Analysis of parallel performance of MPI based parallel FDTD on supercomputer

摘要

With the development of multi-core processors, computer clusters and the application of parallel MPI technique, the study of electromagnetic fields numerical arithmetic can improve the capacity of solving large problems. In this paper, based on the platform from National Supercomputing Center in Tianjin (NSCC-TJ: Ranked first in the world in 2010.11), some tests about the parallel efficiency of the parallel FDTD code are made to provide a general rule for a best performance of FDTD on multi-core computers. The results show that following this rule, we can achieve the highest computational efficiency on a more advanced platform.