PARALLEL 3D FINITE-DIFFERENCE TIME-DOMAIN METHOD ON MULTI-GPU SYSTEMS

摘要

Finite-di®erence time-domain (FDTD) is a popular but computational intensive method tonsolve Maxwell's equations for electrical and optical devices simulation. This paper presentsnimplementations of three-dimensional FDTD with convolutional perfect match layer (CPML)nabsorbing boundary conditions on graphics processing unit (GPU). Electromagnetic ¯elds innYee cells are calculated in parallel millions of threads arranged as a grid of blocks with computenuni¯ed device architecture (CUDA) programming model and considerable speedup factors arenobtained versus sequential CPU code. We extend the parallel algorithm to multiple GPUs innorder to solve electrically large structures. Asynchronous memory copy scheme is used in datanexchange procedure to improve the computation e±ciency. We successfully use this techniquento simulate pointwise source radiation and validate the result by comparison to high precisionncomputation, which shows favorable agreements. With four commodity GTX295 graphics cardsnon a single personal computer, more than 4000 million Yee cells can be updated in one second,nwhich is hundreds of times faster than traditional CPU computation.