Installation¶
Installing from Conda Forge¶
Installing PyOpenCL¶
By far the easiest way to install PyOpenCL is to use the packages available in Conda Forge. Conda Forge is a repository of community-maintained packages for the Conda package manager. The following instructions are aimed at Linux and macOS. The analogous steps for Windows should also work.
Install a version of miniforge that fits your system:
curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
bash ./Miniforge3-*.sh
# (answer questions, pick install location)
Then run:
source /WHERE/YOU/INSTALLED/MINIFORGE/bin/activate root
conda install pyopencl
You can install these pieces of software in your user account and do not need root/administrator privileges.
Note
This installs a conda environment based on Conda Forge. This is not interchangeable with a conda environment based on the (more common) anaconda. If you have an existing conda environment sitting around, just following the instructions below will likely not work. Instead, the suggested approach is to create new environment from scratch, starting with miniforge, above.
Enabling access to CPUs and GPUs via (Py)OpenCL¶
Note that PyOpenCL is no fun (i.e. cannot run code) without an OpenCL device
driver (a so-called âICDâ, for âinstallable client driverâ) that provides access
to hardware through OpenCL. If you get an error message like
pyopencl._cl.LogicError: clGetPlatformIDs failed: PLATFORM_NOT_FOUND_KHR
,
that means PyOpenCL installed successfully, but you have no OpenCL drivers installed.
Note that drivers (ICDs) are separate pieces of software from PyOpenCL. They might be provided by your hardware vendor (e.g. for Nvidia or AMD GPUs). If you have such hardware, see below for instructions on how to make those work with PyOpenCL from Conda Forge.
It is important to note that OpenCL is not restricted to GPUs. In fact, no special hardware is required to use OpenCL for computationâyour existing CPU is enough. On Linux or macOS, type:
conda install pocl
to install a CPU-based OpenCL driver. On macOS, PoCL can offer a marked robustness (and, sometimes, performance) improvement over the OpenCL drivers built into the operating system.
On Linux and Windows, you can use Intelâs CPU OpenCL runtime:
conda install intel-opencl-rt
On Linux Intel Broadwell or newer processors with an Intel graphics card, you can use NEO:
conda install intel-compute-runtime
On Linux Intel Sandybridge or newer processors with an Intel graphics card, you can use Beignet:
conda install beignet
On Linux, Windows and macOS, you can use Oclgrind to detect memory access errors:
conda install oclgrind
You are now ready to run code based on PyOpenCL, such as the code examples.
Using vendor-supplied OpenCL drivers (mainly on Linux)¶
The instructions above help you get a basic OpenCL environment going that will work independently of whether you have specialized hardware (such as GPUs or FPGAs) available. If you do have such hardware, read on for how to make it work.
On Linux, PyOpenCL finds which drivers are installed by looking for files with
the extension .icd
in a directory. PyOpenCL as installed from Conda will
look for these files in
/WHERE/YOU/INSTALLED/MINICONDA/etc/OpenCL/vendors
. They are just
simple text files containing either just the file names or the fully
qualified path names of the shared library providing the OpenCL driver.
Note
If you ran the commands above in a
Conda environment
(i.e. if the environment indicator on your command line prompt says anything other
than (root)
), then you may need to use a path like the following instead:
/WHERE/YOU/INSTALLED/MINICONDA/envs/ENVIRONMENTNAME/etc/OpenCL/vendors
Note that you should replace ENVIRONMENTNAME
with the name of your environment,
shown between parentheses on your command line prompt.
This path (for the currently-active conda environment) can be obtained from the
environment variable CONDA_PREFIX
, i.e.,
$CONDA_PREFIX/etc/OpenCL/vendors
(once the Conda environment is activated).
On Linux, if you have other OpenCL drivers installed (such as for your GPU),
those will be in /etc/OpenCL/vendors
. You can make them work with PyOpenCL
from Conda Forge by using the command:
conda install ocl-icd-system
will make sure these system-wide ICDs are also visible in your conda environment.
As an alternative, one may manually copy ICD files from /etc/OpenCL/vendors
into, e.g., $CONDA_PREFIX/etc/OpenCL/vendors
.
