AMD and nVidia support OpenCL 1.0

On the day of the official launch of the first Open Computing Language (OpenCL) version by the Khronos Group, both AMD and nVidia announce that their graphics cards will support the Application Programming Interface (API).  The cross-platform API is developed to easily adress the calculative power of Cpus and GPUs without having to worry about the chip's brand. 

AMD reports that the company is working on the implementation of the OpenCL 1.0 specification in its ATI Stream SDK. It's expected that the  first Stream SDK with OpenCL-support for developers will be released in the first half of 2009. At the moment, AMD uses the open source Brook+ programming language for parallel operation of its GPUs. In version 1.4 of the ATI Stream SDK, which is planned for the first quarter of 2009, the company will offer the option to convert Brook+ programming to OpenCL, increase support for its topmodel Radeon HD 4870 X2 and add compatibility with the FirePro graphics cards.

nVidia is also on the move and has announced full support for the OpenCL 1.0 specification. The company claims that CUDA, its architecture for parallel computing on GPUs, is developed to be compatible with multiple programming languages, such as C, JAVA, Python and also OpenCL.

Both companies declare that the OpenCL 1.0 specification is an important step towards the full adaptation of the GPU calculative power. It's expected the OpenCL will also be taken into account with the development of future software.

The full specification of OpenCL 1.0 can be downloaded from the Khronos Group website, here.  

Tags: cuda gpgpu opencl hd4870 radeon

Post to Slashdot Digg this Add to del.icio.us