基于CUDA平台的遗传算法并行实现研究  

Research on the Parallel Implementation of Genetic Algorithm on CUDA Platform

作　　者：谭彩凤[1] 马安国[1] 邢座程[1] 

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《计算机工程与科学》2009年第A01期68-72,共5页Computer Engineering & Science

基　　金：国家自然科学基金资助项目(60873016);国家863计划资助项目(2009AA01Z102)

摘　　要：CUDA技术方便程序员在GPU上进行通用计算,但并没有提供随机数产生的应用接口。为此,本文提出并实现在CUDA开发平台上并行产生均匀随机数算法,测试证明算法可行。在此基础上优化基本遗传算法,并在GPU上并行实现其所有操作,提高其运行速度和准确度;分析了种群大小和遗传代数对此算法加速比及准确度的影响,并与MAT-LAB工具箱进行比较。实验表明,相比MATLAB遗传算法工具箱,基于CUDA平台实现的遗传算法性能更高,准确度更好。The CUDA technology provides conveniences of general computation for programmers, but there is no application programming interface of generating random number on CUDA. Therefore, this paper presents and implements a method for parallel producing random number algorithm on CUDA, and the methods is proved feasible by testing. On this condition, we implement a parallel implementation of GA on GPU, optimize the efficiency and precision of the standard GA, analyze the influence of population size and generations of evolution to efficiency and accuracy of this algorithm. The experiment shows that compared with GA Toolbox of MATLAB, the performance and the precision of this method is better.

关 键 词：随机数 遗传算法 并行计算 GPGPU CUDA GPU 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]