If you are looking for more information, see ocl-icd and its documentation. Ocl-icd is the
âICD loaderâ used by PyOpenCL when installed from Conda Forge on Linux.
It represents the code behind libOpenCL.so
.
On macOS, using the command:
conda install ocl_icd_wrapper_apple
will make sure that the Apple provided CPU and GPU implementations are available.
On Windows, the packaging of PyOpenCL for Conda Forge relies on the Khronos ICD Loader, and it is packaged so that the OpenCL drivers that are registered in the OS using registry keys are automatically available.
Installing from PyPI wheels¶
PyOpenCL distributes wheels for most popular OSs and Python versions. To check available versions please visit PyPI page for PyOpenCL.
On Linux, the wheels come with OCL-ICD
bundled and configured to use any OpenCL implementation supporting the ICD
interface and listed in /etc/OpenCL/vendors
.
Wheels for Windows and MacOS are built using the ICD Loader from the Khronos Group.
To install, type:
pip install pyopencl
You can also install the following CPU based OpenCL implementation using pip shipped as binary wheels. Note that pyopencl has to be installed using a wheel for pyopencl to recognize these wheels.
To install pyopencl with PoCL, a CPU based implementation do:
pip install pyopencl[pocl]
To install pyopencl with oclgrind, an OpenCL debugger do:
pip install pyopencl[oclgrind]
Note
Avoid mixing components installed from Conda Forge and PyPI. For example, installing PyOpenCL from pip followed by OCL-ICD from Conda Forge can redirect the ICD loader, removing access to system-wide ICDs.
Installing from source¶
Installing PyOpenCL from source should mostly not be necessary unless you
have very specific needs or would like to modify PyOpenCL yourself.
You can find generic installation instructions for nanobind
-based packages here.
For PyOpenCL, the basic process is as follows:
$ cd pyopencl
# non-editable install:
$ pip install -v .
# editable install - make sure to disable build isolation:
$ pip install nanobind scikit-build-core[pyproject] numpy ninja
$ pip install --no-build-isolation -ve .
# editable install with automatic recompilation if needed (somewhat experimental):
$ pip install --no-build-isolation -Ceditable.rebuild=true -Cbuild-dir=build -ve .
PyOpenCL will attempt to automatically find and use the OpenCL headers and libraries while building. You can also specify the paths to the OpenCL headers and libraries manually:
# Option 1: specify the paths via environment variables:
$ export CL_INC_DIR=</path/to/OpenCL/include>
$ export CL_LIB_DIR=</path/to/OpenCL/lib>
$ export CL_LIBNAME=<OpenCL libname>
# Option 2: specify the paths via arguments to pip install:
$ pip install -v . --config-settings='cmake.args=-DCL_INC_DIR=/path/to/OpenCL/include;-DCL_LIB_DIR=/path/to/OpenCL/lib'
Tips¶
Syntax highlighting¶
You can obtain Vim syntax highlighting for OpenCL C inlined in Python by checking this file.
Note that the triple-quoted strings containing the source must start with
"""//CL// ..."""
.
IPython integration¶
PyOpenCL comes with IPython integration, which lets you seamlessly integrate PyOpenCL kernels into your IPython notebooks. Simply load the PyOpenCL IPython extension using:
%load_ext pyopencl.ipython_ext
and then use the %%cl_kernel
âcell-magicâ command. See this notebook
(which ships with PyOpenCL) for a demonstration.
You can pass build options to be used for building the program executable by
using the -o
flag on the first line of the cell (next to the %%cl_kernel
directive). For example:
%%cl_kernel -o "-cl-fast-relaxed-math"
There are also line magics: cl_load_edit_kernel
which will load a file into
the next cell (adding cl_kernel
to the first line) and cl_kernel_from_file
which will compile kernels from a file (as if you copy-and-pasted the contents of
the file to a cell with cl_kernel
). Both of these magics take options -f
to specify the file and optionally -o
for build options.
Added in version 2014.1.
Guidelines¶
API Stability¶
I consider PyOpenCLâs API âstableâ. That doesnât mean it canât change. But if it does, your code will generally continue to run. It may however start spewing warnings about things you need to change to stay compatible with future versions.
