pyopencl
Python wrapper for OpenCL
Decision gist · record as of 2026-08-14
Yes, if you need GPU acceleration and want to target diverse hardware via OpenCL. The package is production-stable, actively maintained, permissively licensed, and widely used. Install friction is moderate due to compiled dependencies and OpenCL driver requirements, but pre-built wheels and conda packages ease setup. No known security vulnerabilities. Not necessary if you are locked into a single vendor (CUDA-only or Metal-only) or do not need GPU compute.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires an OpenCL implementation (vendor drivers from Apple, AMD, Nvidia, or open-source alternatives) installed on your system; pre-built wheels simplify setup but do not include OpenCL drivers themselves.
- Medium install friction due to compiled C++ components and OpenCL driver dependencies.
- The package is actively maintained with recent commits and offers pre-built wheels for common platforms (Linux, macOS, Windows across multiple architectures), though building from source requires a C++17-compatible compiler and an OpenCL implementation.
License · maintenance · safety
MIT (permissive) — MIT license permits commercial, academic, and private use with minimal restrictions, making it suitable for proprietary and open-source projects alike.
last release 2026-01-16 (210 days) · last repo commit 2026-08-10 · 1,149 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 175,184 downloads/mo, #10,268 on PyPI
Alternatives
Verify before relying
pip install pyopencl
import pyopencl as cl
import numpy as np
platforms = cl.get_platforms()
devices = platforms[0].get_devices()
ctx = cl.Context(devices)
queue = cl.CommandQueue(ctx)- Whether pre-built wheels bundle OpenCL runtime or only headers/bindings
- Performance characteristics on specific hardware (GPUs, accelerators) relative to alternatives
- Maturity of support for recent OpenCL standards beyond 1.2
What it is and what it does
PyOpenCL is a Python wrapper around the OpenCL API that lets you write code to run on GPUs and other massively parallel compute devices. It follows the design philosophy of its sister project PyCUDA, offering automatic resource cleanup via RAII patterns, complete access to OpenCL's API surface, automatic error translation to Python exceptions, and a C++-based implementation for performance. The package supports Apple, AMD, and Nvidia OpenCL implementations and is actively maintained with broad platform coverage.
Typical use involves creating an OpenCL context tied to available compute devices, building or loading kernels written in OpenCL C, and managing data transfer between host and device memory. The runtime dependencies (numpy, pytools, platformdirs, importlib-resources, typing_extensions) support array handling, platform detection, and utility functions. Installation via conda or pre-built wheels is recommended to avoid building from source, which requires a C++17 compiler and OpenCL headers.
Use it for
- Accelerate numerical computations (linear algebra, FFTs, simulations) by offloading to GPU
- Implement custom compute kernels for image processing or signal processing on parallel hardware
- Prototype heterogeneous computing workflows that mix CPU and GPU workloads in a single Python script
- Deploy machine learning inference on embedded or mobile GPUs via OpenCL
- Run scientific simulations (physics, chemistry) that benefit from massively parallel execution
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you need GPU acceleration and want to target diverse hardware via OpenCL.
The package is production-stable, actively maintained, permissively licensed, and widely used. Install friction is moderate due to compiled dependencies and OpenCL driver requirements, but pre-built wheels and conda packages ease setup. No known security vulnerabilities. Not necessary if you are locked into a single vendor (CUDA-only or Metal-only) or do not need GPU compute.
Install
pyopencl on PyPI
Before you install
Medium install friction due to compiled C++ components and OpenCL driver dependencies. The package is actively maintained with recent commits and offers pre-built wheels for common platforms (Linux, macOS, Windows across multiple architectures), though building from source requires a C++17-compatible compiler and an OpenCL implementation.
Requires an OpenCL implementation (vendor drivers from Apple, AMD, Nvidia, or open-source alternatives) installed on your system; pre-built wheels simplify setup but do not include OpenCL drivers themselves.
License in practice
MIT license permits commercial, academic, and private use with minimal restrictions, making it suitable for proprietary and open-source projects alike.
