×

Enhanced distributed computing message passing strategies for FDTD. (English) Zbl 0957.78017

Summary: The finite difference time domain method (FDTD) solves Maxwell’s equations by employing numerically and storage intensive computation to map the electric and magnetic fields within a finite volume as an explicit function of time. Distributed computation, using heterogeneous networks of computers, is a costeffective way of applying FDTD, however interprocessor communication is the rate-limiting step. In particular, in most laboratories there is a surplus of unused, desktop CPU cycles during the evening and night. We have investigated ways of utilizing this resource by examining various communication harnesses for distributed computing and have for the first time in a distributed FDTD EM application, employed data packing mechanisms to alleviate the bottleneck caused by network latency and limited bandwidth availability.

MSC:

78A55 Technical applications of optics and electromagnetic theory
78M20 Finite difference methods applied to problems in optics and electromagnetic theory
78A25 Electromagnetic theory (general)
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Yee, IEEE Trans. Antennas and Propagation AP-14 pp 302– (1966)
[2] Computer Law, Longman, London, p. 87, 1986.
[3] and , Server-client strategies applied to computational electromagnetics, PIERS ’98, Nantes, France, pp. 602-603, July 1998.
[4] Rodohan, Int. J. Numer. Model. 8 pp 283– (1995)
[5] Computational Electrodynamics, The FDTD Method, Artech House, Boston, U.S.A., 1995.
[6] The Data Compression Book, M&T Books, Redwood City, CA, pp. 34-38, 1991.
[7] Sunderam, Parallel Computing 20 pp 531– (1994) · Zbl 0825.68199
[8] The Data Compression Handbook, M&T Books, Redwood City, CA, p. 34ff, 1991.
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.