Parallel Computation Applied to Electromagnetic Scattering and Radiation Analysis
We have been applying the computational power of parallel processing to the solution of large-scale electromagnetic scattering and radiation problems. Several analysis codes have been implemented on the Jet Propulsion Laboratory/California Institute of Technology Mark IIIfp Hypercubes. The first cod...
Gespeichert in:
Veröffentlicht in: | Electromagnetics 1990-01, Vol.10 (1-2), p.21-39 |
---|---|
Hauptverfasser: | , , , , , , , , |
Format: | Artikel |
Sprache: | eng |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | We have been applying the computational power of parallel processing to the solution of large-scale electromagnetic scattering and radiation problems. Several analysis codes have been implemented on the Jet Propulsion Laboratory/California Institute of Technology Mark IIIfp Hypercubes. The first code to be implemented was the Numerical Electromagnetics Code (NEC-2) from Lawrence Livermore National Laboratory. At first we simply ported it to run in the parallel processing environment. Since that time, taking advantage of the large hypercube memory and fast computation. we have enhanced parallel NEC to permit iterative design and analysis. Three other codes, frequency domain finite elements, time domain finite difference, and frequency selective surfaces, have been largely or completely developed within this parallel processing environment. Because of the massive problem size of the typical electromagnetics problem, our work is an important influence in determining the development of hardware, system software, and user-friendly tools for JPL/Caltech parallel processor systems. This paper will summarize our experience in utilizing parallel processing for electromagnetics applications. Included is a discussion of techniques for implementing parallel codes, means of measuring performance, attributes of the parallel processing system which are particularly important for EM solutions, the issue of portability, and directions for further development of parallel processing as it enters the mainstream of EM computing. |
---|---|
ISSN: | 0272-6343 1532-527X |
DOI: | 10.1080/02726349008908227 |