«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

期数:

2013年02期

页码:

115-118

栏目:

智能、算法、系统工程

出版日期:

1900-01-01

文章信息/Info

Title:

Research on Pseudo-random Number Generator Based on CUDA Platform

文章编号:

1673-629X（2013）02-0115-04

作者:

郭海凤

[1]金陵科技学院 信息技术学院;[2]江苏省信息分析工程实验室

Author(s):

GUO Hai-feng

关键词:

伪随机数; 线性同余; 并行计算; 加速比

Keywords:

pseudo-random number; linear congruence; parallel computing; accelerator

文献标志码:

A

摘要:

针对线性同余算法在产生海量随机数序列时,速度较慢的情况,文中提出并实现了一种产生伪随机数的改进方法,即基于CUDA平台的并行线性同余法,测试证明改进算法可行.该方法通过利用GPU的并行架构,实现并行产生伪随机数.通过实验结果比较,与传统的线性同余算法相比,改进算法产生的随机序列周期较长、速度较快,当其产生大量伪随机数序列时其加速比接近100,解决了传统线性同余算法产生伪随机数周期短、速度较慢的问题,系统仿真效果较优

Abstract:

For linear congruence algorithm (LCA) has low speed of producing huge amounts of random number sequences,propose a parallel pseudo-random number generator based on CUDA platform. Test proved the improved algorithm is feasible. The method by using GPU parallel architecture generates parallel pseudo random number. Compared with traditional linear congruence algorithm,improved al-gorithm has longer cycles and faster speed of generating random sequence,its acceleration ratio is close to 100,solves the slow speed of traditional algorithm. And its system simulation shows it is better than the traditional one

常用功能

更新日期/Last Update: 1900-01-01