Acceleration of hydro poro-elastic damage simulation in a shared-memory environment

Harel Levin; Gal Oren; Eyal Shalev; Vladimir Lyakhovsky

doi:10.3233/APC200059

Acceleration of hydro poro-elastic damage simulation in a shared-memory environment

Harel Levin, Gal Oren, Eyal Shalev, Vladimir Lyakhovsky

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

2 Scopus citations

Abstract

Hydro-PED [1] is a numerical simulation software which models nucleation and propagation of damage zones and seismicity patterns induced by wellbore fluid injection. While most of the studies in geo-physical simulation acceleration and parallelization usually focus on exascale scenarios which are translated into vast meshes, encouraging a distributed fashion of parallelization, the nature of the current simulations of Hydro-PED dictates amount of data that can conveniently fit on a single compute node - NUMA and accelerator memory alike. Thus shared-memory parallelization (such as OpenMP) can be fully implemented. In order to utilize this insight, Hydro-PED was interfaced with Trilinos [2] linear algebra solvers package, which enabled an evolution to iterative methods such as CG and GMRES. Additionally, several code sectors were parallelized and offloaded to an accelerator using OpenMP in a fine grained manner. The changes implemented in Hydro-PED gained a total speedup of x5-x12, which will enable Hydro-PED to calculate long-term simulation scenarios of hundreds of years in a feasible time - a few weeks rather than a year.

Original language	English
Title of host publication	Parallel Computing
Subtitle of host publication	Technology Trends
Editors	Ian Foster, Gerhard R. Joubert, Ludek Kucera, Wolfgang E. Nagel, Frans Peters
Publisher	IOS Press BV
Pages	341-353
Number of pages	13
ISBN (Electronic)	9781643680705
DOIs	https://doi.org/10.3233/APC200059
State	Published - 2020
Externally published	Yes

Publication series

Name	Advances in Parallel Computing
Volume	36
ISSN (Print)	0927-5452
ISSN (Electronic)	1879-808X

Bibliographical note

Publisher Copyright:
© 2020 The authors and IOS Press.

Funding

1This work was supported by the Lynn and William Frankel Center for Computer Science. Computational support was provided by the NegevHPC project [3]. Special thanks goes to A. Tarabay, M. Homel and J. White, our colleagues in Lawrence Livermore National Laboratory for their priceless support and guidance during this work. Source code is accessible at https://github.com/harellevin/Hydro-PED.

Funders	Funder number
Lynn and William Frankel Center for Computer Science

Keywords

Accelerators
Geological Simulation
Numerical Linear Algebra
Shared Memory
Trilinos

Access to Document

10.3233/APC200059

Cite this

@inproceedings{ee9b010c247845eaa9de9d8e9201b0e1,

title = "Acceleration of hydro poro-elastic damage simulation in a shared-memory environment",

abstract = "Hydro-PED [1] is a numerical simulation software which models nucleation and propagation of damage zones and seismicity patterns induced by wellbore fluid injection. While most of the studies in geo-physical simulation acceleration and parallelization usually focus on exascale scenarios which are translated into vast meshes, encouraging a distributed fashion of parallelization, the nature of the current simulations of Hydro-PED dictates amount of data that can conveniently fit on a single compute node - NUMA and accelerator memory alike. Thus shared-memory parallelization (such as OpenMP) can be fully implemented. In order to utilize this insight, Hydro-PED was interfaced with Trilinos [2] linear algebra solvers package, which enabled an evolution to iterative methods such as CG and GMRES. Additionally, several code sectors were parallelized and offloaded to an accelerator using OpenMP in a fine grained manner. The changes implemented in Hydro-PED gained a total speedup of x5-x12, which will enable Hydro-PED to calculate long-term simulation scenarios of hundreds of years in a feasible time - a few weeks rather than a year.",

keywords = "Accelerators, Geological Simulation, Numerical Linear Algebra, Shared Memory, Trilinos",

