NVIDIA after half year public beta- testing declared about the release of the first final version SDK (collection for developers), optimized for calculations on graphic chips , CUDA (Compute unified device architecture).
Let us recall, that CUDA includes the first in industry compiler C for GPU.
This development gives developers tools for solution of new problems in heavy applications, such as the development of products, data analysis, technical calculations and gaming physics.
In the number of supported by CUDA video card we have the geForce 8800, and also the last model of professional solutions series quadro and Tesla.
GPU with CUDA works either as flexible stream processor, where thousands of computational programs, called stream, together solve complex problems or as stream processor in specific applications, for example, for image conclusion , where the stream are not connected together. Applications with CUDA support use GPU for fine-structure, intensive calculations and multi core CPU for complex large-module problems, for example, data control .
The development kit tools for CUDA is at present accessible for developers and researchers through the program NVIDIA for registered developers. There it is possible to download the program code.