Efficient stochastic FDTD algorithm with optimised GPU acceleration

Author

Papadimopoulos, A.N. ; Pyrialakos, G.G. ; Kantartzis, N.V. ; Tsiboukis, T.D.

Author_Institution

Dept. of Electr. & Comput. Eng., Aristotle Univ. of Thessaloniki, Thessaloniki, Greece

fYear

2014

fDate

March 31 2014-April 1 2014

Firstpage

1

Lastpage

2

Abstract

A robust implementation of a stochastic finite-difference time-domain (S-FDTD) technique on modern graphics processing units (GPU) is introduced in this paper for complex media with high levels of statistically-variable heterogeneities. The featured method is compared to the multiple-realisation Monte-Carlo FDTD process, while both schemes are developed on an optimal CUDA (compute unified device architecture) programming platform. Therefore, via the GPU´s enhanced parallelised features, a significant speedup is attained with regard to serialised computations as verified by several numerical results.

Keywords

finite difference time-domain analysis; graphics processing units; parallel architectures; stochastic processes; compute unified device architecture; finite-difference time-domain; graphics processing unit; optimal CUDA programming; optimised GPU acceleration; stochastic FDTD algorithm; FDTD methods; GPU computing; Monte-Carlo techniques; statistical electromagnetics; stochastic schemes;

fLanguage

English

Publisher

iet

Conference_Titel

Computation in Electromagnetics (CEM 2014), 9th IET International Conference on

Conference_Location

London

Electronic_ISBN

978-1-84919-817-2

Type

conf

DOI

10.1049/cp.2014.0178

Filename

6826807