A novel scheme for high performance Finite-Difference Time-Domain (FDTD) computations based on GPU

Abstract

Finite-Difference Time-Domain (FDTD) has been proved to be a very useful computational electromagnetic algorithm. However, the scheme based on traditional general purpose processors can be computationally prohibitive and require thousands of CPU hours, which hinders the large-scale application of FDTD. With rapid progress on GPU hardware capability and its programmability, we propose in this paper a novel scheme in which GPU is applied to accelerate three-dimensional FDTD with UPML absorbing boundary conditions. This GPU-based scheme can reduce the computation time significantly, while obtaining high accuracy as compared with the CPU-based scheme. With only one AMD ATI HD4850 GPU, when computational domain is up to (180×180×180), our implementation of the GPU-based FDTD performs approximately 93 times faster than the one running with Intel E2180 dual cores CPU. © Springer-Verlag Berlin Heidelberg 2010.