Deprecation warnings will be around for a whole year, as identified by the first number in the release name. (the â2014â in â2014.1â) I.e. a function that was deprecated in 2014.n will generally be removed in 2015.n (or perhaps later). Further, the stability promise applies for any code thatâs part of a released version. It doesnât apply to undocumented bits of the API, and it doesnât apply to unreleased code downloaded from git.
Relation with OpenCLâs C Bindings¶
Weâve tried to follow these guidelines when binding the OpenCLâs C interface to Python:
Remove the
cl_
,CL_
andcl
prefix from data types, macros and function names.Follow PEP 8, i.e.
Make function names lowercase.
If a data type or function name is composed of more than one word, separate the words with a single underscore.
get_info
functions become attributes.Object creation is done by constructors, to the extent possible. (i.e. minimize use of âfactory functionsâ)
If an operation involves two or more âcomplexâ objects (like e.g. a kernel enqueue involves a kernel and a queue), refuse the temptation to guess which one should get a method for the operation. Instead, simply leave that command to be a function.
Interoperability with other OpenCL software¶
Just about every object in pyopencl
supports the following
interface (here shown as an example for pyopencl.MemoryObject
,
from which pyopencl.Buffer
and pyopencl.Image
inherit):
This allows retrieving the C-level pointer to an OpenCL object as a Python
integer, which may then be passed to other C libraries whose interfaces expose
OpenCL objects. It also allows turning C-level OpenCL objects obtained from
other software to be turned into the corresponding pyopencl
objects.
Added in version 2013.2.
User-visible Changes¶
Unreleased¶
Note
This version is currently under development. You can get snapshots from PyOpenCLâs git repository.
Version 2022.2¶
Added opaque-style SVM and
pyopencl.SVMPointer
.Added
pyopencl.tools.SVMPool
.Added automatic queue-synchronized deallocation of SVM.
Version 2020.2¶
Drop Python 2 support.
Add
allow_empty_ndrange
to kernel enqueue.Bug fixes.
Version 2018.2¶
Use pybind11.
Many bug fixes.
Support arrays with offsets in scan kernels.
Version 2018.1¶
Introduce eliminate_empty_output_lists argument of
pyopencl.algorithm.ListOfListsBuilder
.Many bug fixes.
Version 2017.2¶
Many bug fixes.
Version 2017.1¶
Introduce
pyopencl.cltypes
Version 2016.2¶
Deprecate RANLUXCL. It will be removed in the 2018.x series of PyOpenCL.
Introduce Random123 random number generators. See
pyopencl.clrandom
for more information.Add support for range and slice kwargs and data-less reductions to
pyopencl.reduction.ReductionKernel
.Add support for SPIR-V. (See
pyopencl.Program
.)Add support for Shared Virtual Memory (SVM).
pyopencl.MemoryMap
is usable as a context manager.
Version 2016.1¶
The
from_int_ptr
methods now take a retain parameter for more convenient ownership management.Kernel build options (if passed as a list) are now properly quoted. (This is a potentially compatibility-breaking change.)
Many bug fixes. (GL interop, Windows, event callbacks and more)
Version 2015.2.4¶
Fix building on Windows, using mingwpy and VS 2015.
Version 2015.2.3¶
Fix one more Ubuntu 14.x build issue.
Version 2015.2.2¶
Fix compatibility with CL 1.1
Fix compatibility with Ubuntu 14.x.
Various bug fixes
Version 2015.2.1¶
Fix global_offset kernel launch parameter
Version 2015.2¶
[INCOMPATIBLE] Changed PyOpenCLâs complex numbers from
float2
anddouble2
OpenCL vector types to customstruct
. This was changed because it very easily introduced bugs wherecomplex*complex
real+complex
look like they may do the right thing, but silently do the wrong thing.
Rewrite of the wrapper layer to be based on CFFI
Pypy compatibility
Faster kernel invocation through Python launcher code generation
PoCL compatibility
Version 2015.1¶
Support for new-style buffer protocol
Numerous fixes
Version 2014.1¶
Bug fixes
Version 2013.2¶
Support strides on
pyopencl.enqueue_map_buffer()
andpyopencl.enqueue_map_image()
.pyopencl.ImageFormat
was made comparable and hashable.pyopencl.reduction
supports slicing (contributed by Alex Nitz)Bug fixes
Version 2013.1¶
Vastly improved Prefix Sums (âscanâ).
