摘要
从单个GPU异构并行系统来看,其性能还是比较有限,文章阐述了基于GPU集群的CUDA架构实现过程,详细分析基于GPU集群进行MD5快速破解的编译过程,并对破解程序进行测试研究,对其运行结果进行测试和分析,探讨搭建GPU高性能计算集群及其进行MD5算法的快速破解的过程。
View from a single GPU heterogeneous parallel system, its performance is still relatively limited, the paper described the CUDA architecture based on GPU cluster implementation process, analyzed the process of MD5 fast crack compilation detailed based on GPU cluster MD5, and testing research of the crack program, and analysis of its operating results, explored the structures of GPU high-performance computing cluster and its rapid crack MD5 algorithm.
出处
《电脑与信息技术》
2013年第2期54-56,共3页
Computer and Information Technology