Parallel FDTD solver with optimal topology and dynamic balancing

Abstract

Finite-difference time-domain method (FDTD) is widely used for modeling of computational electrodynamics by numerically solving Maxwell’s equations and finding approximate solution at each time step. The FDTD method was originally developed by K. Yee in 1966 and is still improving to fulfill the needs of researchers. Highly parallel Maxwell’s equations solvers based on the FDTD method allow to model sophisticated structures on large grids with acceptable performance and required accuracy. This article describes parallel FDTD solver for different dimensions with comparison method for virtual topologies of computational nodes’ grid, which allows to choose the best virtual topology for target architecture. Developed solver also incorporates dynamic balancing of computations between computational nodes. Measurements for presented algorithms are provided for IBM Blue Gene/P supercomputer. Further directions for optimizations are also discussed.