摘要
针对浅水波方程(shallow-water wave)的差分格式,通过分析数据关系,应用数据划分的策略方法,将参加计算的数据分配到4个进程中并行进行计算,提高了在机群系统中求解大型数值的速度和效率。并给出了网格点数分别是64×64,128×128和256×256的3种问题规模下在4台工作站机群(C luster)系统中浅水波方程并行程序的运行结果。
The shallow-water equations and the computational methods are given and described. The finite difference scheme is used to attain the discrete forms of the shallow-water wave equations. The solution to PDE can achieve the high efficiency by parallel computation methods. By adopting partitioning, the data is divided into 4 nodes. The results demonstrate the high efficiency achievable on workstations cluster when grid number is 64 ×64,128 × 128 or 256×256 respectively.
出处
《西安科技大学学报》
CAS
北大核心
2006年第4期524-528,共5页
Journal of Xi’an University of Science and Technology
关键词
浅水波
差分格式
MPI
机群
数据划分
进程通信
幻象点
shallow-water
finite difference scheme
MPI
workstation cluster
partitioning
synchronization of nodes
ghost point
