基于MPI的弹道仿真的并行计算研究  

The Parallel Computing Research of Trajectory Simulation Based on MPI

作　　者：钟新玉 王恂[1] 张学军[1] 

机构地区：[1]北京控制与电子技术研究所,北京100038

出　　处：《航天控制》2015年第3期63-67,共5页Aerospace Control

摘　　要：介绍了MPI的基本概念,列举了提高并行计算效率的一些措施,并实现了基于MPI消息传递机制的弹道仿真计算模型的并行编程,最后对并行计算和串行计算的结果进行了比较。研究结果表明,将并行计算技术应用于弹道仿真中是可行的,并获得了较高的加速比和计算效率;与单机运行相比,计算机机群可以大大缩短计算时间,提高计算效率,而且可以扩大计算规模,系统适应性和延展性较好。The basic concept of MPI is introduced, and some measures are listed to improve the efficiency of parallel computing, and parallel programming of trajectory simulation model based on MPI message - passing mechanism is implemented in this paper. Finally, the results of parallel and serial calculation are compared. The research results show that the parallel computing technology applied to the trajectory simula- tion is feasible, and high speed - up ratio and calculation efficiency are achieved. By comparing with the single-unit operation, the computer cluster operation not only can greatly reduce the calculation time and improve computational efficiency, but also can expand the scale of calculation. Thus, the adaptability and ductility of system are better.

关 键 词：弹道仿真 并行计算MPI 计算机机群 加速比 并行计算效率 

分 类 号：V525[航空宇航科学与技术—人机与环境工程]