author = "Harel Levin and Gal Oren and Eyal Shalev and Vladimir Lyakhovsky",

note = "Publisher Copyright: {\textcopyright} 2020 The authors and IOS Press.",

year = "2020",

doi = "10.3233/APC200059",

language = "אנגלית",

series = "Advances in Parallel Computing",

publisher = "IOS Press BV",

pages = "341--353",

editor = "Ian Foster and Joubert, {Gerhard R.} and Ludek Kucera and Nagel, {Wolfgang E.} and Frans Peters",

booktitle = "Parallel Computing",

address = "הולנד",

}

Acceleration of hydro poro-elastic damage simulation in a shared-memory environment. / Levin, Harel; Oren, Gal; Shalev, Eyal et al.
Parallel Computing: Technology Trends. ed. / Ian Foster; Gerhard R. Joubert; Ludek Kucera; Wolfgang E. Nagel; Frans Peters. IOS Press BV, 2020. p. 341-353 (Advances in Parallel Computing; Vol. 36).

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

TY - GEN

T1 - Acceleration of hydro poro-elastic damage simulation in a shared-memory environment

AU - Levin, Harel

AU - Oren, Gal

AU - Shalev, Eyal

AU - Lyakhovsky, Vladimir

PY - 2020

Y1 - 2020

N2 - Hydro-PED [1] is a numerical simulation software which models nucleation and propagation of damage zones and seismicity patterns induced by wellbore fluid injection. While most of the studies in geo-physical simulation acceleration and parallelization usually focus on exascale scenarios which are translated into vast meshes, encouraging a distributed fashion of parallelization, the nature of the current simulations of Hydro-PED dictates amount of data that can conveniently fit on a single compute node - NUMA and accelerator memory alike. Thus shared-memory parallelization (such as OpenMP) can be fully implemented. In order to utilize this insight, Hydro-PED was interfaced with Trilinos [2] linear algebra solvers package, which enabled an evolution to iterative methods such as CG and GMRES. Additionally, several code sectors were parallelized and offloaded to an accelerator using OpenMP in a fine grained manner. The changes implemented in Hydro-PED gained a total speedup of x5-x12, which will enable Hydro-PED to calculate long-term simulation scenarios of hundreds of years in a feasible time - a few weeks rather than a year.

AB - Hydro-PED [1] is a numerical simulation software which models nucleation and propagation of damage zones and seismicity patterns induced by wellbore fluid injection. While most of the studies in geo-physical simulation acceleration and parallelization usually focus on exascale scenarios which are translated into vast meshes, encouraging a distributed fashion of parallelization, the nature of the current simulations of Hydro-PED dictates amount of data that can conveniently fit on a single compute node - NUMA and accelerator memory alike. Thus shared-memory parallelization (such as OpenMP) can be fully implemented. In order to utilize this insight, Hydro-PED was interfaced with Trilinos [2] linear algebra solvers package, which enabled an evolution to iterative methods such as CG and GMRES. Additionally, several code sectors were parallelized and offloaded to an accelerator using OpenMP in a fine grained manner. The changes implemented in Hydro-PED gained a total speedup of x5-x12, which will enable Hydro-PED to calculate long-term simulation scenarios of hundreds of years in a feasible time - a few weeks rather than a year.

KW - Accelerators

KW - Geological Simulation

KW - Numerical Linear Algebra

KW - Shared Memory

KW - Trilinos

UR - http://www.scopus.com/inward/record.url?scp=85083050812&partnerID=8YFLogxK

U2 - 10.3233/APC200059

DO - 10.3233/APC200059

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:85083050812

T3 - Advances in Parallel Computing

SP - 341

EP - 353

BT - Parallel Computing

A2 - Foster, Ian

A2 - Joubert, Gerhard R.

A2 - Kucera, Ludek

A2 - Nagel, Wolfgang E.

A2 - Peters, Frans

PB - IOS Press BV

ER -

Acceleration of hydro poro-elastic damage simulation in a shared-memory environment

Abstract

Publication series

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this