一种OpenMP并行优化的AoI兴趣过滤算法  

OpenMP based Parallel Optimized AoI Interest Filtering Algorithm

作　　者：张佳斌[1] 张琳[1] 杨硕磊[1] 郝爱民[1] 

机构地区：[1]北京航空航天大学虚拟现实技术与系统国家重点实验室,北京100083

出　　处：《系统仿真学报》2010年第6期1386-1391,共6页Journal of System Simulation

基　　金：国家863计划(2006AA01Z331);国家自然科学基金(60603084)

摘　　要：AoI(area of interest)兴趣过滤根据虚拟环境中数据的相关性进行过滤,它能有效降低网络通信量,提高系统可扩展性。在多人在线游戏中,随着虚拟环境中实体规模的增大,服务器进行兴趣过滤的计算开销会显著增大,影响游戏的可扩展性。针对近年来成本大幅度降低的多核CPU(central processing unit)架构,使用OpenMP编程接口将兴趣过滤计算的负载分配到多个CPU处理单元上进行并行计算,实验结果表明该并行算法在多核CPU架构上可获得较大的性能提升。AoI （area of interest） interest Filtering algorithm is a kind of algorithm based on data relevance which filters data to reduce the communication traffic and enhance system scalability. As the number of entities rising in virtual environment of Multipalyer Online Games, the AoI related data filtering which holds the most of computing cost becomes a bottleneck of the scalability of games. Utilizing features of low-cost multicore CPU architecture, with OpenMP API, the AoI relevance data filtering computing to each processing unit was distributed in order to realize parallelism. Experiment results show that the parallel algorithm gains performance enhancement on multicore CPU architecture.

关 键 词：AOI 虚拟环境 并行计算 OPENMP MOG 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]