近年来，计算机图形处理器(GI．apllics Processing Unit,GPU)得到了极大地发展，从最初局限于图形渲染的图形卡，发展为如今可编程的并行计算平台。与CPU的串行计算模式不同，GPU是一种高度并行的流处理器，具有更强的浮点计算能力，在物理模拟、信号分析等许多领域人们将高强度的计算任务以适当的方式转化为流数据计算模式，通过编程在GPU中进行加速计算，通常能获得一个数量级的速度提高，这也是当今的热点研究问题之一。-In recent years, computer graphics processor (GI.apllics Processing Unit, GPU) has been greatly developed, from the initial limited to graphics rendering of the graphics card, development of a programmable parallel computing platform today. Serial computing model with the CPU difference, GPU is a highly parallel stream processors, with more floating-point computing power, in the physical simulation, signal analysis and many other areas that people will be high-strength computing tasks in an appropriate way into computing model for the flow of data through the program carried out in the GPU to accelerate the calculation, usually can get an order of magnitude faster, which is one of today' s hot research issues.