NVIDIA company counted on its own program platform CUDA for the acceleration of calculations using graphic chips , whereas AMD will support the standard openCL and the graphic API directX future versions.
In AMD opinion , the conventional platforms openCL will extrude CUDA from this actively developing market segment .
This week khronos group published the openCL 1.0 specifications- the open free cross- platform standard for processors programming with the high ability of parallel load execution .
We note that OpenCL allow to program not only the graphic chips, but also central processors, digital signal processors, or multi-core chips like Cell.
Actually, the applications developed under OpenCL can use immediately several different processors and chips, which are located in one system.
AMD company, for example, intends to use this possibility within fusion before the physical integration of graphic cores and central processors. AMD plans to release the SDK stream version with openCL 1.0 support in first half of 2009.
NVIDIA company also stated about the support Of openCL 1.0 - all video chips with CUDA support can work, also, with this standard. In this case we deal with video cards Geforce 8xxx and above .
The support Of openCL at the operating system level will be injected by apple company with the release Of mac OS X 10.6 (Snow leopard).
NVIDIA representatives actively greet this step . It is possible to assume that and Intel company will use OpenCL for the realization of computational potential for its graphic products larrabee generation .