Quickstart
pip install pyopencl
import pyopencl as cl
import numpy as np
platforms = cl.get_platforms()
devices = platforms[0].get_devices()
ctx = cl.Context(devices)
queue = cl.CommandQueue(ctx)Verify before relying
- Whether pre-built wheels bundle OpenCL runtime or only headers/bindings
- Performance characteristics on specific hardware (GPUs, accelerators) relative to alternatives
- Maturity of support for recent OpenCL standards beyond 1.2
Package facts
| License | MIT permissive |
| Python support | Supports the current Python release ~=3.10 |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 5 packagesimportlib-resourcesnumpyplatformdirspytoolstyping_extensions |
| Maintenance | Actively maintained 210 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 175,184 / month, #10,268 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Development Status :: 5 - Production/StableEnvironment :: ConsoleIntended Audience :: DevelopersIntended Audience :: Other AudienceIntended Audience :: Science/ResearchNatural Language :: EnglishProgramming Language :: C++Programming Language :: PythonProgramming Language :: Python :: 3 :: OnlyTopic :: Scientific/EngineeringTopic :: Scientific/Engineering :: MathematicsTopic :: Scientific/Engineering :: Physics |
Evidence: pyopencl-2026.1.2-cp310-cp310-macosx_10_14_x86_64.whl; pyopencl-2026.1.2-cp310-cp310-macosx_11_0_arm64.whl; pyopencl-2026.1.2-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; pyopencl-2026.1.2-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pyopencl-2026.1.2-cp310-cp310-musllinux_1_2_aarch64.whl; pyopencl-2026.1.2-cp310-cp310-musllinux_1_2_x86_64.whl; pyopencl-2026.1.2-cp310-cp310-win_amd64.whl; pyopencl-2026.1.2-cp311-cp311-macosx_10_14_x86_64.whl; pyopencl-2026.1.2-cp311-cp311-macosx_11_0_arm64.whl; pyopencl-2026.1.2-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; pyopencl-2026.1.2-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pyopencl-2026.1.2-cp311-cp311-musllinux_1_2_aarch64.whl; pyopencl-2026.1.2-cp311-cp311-musllinux_1_2_x86_64.whl; pyopencl-2026.1.2-cp311-cp311-win_amd64.whl; pyopencl-2026.1.2-cp312-cp312-macosx_10_14_x86_64.whl; pyopencl-2026.1.2-cp312-cp312-macosx_11_0_arm64.whl; pyopencl-2026.1.2-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; pyopencl-2026.1.2-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pyopencl-2026.1.2-cp312-cp312-musllinux_1_2_aarch64.whl; pyopencl-2026.1.2-cp312-cp312-musllinux_1_2_x86_64.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “OpenCL Python bindings”
- pyopenclPyOpenCL provides Python bindings to OpenCL, enabling you to write…
- intel-opencl-rtProvides Intel OpenCL runtime libraries needed to run OpenCL…
- onemkl-sycl-sparseProvides C and Data Parallel C++ (DPC++) interfaces to Intel's oneAPI…
Give your agent the search over MCP, or paste the wish link into any chat.
More Scientific/Engineering packages
NumPy provides an N-dimensional array object and a comprehensive suite of mathematical, linear algebra, Fourier transform, and random number functions for scientific computing in Python.
pandas provides fast, flexible data structures (Series and DataFrame) for loading, cleaning, transforming, and analyzing labeled or relational data in Python.
scipy provides numerical algorithms for mathematics, science, and engineering—including optimization, integration, linear algebra, Fourier transforms, signal and image processing, and ODE solvers—built on numpy arrays.
scikit-learn provides a comprehensive Python library for supervised and unsupervised machine learning, including classification, regression, clustering, dimensionality reduction, and model evaluation tools built on NumPy and SciPy.
Install it if you need to train, evaluate, or deploy supervised or unsupervised learning models.
dill extends Python's pickle module to serialize and deserialize a much wider range of Python objects, including functions, lambdas, classes, and interpreter sessions, to byte streams for storage or network transmission.
Multiprocess is an enhanced fork of Python's standard multiprocessing library that uses dill for better serialization, allowing you to spawn processes with a threading-like API and share complex objects between them.
Install it if you use multiprocessing and encounter pickle serialization limits with lambdas or complex objects.
See also pycuda · nvidia-cuda-cccl · nvidia-cuda-cccl-cu12 · nvidia-cuda-crt · nvidia-cuda-runtime · nvidia-cublas-cu12 · nvidia-cufile · cuda-python · nvidia-cuda-runtime-cu11 · nvidia-cufile-cu12