Add
pyopencl.tools.match_dtype_to_c_struct()
, for better integration of the CL andnumpy
type systems.More/improved Bessel functions. See the source.
Add
PYOPENCL_NO_CACHE
environment variable to aid debugging. (e.g. with AMDâs CPU implementation, see their programming guide)Deprecated
pyopencl.tools.register_dtype()
in favor ofpyopencl.tools.get_or_register_dtype()
.Clean up the
pyopencl.array.Array
constructor interface.Deprecate
pyopencl.array.DefaultAllocator
.Deprecate
pyopencl.tools.CLAllocator
Introduce
pyopencl.tools.DeferredAllocator
,pyopencl.tools.ImmediateAllocator
.Allow arrays whose beginning does not coincide with the beginning of their
pyopencl.array.Array.data
pyopencl.Buffer
. Seepyopencl.array.Array.base_data
andpyopencl.array.Array.offset
. Note that not all functions in PyOpenCL support such arrays just yet. These will fail withpyopencl.array.ArrayHasOffsetError
.Add
pyopencl.array.Array.__getitem__()
andpyopencl.array.Array.__setitem__()
, supporting generic slicing.It is possible to create non-contiguous arrays using this functionality. Most operations (elementwise etc.) will not work on such arrays.
Note also that some operations (specifically, reductions and scans) on sliced arrays that start past the beginning of the original array will fail for now. This will be fixed in a future release.
pyopencl.CommandQueue
may be used as a context manager (in awith
statement)
Note
The addition of pyopencl.array.Array.__getitem__()
has an unintended
consequence due to numpy bug 3375. For instance, this
expression:
numpy.float32(5) * some_pyopencl_array
may take a very long time to execute. This is because numpy
first
builds an object array of (compute-device) scalars (!) before it decides that
thatâs probably not such a bright idea and finally calls
pyopencl.array.Array.__rmul__
.
Note that only left arithmetic operations of pyopencl.array.Array
by numpy
scalars are affected. Pythonâs number types (float
etc.)
are unaffected, as are right multiplications.
If a program that used to run fast suddenly runs extremely slowly, it is likely that this bug is to blame.
Hereâs what you can do:
Version 2012.1¶
Support for complex numbers.
Support for Bessel functions. (experimental)
Numerous fixes.
Version 2011.2¶
IMPORTANT BUGFIX: Kernel caching was broken for all the 2011.1.x releases, with severe consequences on the execution time of
pyopencl.array.Array
operations. Henrik Andresen at a PyOpenCL workshop at DTU first noticed the strange timings.All comparable PyOpenCL objects are now also hashable.
Add
pyopencl.tools.context_dependent_memoize
to the documented functionality.Base
pyopencl.clrandom
on RANLUXCL (https://bitbucket.org/ivarun/ranluxcl>
), add functionality.Add
pyopencl.NannyEvent
objects.Ensure compatibility with OS X Lion.
Add
pyopencl.tools.register_dtype()
to enable scan/reduction on struct types.pyopencl.enqueue_migrate_mem_objects()
was renamedpyopencl.enqueue_migrate_mem_objects_ext
.pyopencl.enqueue_migrate_mem_objects()
now refers to the OpenCL 1.2 function of this name, if available.pyopencl.Device.create_sub_devices()
was renamedpyopencl.Device.create_sub_devices_ext
.pyopencl.Device.create_sub_devices()
now refers to the OpenCL 1.2 function of this name, if available.Alpha support for OpenCL 1.2.
Version 2011.1.2¶
More bug fixes.
Version 2011.1.1¶
Fixes for Python 3 compatibility. (with work by Christoph Gohlke)
Version 2011.1¶
All is_blocking parameters now default to True to avoid crashy-by-default behavior. (suggested by Jan Meinke) In particular, this change affects
pyopencl.enqueue_read_buffer
,pyopencl.enqueue_write_buffer
,pyopencl.enqueue_read_buffer_rect
,pyopencl.enqueue_write_buffer_rect
,pyopencl.enqueue_read_image
,pyopencl.enqueue_write_image
,pyopencl.enqueue_map_buffer
,pyopencl.enqueue_map_image
.Add
pyopencl.reduction
.Add Reductions.
