The AMD JavaLabs team announced the open source release of Aparapi project for Java-OpenCL

"Aparapi is an API for expressing data parallel workloads in Java and a runtime component capable of converting the Java bytecode of compatible workloads into OpenCL™ so that it can be executed on a variety of GPU devices.

Originally an AMD product, Aparapi was released to open source on September 14, 2011. ReadGary Frost’s blog to learn more.

It seems similar with Stanford PPL's Delite,https://github.com/stanford-ppl/Delite(I heard EPFL joined this) Might they collaborate.

Re: The AMD JavaLabs team announced the open source release of A

Added by monster on 2011-11-09, 11:57

How does that compare to JavaCL and ScalaCL? Is it truly hardware-
independent, since it comes from one specific hardware maker?

Re: The AMD JavaLabs team announced the open source release of A

Added by Gary Frost on 2011-11-12, 23:07

monster googlemail.com> writes:

>
> How does that compare to JavaCL and ScalaCL? Is it truly hardware-
> independent, since it comes from one specific hardware maker?
>
>

Aparapi supports any OpenCL 1.1 compatible runtime/device. The initial version
only supported AMD devices, the first patch added MacOS support and relaxed the
restriction. It will work with AMD, NVidia and Intel OpenCL runtimes.

Full disclosure, I am the Aparapi developer.

Re: Re: The AMD JavaLabs team announced the open source release

Added by Solomon Sun on 2011-11-14, 17:17
Well start,here is another excellent technology to utilize the GPGPU for Java:http://www.ateji.com/px/java-on-gpu.html

2011/11/13 Gary Frost <frost [dot] gary [at] gmail [dot] com>
monster <skunkiferous <at> googlemail.com> writes:

>
> How does that compare to JavaCL and ScalaCL? Is it truly hardware-
> independent, since it comes from one specific hardware maker?
>
>

Aparapi supports any OpenCL 1.1 compatible runtime/device.  The initial version
only supported AMD devices, the first patch added MacOS support and relaxed the
restriction.  It will work with AMD, NVidia and Intel OpenCL runtimes.

Full disclosure, I am the Aparapi developer.





--
SolomonHUAWEIGoogle+: Qihui Sun


Re: The AMD JavaLabs team announced the open source release of A

Added by DaveScala on 2011-11-14, 23:37

After installing the latest gpu driver with OpenCL 1.1 support
( http://developer.nvidia.com/opencl ）the aparapi examples detected my
gpu.

Would be nice if it somehow could work with Scala parallel
collections.
I have now 2 core cpu and 8 CUDA cores gpu.
Then I would have 10 logical processors :D

Copyright © 2013 École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland