Acceleration of FDTD Code Using MATLAB's Parallel Computing Toolbox

The well‐known finite difference time‐domain (FDTD) method has been used for several years by many researchers, but with continued desire to speed up the simulation time. One of the most effective, but not straightforward, procedures is to run the FDTD simulations on graphical processing units (GPUs). However, this requires major changes in the way the method is being coded and good knowledge of the GPU's hardware to achieve significantly reduced simulation time.It is common for many new researchers in the field to use MATLAB programming in its traditional way. Unfortunately, this does not provide efficient FDTD simulation. Therefore, the aim of this chapter is to presents ways to adequately reduce the FDTD simulations time by exploring the parallelization of the FDTD method and the use of parallel computing toolbox in MATLAB. The methods presented here do not require significant changes in coding or learning new programming language. Several examples showing the use of multi‐cores and multi‐GPUs within MATLAB are presented. Speed up factor for a parallel FDTD code reaches 10 times when using multi‐cores, and 70 times when using 2 or 4 GPUs relative to CPU simulations of unparalyzed FDTD code.