Add
pyopencl.scan
.Deprecate context arguments of
pyopencl.array.to_device()
,pyopencl.array.zeros()
,pyopencl.array.arange()
.Make construction of
pyopencl.array.Array
more flexible (cqa argument.)Add Memory Pools.
Add vector types, see
pyopencl.array.vec
.Add
pyopencl.array.Array.strides
,pyopencl.array.Array.flags
. Allow the creation of arrays in C and Fortran order.Add
pyopencl.enqueue_copy()
. Deprecate all other transfer functions.Add support for numerous extensions, among them device fission.
Add a compiler cache.
Add the âg_times_lâ keyword arg to kernel execution.
Version 0.92¶
Add support for OpenCL 1.1.
Add support for the cl_khr_gl_sharing extension, leading to working GL interoperability.
The call signature of
pyopencl.Kernel.__call__()
changed to emphasize the importance of local_size.Add support for the cl_nv_device_attribute_query extension.
Add
pyopencl.array.Array()
and related functionality.Make build not depend on Boost C++.
Version 0.91.5¶
Add
pyopencl.ImageFormat.channel_count
,pyopencl.ImageFormat.dtype_size
,pyopencl.ImageFormat.itemsize
.Add missing
pyopencl.enqueue_copy_buffer
.Add
pyopencl.enqueue_barrier()
, which was previously missing.
Version 0.91.4¶
A bugfix release. No user-visible changes.
Version 0.91.3¶
All parameters named host_buffer were renamed hostbuf for consistency with the
pyopencl.Buffer
constructor introduced in 0.91. Compatibility code is in place.The
pyopencl.Image
constructor does not need a shape parameter if the given hostbuf has hostbuf.shape.The
pyopencl.Context
constructor can now be called without parameters.
Version 0.91.2¶
pyopencl.Program.build()
now captures build logs and adds them to the exception text.Deprecate
pyopencl.create_context_from_type
in favor of second form ofpyopencl.Context
constructorIntroduce
pyopencl.LocalMemory
.Document kernel invocation and
pyopencl.Kernel.set_arg()
.
Version 0.91.1¶
Fixed a number of bugs, notably involving
pyopencl.Sampler
.pyopencl.Device
,pyopencl.Platform
,pyopencl.Context
now have nicer string representations.Add
pyopencl.Image.shape
. (suggested by David Garcia)
Version 0.91¶
Add a test suite.
Fix numerous
get_info
bugs. (reports by David Garcia and the test suite)Add
pyopencl.addressing_mode.to_string()
and colleagues.The
pitch
arguments topyopencl.create_image_2d
,pyopencl.create_image_3d
,pyopencl.enqueue_read_image
, andpyopencl.enqueue_write_image
are now defaulted to zero. The argument order ofenqueue_{read,write}_image
has changed for this reason.Deprecate
pyopencl.create_image_2d
,pyopencl.create_image_3d
in favor of thepyopencl.Image
constructor.Deprecate
pyopencl.create_program_with_source
,pyopencl.create_program_with_binary
in favor of thepyopencl.Program
constructor.Deprecate
pyopencl.create_buffer
,pyopencl.create_host_buffer
in favor of thepyopencl.Buffer
constructor.pyopencl.Image.get_image_info()
now actually exists.Add
pyopencl.Image.info
.Fix API tracing.
Add constructor arguments to
pyopencl.ImageFormat
. (suggested by David Garcia)
Version 0.90.4¶
Add build fixes for Windows and OS X.
Version 0.90.3¶
Fix a GNU-ism in the C++ code of the wrapper.
Version 0.90.2¶
Fix passing properties to
pyopencl.CommandQueue
. Also fix related documentation.
Version 0.90.1¶
Fix building on the Mac.
Version 0.90¶
Initial release.
