Parallel computing in sound field analysis by bread slicing method

摘要

Numerical analyses using the explicit scheme of the finite difference method greatly reduce the number of known variables that are needed to solve an unknown variable in comparison with implicit methods. This fact indicates that the amount of communication decreases among the nodes of a computer cluster on parallel computing Generally, on the parallel computing of the distributed memory system, the communicating time occupies most of the total calculating time. Therefore, a reduction in the amount of communication directly causes a reduction in calculating time The parallel computing using the explicit scheme of the finite difference method enables the building up of high-performance algorithms. To reduce the amount of communication among nodes, we have proposed a method of dividing the field of numerical analyses. It is called the "bread slicing method" With this method, the field must be divided into pieces parallel to the section that has the smallest area. Results showed that the method in this study enables for the total amount of communication in the computer cluster to be reduced and the geometries of communication to be made simpler.