字符串匹配算法的实现:CPU vs.GPU vs.FPGA

Hands-on Experience with a String Matching: CPU vs. GPU vs. FPGA

针对字符串匹配算法在各平台实现的性能问题,将算法在CPU、GPU及FPGA上做了测试对比。GPU具有计算单元多的特点,使得GPU对计算密集型应用有较大的效率提升;而FPGA具有级强的灵活性、可编程性及大量的逻辑运算单元,在处理字符串匹配时的处理速度快。通过对3种实现方式在Snort规则库下做的分析,其结果表明,FPGA的处理速度最快,相比GPU的处理速度提升了10倍。而CPU的串行处理速度最慢,且FPGA的资源消耗最多,GPU次之,CPU的资源消耗最少,且实现最简单。

A comparison is made of the performance of string matching algorithm on CPU, GPU and FPGA platforms. GPU has more computing unit, greatly enhancing the efficiency for compute-intensive applications; while the programmable FPGA has great flexibility and a large number of logic unit, thus faster string matching. The anal- ysis in the three implementations base on the snort shows that FPGA is the fastest （ 10 times faster than GPU） with CPU as the slowest; and that FPGA has the greatest resources consumption, followed by GPU, while CPU is the least and with simplest realization.