In this paper, we explore the use of hybrid genetic algorithm for optimized clustering and distribution of Finite Difference Time Domain (FDTD) computation over a large number of desktop computers and servers. Given a large number of computers, we first attempt to compute an optimal set of clusters. The clustering takes into consideration similarity of machine capability plus the interconnection speed. It considers as well the predicted availability pattern for each computer. Then, for each cluster, we optimize the distribution of FDTD workload over its computers. Hence, the overall optimization procedure optimizes clustered distributed FDTD. We show in this paper how pure as well as hybrid genetic algorithm can effectively be used to perform the optimization. © 2013 ICIC International.