Parallel FDTD Simulation Using Numa Acceleration Technique

In this paper, we introduce a new non-uniform memory access (NUMA) acceleration algorithm for parallel finite-difference time-domain (FDTD) method on NUMA architecture workstation. We compare the performance of parallel FDTD method with and without the NUMA acceleration technique. An ideal test case and an engineering example show that the NUMA acceleration technique can efficiently improve the computing performance of FDTD parallel applications.