NVidia releases new compiler and OpenGL specifications

NVidia has just released CUDA, a new special compiler that has C extensions to program the GPU. This allows a high-level direct programming of the GPU, using it almost as a coprocessor. The approach is different from ATI's CTM (which provides a low-level assembly API and leaves to the developers the development of the libraries).

They have also updated their OpenGL extension specifications to support DirectX 10 capabilities, such as the geometry shaders.

You can read more info about CUDA and CTM at ars technica and at NVidia.


Posted by Kartones on 2007-02-19