CUDA is a parallel computing framework developed by graphics card manufacturer NVIDIA . CUDA can be used to implement software that will run on recent NVIDIA graphics cards. They provide drivers, documentation, links to tutorials and links to applications of CUDA in scientific computing.
CUDA Website with links to examples and various applications
Dr. Dobb's Tutorial provides a good introductory tutorial with coded examples of scientific computing using CUDA on GPUs
GPGPU is a news site and forum covering general purpose computation on GPUs
NVIDIA's Education Page
gpucomputing.net is a website dedicated to fostering collaborative and interdisciplinary work on the various disciplines that benefit from GPU computing.
Mike Giles' page on using GPUs for Computational Finance
AccelerEyes have an engine called Jacket that allows MATLAB code to run on CUDA-capable NVIDIA cards
R+GPU is a site which reports R functions migrated to run on CUDA-capable GPUs
OpenCL is an emerging framework which provides a uniform programming environment for developers that enables them to write portable code for a variety of parallel devices, including GPUs and CPUs. Industry partners include AMD, IBM, Intel and NVIDIA.
Main OpenCL Page
NVIDIA's OpenCL Page
AMD's OpenCL Page
Many-core technology is developing rapidly with respect to hardware. Therefore, it is too difficult to maintain a comprehensive list of important links. However, one can expect this page to keep abreast of major developments.