摘要
BF算法是串匹配算法经典算法之一,但并不适合GPU这种并行体系结构。提出了基于统一计算设备架构(CUDA)架构的解决方案,通过对需要处理的数据增加一定比例的冗余信息,设计了适合CUDA计算数据的独立性特点的并行BF算法。实验结果表明,基于CUDA架构的并行串匹配算法比同等CPU算法获得约10倍的加速比。此外还对该算法性能的影响因子做了分析。
BF algorithm is one of classical string match algorithms, but it is not suited to Graphic Processor Unit (GPU) with is parallel architecture. The authors designed a parallel BF algorithm conforming to the profiles of data processing in Compute Unified Device Architecture (CUDA) by adding some extent redundancy to the data to be matched. Experiment shows the proposed solution is about 10 times faster than equivalent CPU algorithm. Furthermore, some factors that will infect string matching performance were analyzed.
出处
《计算机应用》
CSCD
北大核心
2009年第B06期399-401,共3页
journal of Computer Applications
基金
国家973计划项目(2007CB311100)
关键词
统一计算设备架构
单指令多线程
并行
串匹配算法
Compute Unified Device Architecture (CUDA)
single instruction multiple threads
parallel
string matching algorithm