Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks

Allison H. Baker; Axel Klawonn; Tzanio Kolev; Martin Lanser; Oliver Rheinbach; Ulrike Meier Yang

doi:10.1007/978-3-319-40528-5_6

Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks

Allison H. Baker, Axel Klawonn, Tzanio Kolev, Martin Lanser, Oliver Rheinbach, Ulrike Meier Yang

Application Scalability and Performance Group

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

23 Scopus citations

Abstract

The parallel performance of several classical AlgebraicMultigrid (AMG) methods applied to linear elasticity problems is investigated. These methods include standard AMG approaches for systems of partial differential equations such as the unknown and hybrid approaches, as well as the more recent globalmatrix (GM) and local neighborhood (LN) approaches, which incorporate rigid body modes (RBMs) into the AMG interpolation operator. Numerical experiments are presented for both two- and three-dimensional elasticity problems on up to 131,072 cores (and 262,144 MPI processes) on the Vulcan supercomputer (LLNL, USA) and up to 262,144 cores (and 524,288 MPI processes) on the JUQUEEN supercomputer (JSC, Jülich, Germany). It is demonstrated that incorporating all RBMs into the interpolation leads generally to faster convergence and improved scalability.

Original language	English
Title of host publication	Software for Exascale Computing - SPPEXA 2013-2015
Editors	Wolfgang E. Nagel, Hans-Joachim Bungartz, Philipp Neumann
Publisher	Springer Verlag
Pages	113-140
Number of pages	28
ISBN (Print)	9783319405261
DOIs	https://doi.org/10.1007/978-3-319-40528-5_6
State	Published - 2016
Event	International Conference on Software for Exascale Computing, SPPEXA 2015 - Munich, Germany Duration: Jan 25 2016 → Jan 27 2016

Publication series

Name	Lecture Notes in Computational Science and Engineering
Volume	113
ISSN (Print)	1439-7358

Conference

Conference	International Conference on Software for Exascale Computing, SPPEXA 2015
Country/Territory	Germany
City	Munich
Period	01/25/16 → 01/27/16

Access to Document

10.1007/978-3-319-40528-5_6

Cite this

Baker, A. H., Klawonn, A., Kolev, T., Lanser, M., Rheinbach, O., & Yang, U. M. (2016). Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks. In W. E. Nagel, H.-J. Bungartz, & P. Neumann (Eds.), Software for Exascale Computing - SPPEXA 2013-2015 (pp. 113-140). (Lecture Notes in Computational Science and Engineering; Vol. 113). Springer Verlag. https://doi.org/10.1007/978-3-319-40528-5_6

@inproceedings{bb204f27c2e545878f8eb3f1b979d203,

title = "Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks",

abstract = "The parallel performance of several classical AlgebraicMultigrid (AMG) methods applied to linear elasticity problems is investigated. These methods include standard AMG approaches for systems of partial differential equations such as the unknown and hybrid approaches, as well as the more recent globalmatrix (GM) and local neighborhood (LN) approaches, which incorporate rigid body modes (RBMs) into the AMG interpolation operator. Numerical experiments are presented for both two- and three-dimensional elasticity problems on up to 131,072 cores (and 262,144 MPI processes) on the Vulcan supercomputer (LLNL, USA) and up to 262,144 cores (and 524,288 MPI processes) on the JUQUEEN supercomputer (JSC, J{\"u}lich, Germany). It is demonstrated that incorporating all RBMs into the interpolation leads generally to faster convergence and improved scalability.",

author = "Baker, \{Allison H.\} and Axel Klawonn and Tzanio Kolev and Martin Lanser and Oliver Rheinbach and Yang, \{Ulrike Meier\}",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2016.; International Conference on Software for Exascale Computing, SPPEXA 2015 ; Conference date: 25-01-2016 Through 27-01-2016",

