Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

https://github.com/flyx/OpenCLAda

An Ada binding for the OpenCL host API
https://github.com/flyx/OpenCLAda

ada ada-binding cl-gl opencl

Last synced: 2 months ago
JSON representation

An Ada binding for the OpenCL host API

Lists

README 

        
# OpenCLAda - An Ada binding for the OpenCL host API



## About

This is OpenCLAda, a thick Ada binding for the OpenCL host API.

This binding enables you to write OpenCL hosts in Ada. It does **not**

enable you to write OpenCL kernels in Ada.



## Prerequisites



OpenCLAda currently supports MacOSX, Linux and Windows. You need to have



 - a GNAT compiler

 - an OpenCL implementation

 - optionally OpenGLAda (if you want to use the cl_gl extension)



available on your system. GNAT GPL Edition is available at

[the AdaCore website][7]. OpenCL is

usually available from your hardware vendor. On MacOSX, it's already part of

the operating system. On Windows, you'll need an `OpenCL.lib` file to link

against. This is usually not part of the OpenCL implementation, but can be

acquired as part of an SDK from your hardware vendor (eg the

[AMD APP SDK][6]).



[OpenGLAda][8] is required for OpenCL's

cl_gl extension. Just download its source and make sure the path to `opengl.gpr`

is included in the `ADA_PROJECT_PATH` environment variable. To compile the tests,

you also need the [GLFW library][9] version 2, which is used

for window construction in the cl_gl tests.



## Compilation



On MacOSX and Linux, open a terminal, navigate to the OpenCLAda directory and do:



	$ make



On Windows, it could work the same way if you're using MinGW or Cygwin.

However, I didn't try either one. Anyway, to compile without make, just do



	$ gnatmake -p -Popencl-cl_gl.gpr -XWindowing_System=windows



*Note: The variable __Windowing_System__ is shared with OpenGLAda. You have to

provide it even when compiling without OpenGL support because it defines the way

OpenCLAda links with your system libraries.*



On Windows, the compiler needs to find the `OpenCL.lib` file mentioned above. If

you're unsure how to achieve this, just copy it into `C:\GNAT\[version]\lib` or

wherever you installed your GNAT compiler.



*Note: The availability of an OpenCL implementation will not be tested when

building OpenCLAda. So if you want to make sure that OpenCL is available,

build the tests and see if they are linked properly (see below).*



If you want to build OpenCLAda without the cl_gl extension, do



   $ gprbuild -p -P opencl.gpr -XWindowing_System={windows|quartz|x11}



*Note: The makefile does not support switching off cl_gl, because I'm lazy.*



## Installation



OpenCLAda is just a wrapper library and does not include an installation routine.

You can just add it to your project.



## Tests



OpenCLAda comes with some tests (or rather, examples). I wrote them to test

some of the basic functionality of the API. You can build them with



	$ make tests



or



	$ gnatmake -p -P opencl_test.gpr -XGL_Backend={windows|quartz|x11}

	

A basic "hello world" example is also included. After compilation,

the executables will be located in the `bin` directory. They can only be

executed in the `bin` directory, as they load some OpenCL kernel files through

relative paths.



## Usage



There is some

[overview over the OpenCLAda API on the Wiki][1], which is outdated and will

shortly replaced by a new documentation section on the [homepage][2].

For more information, please consult the [Khronos OpenCL API Registry][3].



## Contributing



You're welcome to contribute code or file bug reports on the

[project's page on GitHub][4].



## License



This code is distributed under the terms of the [ISC License][5], which you can

find in the file `COPYING`.



 [1]: https://github.com/flyx/OpenCLAda/wiki/Overview

 [2]: http://flyx.github.io/OpenCLAda

 [3]: http://www.khronos.org/registry/cl/

 [4]: https://github.com/flyx/OpenCLAda

 [5]: http://opensource.org/licenses/ISC

 [6]: http://developer.amd.com/tools-and-sdks/heterogeneous-computing/amd-accelerated-parallel-processing-app-sdk/

 [7]: http://libre.adacore.com/libre/download/

 [8]: https://flyx.github.io/OpenGLAda

 [9]: http://www.glfw.org/