Hybrid MPI/OpenMP Parallelization in FETI-DP Methods

Beitrag in einem Sammelwerk


Details zur Publikation

Autor(en): Klawonn A, Lanser M, Rheinbach O, Stengel H, Wellein G
Titel Sammelwerk: Recent Trends in Computational Engineering - CE2014
Verlag: Springer Link
Verlagsort: -
Jahr der Veröffentlichung: 2015
Titel der Reihe: Lecture Notes in Computational Science and Engineering
Band: 105
Seitenbereich: 67-84
ISBN: 978-3-319-22997-3
ISSN: 1439-7358


Abstract


We present an approach to hybrid MPI/OpenMP parallelization in FETIDP methods using OpenMP with PETScCMPI in the finite element assembly and using the shared memory parallel direct solver Pardiso in the FETI-DP solution phase. Our approach thus uses OpenMP parallelization on subdomains and MPI in between subdomains. We investigate the efficiency of this approach for a benchmark problem from two dimensional nonlinear hyperelasticity. We observe good scalability for up to four threads for each MPI rank on a state-of-the-art Ivy Bridge architecture and incremental improvements for up to ten OpenMP threads for each MPI rank.



FAU-Autoren / FAU-Herausgeber

Stengel, Holger
Hauptbibliothek
Wellein, Gerhard Prof. Dr.
Professur für Höchstleistungsrechnen


Autor(en) der externen Einrichtung(en)
Technische Universität Bergakademie Freiberg
Universität Köln


Zitierweisen

APA:
Klawonn, A., Lanser, M., Rheinbach, O., Stengel, H., & Wellein, G. (2015). Hybrid MPI/OpenMP Parallelization in FETI-DP Methods. In Recent Trends in Computational Engineering - CE2014 (pp. 67-84). -: Springer Link.

MLA:
Klawonn, Axel, et al. "Hybrid MPI/OpenMP Parallelization in FETI-DP Methods." Recent Trends in Computational Engineering - CE2014 -: Springer Link, 2015. 67-84.

BibTeX: 

Zuletzt aktualisiert 2018-10-08 um 22:24