Record link: 
http://hdl.handle.net/20.500.12708/152596
-
Title: 
Collaboration with the SME TAILSIT during the SHAPE project PARTS: Electromagnetic simulations with the finite/boundary element method for large systems using HPC
en
Citation: 
Steiner, S. (2022, May 31). Collaboration with the SME TAILSIT during the SHAPE project PARTS: Electromagnetic simulations with the finite/boundary element method for large systems using HPC [Conference Presentation]. Austrian-Slovenian HPC Meeting 2022 - ASHPC22, Grundlsee, Austria, Austria.
-
Publication Type: 
Presentation - Conference Presentation
en
Language: 
English
-
Authors: 
Steiner, Soner 
-
Organisational Unit: 
E057-09 - Fachbereich VSC Research Center 
-
Date (published): 
31-May-2022
-
Event name: 
Austrian-Slovenian HPC Meeting 2022 - ASHPC22 
-
Event date: 
31-May-2022 - 2-Jun-2022
-
Event place: 
Grundlsee, Austria, Austria
-
Keywords: 
High Performance Computing; Electromagnetic Simulations; Parallel Computing; Finite element method (FEM); Boundary Element Method
en
Abstract: 
TAILSIT is a company based in Styria, Austria, that produces custom-fit simulation software tools for electromagnetic problems and structural analysis. During this SHAPE project TAILSIT’s electromagnetic simulation software has been ported to HPC machines with the support from an expert from the VSC Research Center at TU Wien and using the computing resources of the Vienna Scientific Cluster (VSC, https://vsc.ac.at). A careful performance analysis revealed several bottlenecks and limitations that were addressed by implementing optimised communication strategies with the Message Passing Interface (MPI). While the previous version of TAILSIT’s software simulation tools relying on shared-memory parallelism was restricted to typically less than one million surface degrees of freedom, the new and optimised version employing distributed-memory parallelisation with MPI allows to treat problems up to 50 x 10^9 surface degrees of freedom. In addition, access to the HPC resources allowed to test and further optimise the shared-memory version of the fast multipole method FMM library.
en
Research facilities: 
Vienna Scientific Cluster 
-
Link (external): 
https://www.ashpc.at/ashpc22/
-
Research Areas: 
Mathematical and Algorithmic Foundations: 50%
Modeling and Simulation: 50%
-
Science Branch: 
1020 - Informatik: 50%
1030 - Physik, Astronomie: 30%
1010 - Mathematik: 20%
-
Appears in Collections:
Presentation

Thumbnail
(Graphical) Abstract
Adobe PDF
(672.65 kB)
ASHPC22 – Austrian-Slovenian HPC Meeting 2022, Abstract von Soner Steiner
Open Access CC BY-NC-ND 4.0
Show full item record

Page view(s)

107
checked on Nov 20, 2023

Download(s)

29
checked on Nov 20, 2023

Google ScholarTM

Check