Performance benchmarks for a next generation numerical dynamo model

Hiroaki Matsui, Eric Heien, Julien Aubert, Jonathan M. Aurnou, Margaret Avery, Ben Brown, Bruce A. Buffett, Friedrich Busse, Ulrich R. Christensen, Christopher J. Davies, Nicholas Featherstone, Thomas Gastine, Gary A. Glatzmaier, David Gubbins, Jean Luc Guermond, Yoshi Yuki Hayashi, Rainer Hollerbach, Lorraine J. Hwang, Andrew Jackson, Chris A. JonesWeiyuan Jiang, Louise H. Kellogg, Weijia Kuang, Maylis Landeau, Philippe Marti, Peter Olson, Adolfo Ribeiro, Youhei Sasaki, Nathanaël Schaeffer, Radostin D. Simitev, Andrey Sheyko, Luis Silva, Sabine Stanley, Futoshi Takahashi, Shin Ichi Takehiro, Johannes Wicht, Ashley P. Willis

Research output: Contribution to journalArticle

27 Citations (Scopus)

Abstract

Numerical simulations of the geodynamo have successfully represented many observable characteristics of the geomagnetic field, yielding insight into the fundamental processes that generate magnetic fields in the Earth's core. Because of limited spatial resolution, however, the diffusivities in numerical dynamo models are much larger than those in the Earth's core, and consequently, questions remain about how realistic these models are. The typical strategy used to address this issue has been to continue to increase the resolution of these quasi-laminar models with increasing computational resources, thus pushing them toward more realistic parameter regimes. We assess which methods are most promising for the next generation of supercomputers, which will offer access to O(106) processor cores for large problems. Here we report performance and accuracy benchmarks from 15 dynamo codes that employ a range of numerical and parallelization methods. Computational performance is assessed on the basis of weak and strong scaling behavior up to 16,384 processor cores. Extrapolations of our weak-scaling results indicate that dynamo codes that employ two-dimensional or three-dimensional domain decompositions can perform efficiently on up to ∼106 processor cores, paving the way for more realistic simulations in the next model generation.

Original languageEnglish
Pages (from-to)1586-1607
Number of pages22
JournalGeochemistry, Geophysics, Geosystems
Volume17
Issue number5
DOIs
Publication statusPublished - May 1 2016

Fingerprint

Numerical models
central processing units
Earth core
Earth (planet)
Supercomputers
scaling
Extrapolation
geodynamo
supercomputers
pushing
geomagnetism
geomagnetic field
diffusivity
Magnetic fields
Decomposition
simulation
extrapolation
resources
spatial resolution
Computer simulation

All Science Journal Classification (ASJC) codes

  • Geophysics
  • Geochemistry and Petrology

Cite this

Matsui, H., Heien, E., Aubert, J., Aurnou, J. M., Avery, M., Brown, B., ... Willis, A. P. (2016). Performance benchmarks for a next generation numerical dynamo model. Geochemistry, Geophysics, Geosystems, 17(5), 1586-1607. https://doi.org/10.1002/2015GC006159

Performance benchmarks for a next generation numerical dynamo model. / Matsui, Hiroaki; Heien, Eric; Aubert, Julien; Aurnou, Jonathan M.; Avery, Margaret; Brown, Ben; Buffett, Bruce A.; Busse, Friedrich; Christensen, Ulrich R.; Davies, Christopher J.; Featherstone, Nicholas; Gastine, Thomas; Glatzmaier, Gary A.; Gubbins, David; Guermond, Jean Luc; Hayashi, Yoshi Yuki; Hollerbach, Rainer; Hwang, Lorraine J.; Jackson, Andrew; Jones, Chris A.; Jiang, Weiyuan; Kellogg, Louise H.; Kuang, Weijia; Landeau, Maylis; Marti, Philippe; Olson, Peter; Ribeiro, Adolfo; Sasaki, Youhei; Schaeffer, Nathanaël; Simitev, Radostin D.; Sheyko, Andrey; Silva, Luis; Stanley, Sabine; Takahashi, Futoshi; Takehiro, Shin Ichi; Wicht, Johannes; Willis, Ashley P.

In: Geochemistry, Geophysics, Geosystems, Vol. 17, No. 5, 01.05.2016, p. 1586-1607.

