Implementation & Parallelisation of FDTD code for Electromagnetic Scattering
Department of Aerospace Engineering, Indian Institute of Technology, Bombay, Mumbai – 400076
Indian Institute of Technology, 2014
@phdthesis{ravindra2014implementation,
title={Implementation & Parallelisation of FDTD code for Electromagnetic Scattering},
author={Ravindra, Deshpande Varadendra},
year={2014},
school={Indian Institute of Technology, Bombay Mumbai}
}
The present report deals with the application of the algorithm for computation of electromagnetic field components using FDTD method developed by Kane Yee, to Cartesian meshes using total field formulation. For this purpose, code has been written for electromagnetic scattering computation in C language. For generation of code, some snippets from [1] have been used. Various canonical geometries are studied as PECs and also dielectric materials. The serial code has then been parallelized using OpenMP, MPI and GPU (CUDA) and performance has been compared for various mesh sizes and core/thread availability. It was observed that while OpenMPI and MPI provide a linear scale-up, use of GPU (CUDA) gave a scale-up of about 60x. Also, NTFF transformation, RCS computation are studied in brief for future work.
July 18, 2014 by hgpu
