Ever Heard of Cuda?

No it's not a misspelling of a certain Caribbean island! A few years back it was discovered that graphics processing chips (GPUs) could be used to offset much of the main CPU's work. GPUs are optimized to do parallel floating-point operations and can do many vector operations yielding many orders of magnitude higher performance than a conventional CPU. Still it wasn't so easy exploiting this and so Nvidia has created CUDA ("Compute Unified Device Architecture"), a technology that lets a programmer use the C programming language to code algorithms for execution on the GPU.

This technology is already in use. There are for example six open source projects on SourceForge using Cuda. Cuda itself is a free download from Nvidia )for Windows, Linux and Mac) which includes the nvcc Compiler, a Profiler, a gdb debugger for the GPU (alpha available in March, 2008), a CUDA runtime driver (now also available in the standard NVIDIA GPU driver) and the CUDA programming manual. Of course you have to have a Nvidia video card for this. You can see a full list of supported cards on the Learn Cuda site in the Nvidia Cuda Zone. The best thing with Cuda is that it's a totally free technology that you can use to create commercial applications. There is an impressive list of showcase applications. Also NVidia have also provided a free online book GPU Gems2 which I've added to the free E-Book library.

• Link to Free E-Books page
• Link to C Tutorials