摘要
为了达到提高颗粒流体动力学方法 GHM计算效率的目标,分析了GHM模型的主要计算模块,抽取其中的可并行计算模块,基于多核计算机的硬件环境,应用OpenMP多线程并行计算模型,对采用数值积分方法求解颗粒运动方程的部分,实现求解过程的并行计算。最后通过多次实验验证程序的正确性及算法性能。实验结果表明,在Windows 7系统4核8线程处理器的计算机上,并行程序的并行加速比最高达到了2.5,说明OpenMP多核并行技术能较显著地提高GHM方法的计算性能。
In order to improve the computational efficiency of the granule hydrodynamics method (GHM) model, we propose a multi-thread parallel computing model based on OpenMP for multi-core computers. We analyze the primary computational modules and extract the ones that can be paralleled. We then use the OpenMP compiler directive statements and runtime library to realize the parallel computation of the solving process in the GHM model to solve the granules motion equations by using the numerical integration method. Experimental results on Windows platform with 4 cores and 8 threads CPU show that the parallel program's speedup can reach 2.5, which proves that the OpenMP technology can improve the computational efficiency of the GHM significantly, and that it has advantages of easy programming and good portability as well.
出处
《计算机工程与科学》
CSCD
北大核心
2017年第7期1234-1240,共7页
Computer Engineering & Science
基金
国家自然科学基金(11272061)
关键词
OPENMP
流体力学
并行计算
多核处理器
效率
OpenMP
fluid hydrodynamics
parallel computing
multicore processor
efficiency