High-performance parallel implicit CFD

William D. Gropp, Dinesh K. Kaushik, David E. Keyes, Barry F. Smith

Research output: Contribution to journalArticle

116 Citations (Scopus)

Abstract

Fluid dynamical simulations based on finite discretizations on (quasi-)static grids scale well in parallel, but execute at a disappointing percentage of per-processor peak floating point operation rates without special attention to layout and access ordering of data. We document both claims from our experience with an unstructured grid CFD code that is typical of the state of the practice at NASA. These basic performance characteristics of PDE-based codes can be understood with surprisingly simple models, for which we quote earlier work, presenting primarily experiemental results. The performance models and experimental results motivate algorithmic and software practices that lead to improvements in both parallel scalability and per node performance. This snapshot of ongoing work updates our 1999 Bell Prize-winning simulation on ASCI computers

Original languageEnglish
Pages (from-to)337-362
Number of pages26
JournalParallel Computing
Volume27
Issue number4
DOIs
Publication statusPublished - Mar 2001
Externally publishedYes

Fingerprint

Computational fluid dynamics
High Performance
Unstructured Grid
Snapshot
Floating point
Performance Model
NASA
Percentage
Scalability
Layout
Simulation
Update
Discretization
Grid
Fluid
Software
Fluids
Experimental Results
Vertex of a graph
Model

Keywords

  • Computational fluid dynamics
  • High-performance computing
  • Parallel implicit solvers
  • Unstructured grids

ASJC Scopus subject areas

  • Computer Science Applications
  • Hardware and Architecture
  • Control and Systems Engineering

Cite this

High-performance parallel implicit CFD. / Gropp, William D.; Kaushik, Dinesh K.; Keyes, David E.; Smith, Barry F.

In: Parallel Computing, Vol. 27, No. 4, 03.2001, p. 337-362.

Research output: Contribution to journalArticle

Gropp, William D. ; Kaushik, Dinesh K. ; Keyes, David E. ; Smith, Barry F. / High-performance parallel implicit CFD. In: Parallel Computing. 2001 ; Vol. 27, No. 4. pp. 337-362.
@article{5b2627815bf14ebd96b26c7bbd7926ba,
title = "High-performance parallel implicit CFD",
abstract = "Fluid dynamical simulations based on finite discretizations on (quasi-)static grids scale well in parallel, but execute at a disappointing percentage of per-processor peak floating point operation rates without special attention to layout and access ordering of data. We document both claims from our experience with an unstructured grid CFD code that is typical of the state of the practice at NASA. These basic performance characteristics of PDE-based codes can be understood with surprisingly simple models, for which we quote earlier work, presenting primarily experiemental results. The performance models and experimental results motivate algorithmic and software practices that lead to improvements in both parallel scalability and per node performance. This snapshot of ongoing work updates our 1999 Bell Prize-winning simulation on ASCI computers",
keywords = "Computational fluid dynamics, High-performance computing, Parallel implicit solvers, Unstructured grids",
author = "Gropp, {William D.} and Kaushik, {Dinesh K.} and Keyes, {David E.} and Smith, {Barry F.}",
year = "2001",
month = "3",
doi = "10.1016/S0167-8191(00)00075-2",
language = "English",
volume = "27",
pages = "337--362",
journal = "Parallel Computing",
issn = "0167-8191",
publisher = "Elsevier",
number = "4",

}

TY - JOUR

T1 - High-performance parallel implicit CFD

AU - Gropp, William D.

AU - Kaushik, Dinesh K.

AU - Keyes, David E.

AU - Smith, Barry F.

PY - 2001/3

Y1 - 2001/3

N2 - Fluid dynamical simulations based on finite discretizations on (quasi-)static grids scale well in parallel, but execute at a disappointing percentage of per-processor peak floating point operation rates without special attention to layout and access ordering of data. We document both claims from our experience with an unstructured grid CFD code that is typical of the state of the practice at NASA. These basic performance characteristics of PDE-based codes can be understood with surprisingly simple models, for which we quote earlier work, presenting primarily experiemental results. The performance models and experimental results motivate algorithmic and software practices that lead to improvements in both parallel scalability and per node performance. This snapshot of ongoing work updates our 1999 Bell Prize-winning simulation on ASCI computers

AB - Fluid dynamical simulations based on finite discretizations on (quasi-)static grids scale well in parallel, but execute at a disappointing percentage of per-processor peak floating point operation rates without special attention to layout and access ordering of data. We document both claims from our experience with an unstructured grid CFD code that is typical of the state of the practice at NASA. These basic performance characteristics of PDE-based codes can be understood with surprisingly simple models, for which we quote earlier work, presenting primarily experiemental results. The performance models and experimental results motivate algorithmic and software practices that lead to improvements in both parallel scalability and per node performance. This snapshot of ongoing work updates our 1999 Bell Prize-winning simulation on ASCI computers

KW - Computational fluid dynamics

KW - High-performance computing

KW - Parallel implicit solvers

KW - Unstructured grids

UR - http://www.scopus.com/inward/record.url?scp=0035276480&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035276480&partnerID=8YFLogxK

U2 - 10.1016/S0167-8191(00)00075-2

DO - 10.1016/S0167-8191(00)00075-2

M3 - Article

VL - 27

SP - 337

EP - 362

JO - Parallel Computing

JF - Parallel Computing

SN - 0167-8191

IS - 4

ER -