基于GPU的社会化网络关系计算研究

随着网络不断地社会化、普及化，网络社区的规模也越来越大，这给社会化网络关系的计算带来了巨大的计算量。这些计算包括个人关系及计算与生成、全局关系计算与生成以及关系的挖掘等。虽然这些工作的计算量很大，但却适合并行处理。基于此，本文通过详细分析GPU的高性能运算及其在CUDA编程模型上的具体实现，讨论利用基于CUDA硬件架构的GPU来进行社区用户关系的并行计算。

Computing Reasearch of Socialization Networking Relationship on GPU

With the network being constantly socialized and popular, The scale of the online community is becoming more and more strong,which leads to a large amount of computation to the calculation of social networks relationships.These calculations include personal relationship calculation and generation,global relations calculation and generation and the mining of the relationship,etc.Although the amount of calculation of the work is very large,it is suitable for parallel processing.Based on this,by analysising the high performance computing of GPU and concrete realization on CUDA programming model detailly, the paper discusses the use of GPU based on the hardware architecture of CUDA to the parallel computing of the community customers＇ relationships.