PyOpenCL: Pythonic Access to OpenCL, with Arrays and Algorithms

April 6, 2025 · View on GitHub

.. |badge-gitlab-ci| image:: https://gitlab.tiker.net/inducer/pyopencl/badges/main/pipeline.svg :alt: Gitlab Build Status :target: https://gitlab.tiker.net/inducer/pyopencl/commits/main .. |badge-github-ci| image:: https://github.com/inducer/pyopencl/actions/workflows/ci.yml/badge.svg :alt: Github Build Status :target: https://github.com/inducer/pyopencl/actions/workflows/ci.yml .. |badge-pypi| image:: https://badge.fury.io/py/pyopencl.svg :alt: Python Package Index Release Page :target: https://pypi.org/project/pyopencl/ .. |badge-zenodo| image:: https://zenodo.org/badge/1575307.svg :alt: Zenodo DOI for latest release :target: https://zenodo.org/badge/latestdoi/1575307

PyOpenCL lets you access GPUs and other massively parallel compute devices from Python. It tries to offer computing goodness in the spirit of its sister project PyCUDA <https://mathema.tician.de/software/pycuda>__:

Object cleanup tied to lifetime of objects. This idiom, often called RAII <https://en.wikipedia.org/wiki/Resource_Acquisition_Is_Initialization>__ in C++, makes it much easier to write correct, leak- and crash-free code.
Completeness. PyOpenCL puts the full power of OpenCL's API at your disposal, if you wish. Every obscure get_info() query and all CL calls are accessible.
Automatic Error Checking. All CL errors are automatically translated into Python exceptions.
Speed. PyOpenCL's base layer is written in C++, so all the niceties above are virtually free.
Helpful and complete Documentation <https://documen.tician.de/pyopencl>__ as well as a Wiki <https://wiki.tiker.net/PyOpenCL>__.
Liberal license. PyOpenCL is open-source under the MIT license <https://en.wikipedia.org/wiki/MIT_License>__ and free for commercial, academic, and private use.
Broad support. PyOpenCL was tested and works with Apple's, AMD's, and Nvidia's CL implementations.

Simple 4-step install instructions <https://documen.tician.de/pyopencl/misc.html#installation>__ using Conda on Linux and macOS (that also install a working OpenCL implementation!) can be found in the documentation <https://documen.tician.de/pyopencl/>__.

What you'll need if you do not want to use the convenient instructions above and instead build from source:

g++/clang new enough to be compatible with nanobind (specifically, full support of C++17 is needed)
numpy <https://numpy.org>__, and
an OpenCL implementation. (See this howto <https://wiki.tiker.net/OpenCLHowTo>__ for how to get one.)

Links