile Numba from source, it is recommended to use conda environment to maintain multiple isolated development environments. To create a new environment for Numba development:
$ conda create -p ~/dev/mynumba python numpy llvmlite
To select the installed version, append “=VERSION” to the package name, where, “VERSION” is the version number. For example:
$ conda create -p ~/dev/mynumba python=2.7 numpy=1.6 llvmlite
to use Python 2.7 and Numpy 1.6.
If you need CUDA support, you should also install the CUDA toolkit:
$ conda install cudatoolkit
This installs the CUDA Toolkit version 6.0, which requires driver version 331.00 or later to be installed.
Custom Python Environments
If you’re not using conda, you will need to build llvmlite yourself:
Building and installing llvmlite
See https://github.com/numba/llvmlite for the most up-to-date instructions. You will need a build of LLVM 3.5.
$ git clone https://github.com/numba/llvmlite $ cd llvmlite $ python setup.py install
$ git clone https://github.com/numba/numba.git $ cd numba $ pip install -r requirements.txt $ python setup.py build_ext --inplace $ python setup.py install
$ pip install numba
If you want to enable CUDA support, you will need to install CUDA Toolkit 6.0. After installing the toolkit, you might have to specify environment variables in order to override the standard search paths:
- Path to the CUDA driver shared library
- Path to the CUDA libNVVM shared library file
- Path to the CUDA libNVVM libdevice directory which contains .bc files
Join the numba mailing list firstname.lastname@example.org: https://groups.google.com/a/continuum.io/d/forum/numba-users
or access it through the Gmane mirror: http://news.gmane.org/gmane.comp.python.numba.user
Some old archives are at: http://librelist.com/browser/numba/
See if our sponsor can help you (which can help this project): http://www.continuum.io