Design of asynchronous and scalable MLFMA implementations

Design of asynchronous and scalable MLFMA implementations

Over the past years, we have developed several parallel implementations of the MLFMA in two and three dimensions, for the broadband EM characterization of large structures. We have reviewed how the hierarchical partitioning can result into scalable algorithms. Asynchronous techniques allow for the s...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Fostier, J, Michiels, B, Peeters, J, Bogaert, I, Cools, K, De Zutter, D
Format: Tagungsbericht
Sprache:eng
Schlagworte:
Dielectrics
MLFMA
Multicore processing
Parallel algorithms
Partitioning algorithms
Scattering
Three dimensional displays
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Over the past years, we have developed several parallel implementations of the MLFMA in two and three dimensions, for the broadband EM characterization of large structures. We have reviewed how the hierarchical partitioning can result into scalable algorithms. Asynchronous techniques allow for the simulation of problems with a large number of dielectric objects. Experiments indicate that large scale simulations of over 512 cores are now possible.
ISSN:1522-3965
1947-1491
DOI:10.1109/APS.2010.5561771