NVIDIA company expends many resources on the advance of its CUDA platform, which implies the acceleration of calculations by graphic chips. Partner agreements with the software developers consist on competitions among programmers . Special attention is paid to the possibilities of CUDA, supers-computer take this technology for armament. AMD Company in this sense is less active. NVIDIA Company, also supports OpenCL and DirectCompute;.
AMD representatives acknowledged in an interview that up to 2012 the company actively moves graphic calculation accelerators (GPGPU) in the segment of mass server solutions. In 2012 AMD will already release processors with 16 cores, but a number of loads are better processed by the video card; therefore, AMD will try to harmoniously develop server products.
Working with video and speeches, pattern recognition and data analysis - here only several spheres of GPGPU application. A number of loads will be as before processed rapidly by central processors. The servers in the nearest future will obtain a different structure, which is not rested exclusively on the possibility of central processors.