Parallel computing using CUDA

Abstract

The direction of computing evolves from a "centralized processing" on the CPU to "coprocessing" on the CPU and GPU. To implement the new paradigm of computing, NVIDIA has invented a parallel computing architecture CUDA, currently represented in the graphic processors GeForce, ION, Quadro, and Tesla, and provides the necessary framework for software developers.