A comparison between parallelization approaches in molecular dynamics simulations on GPUs

Lorenzo Rovigatti, Petr Sulc, István Z. Reguly, Flavio Romano

Research output: Contribution to journalArticle

25 Citations (Scopus)

Abstract

We test the relative performances of two different approaches to the computation of forces for molecular dynamics simulations on graphics processing units. A "vertex-based" approach, where a computing thread is started per particle, is compared to an "edge-based" approach, where a thread is started per each potentially non-zero interaction. We find that the former is more efficient for systems with many simple interactions per particle while the latter is more efficient if the system has more complicated interactions or fewer of them. By comparing computation times on more and less recent graphics processing unit technology, we predict that, if the current trend of increasing the number of processing cores - as opposed to their computing power - remains, the "edge-based" approach will gradually become the most efficient choice in an increasing number of cases.

Original languageEnglish (US)
Pages (from-to)1-8
Number of pages8
JournalJournal of Computational Chemistry
Volume36
Issue number1
DOIs
StatePublished - Oct 30 2015
Externally publishedYes

Fingerprint

Parallelization
Molecular Dynamics Simulation
Molecular dynamics
Graphics Processing Unit
Thread
Computer simulation
Interaction
Computing
Processing
Predict
Vertex of a graph
Graphics processing unit
Trends

Keywords

  • Graphics processing unit
  • Molecular dynamics
  • Parallelization
  • Soft matter

ASJC Scopus subject areas

  • Chemistry(all)
  • Computational Mathematics

Cite this

A comparison between parallelization approaches in molecular dynamics simulations on GPUs. / Rovigatti, Lorenzo; Sulc, Petr; Reguly, István Z.; Romano, Flavio.

In: Journal of Computational Chemistry, Vol. 36, No. 1, 30.10.2015, p. 1-8.

Research output: Contribution to journalArticle

Rovigatti, Lorenzo ; Sulc, Petr ; Reguly, István Z. ; Romano, Flavio. / A comparison between parallelization approaches in molecular dynamics simulations on GPUs. In: Journal of Computational Chemistry. 2015 ; Vol. 36, No. 1. pp. 1-8.
@article{4cd89ff81d6c46569372240385228ef5,
title = "A comparison between parallelization approaches in molecular dynamics simulations on GPUs",
abstract = "We test the relative performances of two different approaches to the computation of forces for molecular dynamics simulations on graphics processing units. A {"}vertex-based{"} approach, where a computing thread is started per particle, is compared to an {"}edge-based{"} approach, where a thread is started per each potentially non-zero interaction. We find that the former is more efficient for systems with many simple interactions per particle while the latter is more efficient if the system has more complicated interactions or fewer of them. By comparing computation times on more and less recent graphics processing unit technology, we predict that, if the current trend of increasing the number of processing cores - as opposed to their computing power - remains, the {"}edge-based{"} approach will gradually become the most efficient choice in an increasing number of cases.",
keywords = "Graphics processing unit, Molecular dynamics, Parallelization, Soft matter",
author = "Lorenzo Rovigatti and Petr Sulc and Reguly, {Istv{\'a}n Z.} and Flavio Romano",
year = "2015",
month = "10",
day = "30",
doi = "10.1002/jcc.23763",
language = "English (US)",
volume = "36",
pages = "1--8",
journal = "Journal of Computational Chemistry",
issn = "0192-8651",
publisher = "John Wiley and Sons Inc.",
number = "1",

}

TY - JOUR

T1 - A comparison between parallelization approaches in molecular dynamics simulations on GPUs

AU - Rovigatti, Lorenzo

AU - Sulc, Petr

AU - Reguly, István Z.

AU - Romano, Flavio

PY - 2015/10/30

Y1 - 2015/10/30

N2 - We test the relative performances of two different approaches to the computation of forces for molecular dynamics simulations on graphics processing units. A "vertex-based" approach, where a computing thread is started per particle, is compared to an "edge-based" approach, where a thread is started per each potentially non-zero interaction. We find that the former is more efficient for systems with many simple interactions per particle while the latter is more efficient if the system has more complicated interactions or fewer of them. By comparing computation times on more and less recent graphics processing unit technology, we predict that, if the current trend of increasing the number of processing cores - as opposed to their computing power - remains, the "edge-based" approach will gradually become the most efficient choice in an increasing number of cases.

AB - We test the relative performances of two different approaches to the computation of forces for molecular dynamics simulations on graphics processing units. A "vertex-based" approach, where a computing thread is started per particle, is compared to an "edge-based" approach, where a thread is started per each potentially non-zero interaction. We find that the former is more efficient for systems with many simple interactions per particle while the latter is more efficient if the system has more complicated interactions or fewer of them. By comparing computation times on more and less recent graphics processing unit technology, we predict that, if the current trend of increasing the number of processing cores - as opposed to their computing power - remains, the "edge-based" approach will gradually become the most efficient choice in an increasing number of cases.

KW - Graphics processing unit

KW - Molecular dynamics

KW - Parallelization

KW - Soft matter

UR - http://www.scopus.com/inward/record.url?scp=84918814617&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84918814617&partnerID=8YFLogxK

U2 - 10.1002/jcc.23763

DO - 10.1002/jcc.23763

M3 - Article

VL - 36

SP - 1

EP - 8

JO - Journal of Computational Chemistry

JF - Journal of Computational Chemistry

SN - 0192-8651

IS - 1

ER -