Research output: Contribution to journalArticle

Matsui, H, Heien, E, Aubert, J, Aurnou, JM, Avery, M, Brown, B, Buffett, BA, Busse, F, Christensen, UR, Davies, CJ, Featherstone, N, Gastine, T, Glatzmaier, GA, Gubbins, D, Guermond, JL, Hayashi, YY, Hollerbach, R, Hwang, LJ, Jackson, A, Jones, CA, Jiang, W, Kellogg, LH, Kuang, W, Landeau, M, Marti, P, Olson, P, Ribeiro, A, Sasaki, Y, Schaeffer, N, Simitev, RD, Sheyko, A, Silva, L, Stanley, S, Takahashi, F, Takehiro, SI, Wicht, J & Willis, AP 2016, 'Performance benchmarks for a next generation numerical dynamo model', Geochemistry, Geophysics, Geosystems, vol. 17, no. 5, pp. 1586-1607. https://doi.org/10.1002/2015GC006159
Matsui H, Heien E, Aubert J, Aurnou JM, Avery M, Brown B et al. Performance benchmarks for a next generation numerical dynamo model. Geochemistry, Geophysics, Geosystems. 2016 May 1;17(5):1586-1607. https://doi.org/10.1002/2015GC006159
Matsui, Hiroaki ; Heien, Eric ; Aubert, Julien ; Aurnou, Jonathan M. ; Avery, Margaret ; Brown, Ben ; Buffett, Bruce A. ; Busse, Friedrich ; Christensen, Ulrich R. ; Davies, Christopher J. ; Featherstone, Nicholas ; Gastine, Thomas ; Glatzmaier, Gary A. ; Gubbins, David ; Guermond, Jean Luc ; Hayashi, Yoshi Yuki ; Hollerbach, Rainer ; Hwang, Lorraine J. ; Jackson, Andrew ; Jones, Chris A. ; Jiang, Weiyuan ; Kellogg, Louise H. ; Kuang, Weijia ; Landeau, Maylis ; Marti, Philippe ; Olson, Peter ; Ribeiro, Adolfo ; Sasaki, Youhei ; Schaeffer, Nathanaël ; Simitev, Radostin D. ; Sheyko, Andrey ; Silva, Luis ; Stanley, Sabine ; Takahashi, Futoshi ; Takehiro, Shin Ichi ; Wicht, Johannes ; Willis, Ashley P. / Performance benchmarks for a next generation numerical dynamo model. In: Geochemistry, Geophysics, Geosystems. 2016 ; Vol. 17, No. 5. pp. 1586-1607.
@article{5e5fbd75ef894a7bb306eddb826462bb,
title = "Performance benchmarks for a next generation numerical dynamo model",
abstract = "Numerical simulations of the geodynamo have successfully represented many observable characteristics of the geomagnetic field, yielding insight into the fundamental processes that generate magnetic fields in the Earth's core. Because of limited spatial resolution, however, the diffusivities in numerical dynamo models are much larger than those in the Earth's core, and consequently, questions remain about how realistic these models are. The typical strategy used to address this issue has been to continue to increase the resolution of these quasi-laminar models with increasing computational resources, thus pushing them toward more realistic parameter regimes. We assess which methods are most promising for the next generation of supercomputers, which will offer access to O(106) processor cores for large problems. Here we report performance and accuracy benchmarks from 15 dynamo codes that employ a range of numerical and parallelization methods. Computational performance is assessed on the basis of weak and strong scaling behavior up to 16,384 processor cores. Extrapolations of our weak-scaling results indicate that dynamo codes that employ two-dimensional or three-dimensional domain decompositions can perform efficiently on up to ∼106 processor cores, paving the way for more realistic simulations in the next model generation.",
author = "Hiroaki Matsui and Eric Heien and Julien Aubert and Aurnou, {Jonathan M.} and Margaret Avery and Ben Brown and Buffett, {Bruce A.} and Friedrich Busse and Christensen, {Ulrich R.} and Davies, {Christopher J.} and Nicholas Featherstone and Thomas Gastine and Glatzmaier, {Gary A.} and David Gubbins and Guermond, {Jean Luc} and Hayashi, {Yoshi Yuki} and Rainer Hollerbach and Hwang, {Lorraine J.} and Andrew Jackson and Jones, {Chris A.} and Weiyuan Jiang and Kellogg, {Louise H.} and Weijia Kuang and Maylis Landeau and Philippe Marti and Peter Olson and Adolfo Ribeiro and Youhei Sasaki and Nathana{\"e}l Schaeffer and Simitev, {Radostin D.} and Andrey Sheyko and Luis Silva and Sabine Stanley and Futoshi Takahashi and Takehiro, {Shin Ichi} and Johannes Wicht and Willis, {Ashley P.}",
year = "2016",
month = "5",
day = "1",
doi = "10.1002/2015GC006159",
language = "English",
volume = "17",
pages = "1586--1607",
journal = "Geochemistry, Geophysics, Geosystems",
issn = "1525-2027",
publisher = "American Geophysical Union",
number = "5",

