Performance of a Massively Parallel Higher-Order Method of Moments Code Using Thousands of CPUs

Z. Lin, X. Zhao, Y. Zhang, T. K. Sarkar

Research output: Chapter in Book/Entry/PoemConference contribution

2 Scopus citations

Abstract

The efficiency and scalability of a parallel higher-order method of moments is illustrated using up to 4096 CPU cores on a supercomputer. The scattering problems solved includes the analysis from a full scale airplane. The scattering problem is simulated to demonstrate the efficiency and scalability of the algorithm. Numerical results show that one can achieve above 60% efficiency when the used memory to the total memory ratio is larger than 15%, and the scalability can reach a theoretical value between O(N2) and O(N3), where N is the number of unknowns. Due to its high efficiency and excellent scalability, the algorithm is able to accurately solve large complex electromagnetic radiation and scattering problems including composite and multi-scale structures.

Original languageEnglish (US)
Title of host publicationProceedings of the 2018 20th International Conference on Electromagnetics in Advanced Applications, ICEAA 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages22-25
Number of pages4
ISBN (Electronic)9781538667620
DOIs
StatePublished - Nov 1 2018
Event20th International Conference on Electromagnetics in Advanced Applications, ICEAA 2018 - Cartagena de Indias, Colombia
Duration: Sep 10 2018Sep 14 2018

Publication series

NameProceedings of the 2018 20th International Conference on Electromagnetics in Advanced Applications, ICEAA 2018

Conference

Conference20th International Conference on Electromagnetics in Advanced Applications, ICEAA 2018
Country/TerritoryColombia
CityCartagena de Indias
Period9/10/189/14/18

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Electrical and Electronic Engineering
  • Instrumentation

Fingerprint

Dive into the research topics of 'Performance of a Massively Parallel Higher-Order Method of Moments Code Using Thousands of CPUs'. Together they form a unique fingerprint.

Cite this