摘要
为解决Matlab环境下使用图形处理器(GPU)设备进行计算问题,基于对硬件及软件环境进行的实验,给出了对策的实现方法及步骤,主要包括操作系统、GPU设备及驱动程序、使用GPU进行数据并行计算、编译超时、程序设计上的限制、同时用几个GPU设备等问题。主结论有,Matlab环境下使用GPU进行数据并行仿真计算的实现方式简单,经济可行,可以大幅提高计算速度。
Based on experiments of software and hardware, this paper presents several problems and solutions for Graphic Processing Unit （GPU） parallel computing in Matlab, mainly including operating system, GPU device and their driver, data parallel computing using GPU, compile timeout, restricts of programming, synchronously using several GPU devices, etc. It is concluded that the data parallel computing method using GPU in Matlab is simple and cost-efficient, and it can significantly improve the computing speed.
出处
《计算机与网络》
2013年第11期51-54,共4页
Computer & Network