}

TY - JOUR

T1 - Performance benchmarks for a next generation numerical dynamo model

AU - Matsui, Hiroaki

AU - Heien, Eric

AU - Aubert, Julien

AU - Aurnou, Jonathan M.

AU - Avery, Margaret

AU - Brown, Ben

AU - Buffett, Bruce A.

AU - Busse, Friedrich

AU - Christensen, Ulrich R.

AU - Davies, Christopher J.

AU - Featherstone, Nicholas

AU - Gastine, Thomas

AU - Glatzmaier, Gary A.

AU - Gubbins, David

AU - Guermond, Jean Luc

AU - Hayashi, Yoshi Yuki

AU - Hollerbach, Rainer

AU - Hwang, Lorraine J.

AU - Jackson, Andrew

AU - Jones, Chris A.

AU - Jiang, Weiyuan

AU - Kellogg, Louise H.

AU - Kuang, Weijia

AU - Landeau, Maylis

AU - Marti, Philippe

AU - Olson, Peter

AU - Ribeiro, Adolfo

AU - Sasaki, Youhei

AU - Schaeffer, Nathanaël

AU - Simitev, Radostin D.

AU - Sheyko, Andrey

AU - Silva, Luis

AU - Stanley, Sabine

AU - Takahashi, Futoshi

AU - Takehiro, Shin Ichi

AU - Wicht, Johannes

AU - Willis, Ashley P.

PY - 2016/5/1

Y1 - 2016/5/1

N2 - Numerical simulations of the geodynamo have successfully represented many observable characteristics of the geomagnetic field, yielding insight into the fundamental processes that generate magnetic fields in the Earth's core. Because of limited spatial resolution, however, the diffusivities in numerical dynamo models are much larger than those in the Earth's core, and consequently, questions remain about how realistic these models are. The typical strategy used to address this issue has been to continue to increase the resolution of these quasi-laminar models with increasing computational resources, thus pushing them toward more realistic parameter regimes. We assess which methods are most promising for the next generation of supercomputers, which will offer access to O(106) processor cores for large problems. Here we report performance and accuracy benchmarks from 15 dynamo codes that employ a range of numerical and parallelization methods. Computational performance is assessed on the basis of weak and strong scaling behavior up to 16,384 processor cores. Extrapolations of our weak-scaling results indicate that dynamo codes that employ two-dimensional or three-dimensional domain decompositions can perform efficiently on up to ∼106 processor cores, paving the way for more realistic simulations in the next model generation.

AB - Numerical simulations of the geodynamo have successfully represented many observable characteristics of the geomagnetic field, yielding insight into the fundamental processes that generate magnetic fields in the Earth's core. Because of limited spatial resolution, however, the diffusivities in numerical dynamo models are much larger than those in the Earth's core, and consequently, questions remain about how realistic these models are. The typical strategy used to address this issue has been to continue to increase the resolution of these quasi-laminar models with increasing computational resources, thus pushing them toward more realistic parameter regimes. We assess which methods are most promising for the next generation of supercomputers, which will offer access to O(106) processor cores for large problems. Here we report performance and accuracy benchmarks from 15 dynamo codes that employ a range of numerical and parallelization methods. Computational performance is assessed on the basis of weak and strong scaling behavior up to 16,384 processor cores. Extrapolations of our weak-scaling results indicate that dynamo codes that employ two-dimensional or three-dimensional domain decompositions can perform efficiently on up to ∼106 processor cores, paving the way for more realistic simulations in the next model generation.

UR - http://www.scopus.com/inward/record.url?scp=84976531791&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84976531791&partnerID=8YFLogxK

U2 - 10.1002/2015GC006159

DO - 10.1002/2015GC006159

M3 - Article

AN - SCOPUS:84976531791

VL - 17

SP - 1586

EP - 1607

JO - Geochemistry, Geophysics, Geosystems

JF - Geochemistry, Geophysics, Geosystems

SN - 1525-2027

IS - 5

ER -