year = "2016",

doi = "10.1007/978-3-319-40528-5\_6",

language = "English",

isbn = "9783319405261",

series = "Lecture Notes in Computational Science and Engineering",

publisher = "Springer Verlag",

pages = "113--140",

editor = "Nagel, \{Wolfgang E.\} and Hans-Joachim Bungartz and Philipp Neumann",

booktitle = "Software for Exascale Computing - SPPEXA 2013-2015",

address = "Germany",

}

Baker, AH, Klawonn, A, Kolev, T, Lanser, M, Rheinbach, O & Yang, UM 2016, Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks. in WE Nagel, H-J Bungartz & P Neumann (eds), Software for Exascale Computing - SPPEXA 2013-2015. Lecture Notes in Computational Science and Engineering, vol. 113, Springer Verlag, pp. 113-140, International Conference on Software for Exascale Computing, SPPEXA 2015, Munich, Germany, 01/25/16. https://doi.org/10.1007/978-3-319-40528-5_6

Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks. / Baker, Allison H.; Klawonn, Axel; Kolev, Tzanio et al.
Software for Exascale Computing - SPPEXA 2013-2015. ed. / Wolfgang E. Nagel; Hans-Joachim Bungartz; Philipp Neumann. Springer Verlag, 2016. p. 113-140 (Lecture Notes in Computational Science and Engineering; Vol. 113).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks

AU - Baker, Allison H.

AU - Klawonn, Axel

AU - Kolev, Tzanio

AU - Lanser, Martin

AU - Rheinbach, Oliver

AU - Yang, Ulrike Meier

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2016.

PY - 2016

Y1 - 2016

N2 - The parallel performance of several classical AlgebraicMultigrid (AMG) methods applied to linear elasticity problems is investigated. These methods include standard AMG approaches for systems of partial differential equations such as the unknown and hybrid approaches, as well as the more recent globalmatrix (GM) and local neighborhood (LN) approaches, which incorporate rigid body modes (RBMs) into the AMG interpolation operator. Numerical experiments are presented for both two- and three-dimensional elasticity problems on up to 131,072 cores (and 262,144 MPI processes) on the Vulcan supercomputer (LLNL, USA) and up to 262,144 cores (and 524,288 MPI processes) on the JUQUEEN supercomputer (JSC, Jülich, Germany). It is demonstrated that incorporating all RBMs into the interpolation leads generally to faster convergence and improved scalability.

AB - The parallel performance of several classical AlgebraicMultigrid (AMG) methods applied to linear elasticity problems is investigated. These methods include standard AMG approaches for systems of partial differential equations such as the unknown and hybrid approaches, as well as the more recent globalmatrix (GM) and local neighborhood (LN) approaches, which incorporate rigid body modes (RBMs) into the AMG interpolation operator. Numerical experiments are presented for both two- and three-dimensional elasticity problems on up to 131,072 cores (and 262,144 MPI processes) on the Vulcan supercomputer (LLNL, USA) and up to 262,144 cores (and 524,288 MPI processes) on the JUQUEEN supercomputer (JSC, Jülich, Germany). It is demonstrated that incorporating all RBMs into the interpolation leads generally to faster convergence and improved scalability.

UR - https://www.scopus.com/pages/publications/84989863866

U2 - 10.1007/978-3-319-40528-5_6

DO - 10.1007/978-3-319-40528-5_6

M3 - Conference contribution

AN - SCOPUS:84989863866

SN - 9783319405261

T3 - Lecture Notes in Computational Science and Engineering

SP - 113

EP - 140

BT - Software for Exascale Computing - SPPEXA 2013-2015

A2 - Nagel, Wolfgang E.

A2 - Bungartz, Hans-Joachim

A2 - Neumann, Philipp

PB - Springer Verlag

T2 - International Conference on Software for Exascale Computing, SPPEXA 2015

Y2 - 25 January 2016 through 27 January 2016

ER -

Baker AH, Klawonn A, Kolev T, Lanser M, Rheinbach O, Yang UM. Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks. In Nagel WE, Bungartz HJ, Neumann P, editors, Software for Exascale Computing - SPPEXA 2013-2015. Springer Verlag. 2016. p. 113-140. (Lecture Notes in Computational Science and Engineering). doi: 10.1007/978-3-319-40528-5_6

Scalability of classical algebraic multigrid for elasticity to half a million parallel tasks

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this