License¶
PyOpenCL is licensed to you under the MIT/X Consortium license:
Copyright (c) 2009-13 Andreas Klöckner and Contributors.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the âSoftwareâ), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED âAS ISâ, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
PyOpenCL includes derivatives of parts of the Thrust computing package (in particular the scan implementation). These parts are licensed as follows:
Copyright 2008-2011 NVIDIA Corporation
Licensed under the Apache License, Version 2.0 (the âLicenseâ); you may not use this file except in compliance with the License. You may obtain a copy of the License at
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an âAS ISâ BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
Note
If you use Apache-licensed parts, be aware that these may be incompatible with software licensed exclusively under GPL2. (Most software is licensed as GPL2 or later, in which case this is not an issue.)
PyOpenCL includes parts of the Random123 suite of random number generators:
Copyright 2010-2012, D. E. Shaw Research. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions, and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions, and the following disclaimer in the documentation and/or other materials provided with the distribution.
Neither the name of D. E. Shaw Research nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS âAS ISâ AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Frequently Asked Questions¶
The FAQ is maintained collaboratively on the Wiki FAQ page.
Citing PyOpenCL¶
We are not asking you to gratuitously cite PyOpenCL in work that is otherwise unrelated to software. That said, if you do discuss some of the development aspects of your code and would like to highlight a few of the ideas behind PyOpenCL, feel free to cite this article:
Andreas Klöckner, Nicolas Pinto, Yunsup Lee, Bryan Catanzaro, Paul Ivanov, Ahmed Fasih, PyCUDA and PyOpenCL: A scripting-based approach to GPU run-time code generation, Parallel Computing, Volume 38, Issue 3, March 2012, Pages 157-174.
Hereâs a Bibtex entry for your convenience:
@article{kloeckner_pycuda_2012,
author = {{Kl{\"o}ckner}, Andreas
and {Pinto}, Nicolas
and {Lee}, Yunsup
and {Catanzaro}, B.
and {Ivanov}, Paul
and {Fasih}, Ahmed },
title = "{PyCUDA and PyOpenCL: A Scripting-Based Approach to GPU Run-Time Code Generation}",
journal = "Parallel Computing",
volume = "38",
number = "3",
pages = "157--174",
year = "2012",
issn = "0167-8191",
doi = "10.1016/j.parco.2011.09.001",
}
Acknowledgments¶
Contributors¶
Too many to list. Please see the commit log for detailed acknowledgments.
Funding¶
Work on pytential was supported in part by
the US National Science Foundation under grant numbers DMS-1418961, DMS-1654756, SHF-1911019, and OAC-1931577, and
the Department of Energy, National Nuclear Security Administration, under Award Number DE-NA0003963.
AK also gratefully acknowledges a hardware gift from Nvidia Corporation.
The views and opinions expressed herein do not necessarily reflect those of the funding agencies.
Documentation Cross-References¶
Numpy¶
- class numpy.int8¶
See
numpy.generic
.
- class numpy.int32¶
See
numpy.generic
.
- class numpy.float64¶
See
numpy.generic
.
OpenCL Specification¶
-
type cl_platform_id¶
See the CL specification.
-
type cl_device_id¶
See the CL specification.
-
type cl_context¶
See the CL specification.
-
type cl_command_queue¶
See the CL specification.
-
type cl_mem¶
See the CL specification.
-
type cl_program¶
See the CL specification.
-
type cl_kernel¶
See the CL specification.
-
type cl_sampler¶
See the CL specification.
-
type cl_event¶
See the CL specification.
-
void clCreateCommandQueueWithProperties()¶
See the CL specification.
-
void clCreateSamplerWithProperties()¶
See the CL specification.
-
void clCreatePipe()¶
See the CL specification.
Internal Types¶
- class pyopencl._cl.Platform¶
See
pyopencl.Platform
.
- class pyopencl._cl.Device¶
See
pyopencl.Device
.
- class pyopencl._cl.CommandQueue¶
- class pyopencl._cl.Context¶
See
pyopencl.Context
.
- class pyopencl._cl.Event¶
See
pyopencl.Event
.
- class pyopencl._cl.SVMAllocation¶
- class pyopencl._cl.MemoryMap¶
See
pyopencl.MemoryMap
.
- class pyopencl._cl.Sampler¶
See
pyopencl.Sampler
.
- class pyopencl._cl.Program¶
See
pyopencl.Program
.
- class pyopencl._cl._Program¶
See
pyopencl.Program
.
- class pyopencl._cl.Kernel¶
See
pyopencl.Kernel
.