首页> 外文会议>International Conference on Virtual Reality and Visualization >How Parallelization Helps Crowd Simulation: Study of an OpenMP-Based System
【24h】

How Parallelization Helps Crowd Simulation: Study of an OpenMP-Based System

机译:平行化如何有助于人群模拟:基于OpenMP的系统研究

获取原文

摘要

This paper analyzes the parallelization efficiency of Menge [1], an open source virtual crowd simulation system widely used for algorithm benchmarking, with focuses on three aspects: performance of the existing parallel processing scheme, bottleneck of parallel processing, and improvement opportunities for parallel efficiency of the system. First, we calculate the speedup ratio of each Menge module by analyzing the data collected under with and without OpenMP scenarios. We then identify the bottleneck of the parallel computing from the empirical study. Secondly, the possibility of improving the performance through hardware configuration is analyzed by testing the performance of the system on different computers which have the similar clock frequencies but different number of cores. The experimental results show that there is still plenty of room for improvement in the parallelization performance of the system.
机译:本文分析了Menge [1]的平行化效率,广泛用于算法基准的开源虚拟人群仿真系统,专注于三个方面:现有的并行处理方案的性能,并行处理的瓶颈,以及平行效率的改进机会系统。首先,我们通过分析在没有OpenMP场景下收集的数据来计算每个Menge模块的加速比。然后,我们从实证研究中识别并行计算的瓶颈。其次,通过在具有类似时钟频率但不同数量的核心的不同计算机上测试系统的性能来分析通过硬件配置改善性能的可能性。实验结果表明,仍然有足够的空间改善了系统的并行化性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号