OpenMP architecture to simulate 2D water oscillation on paraboloid

Abstract

This paper discusses about implementation of shared parallel architecture OpenMP platform on two-dimensional shallow water equations. Finite volume method that is used to approximate these equations is called staggered grid scheme. The performance of numerical implementation is shown for number of grids (Nx, Ny) more than (16, 16). The results using parallel processing OpenMP for two numbers of threads (4 and 8) are shown better than serial processing. Moreover, maximum speedup by using 4 threads achieved around 3.38 times of serial processing, and around 6.05 times by using 8 threads. Then, the maximum efficiency in this simulation achieved around 84.4% by using 4 threads at number of grids (Nx, Ny) = (256, 256). However by using 8 threads, maximum efficiency achieved around 75.6% at number of grids (Nx, Ny) = (256, 256).