Implementation of implicit finite element methods for incompressible flows on the CM-5

J. G. Kennedy, M. Behr, V. Kalro, Tayfun E. Tezduyar

Research output: Contribution to journalArticle

26 Citations (Scopus)

Abstract

A parallel implementation of an implicit finite element formulation for incompressible fluids on a distributed-memory massively parallel computer is presented. The dominant issue that distinguishes the implementation of finite element problems on distributed-memory computers from that on traditional shared-memory scalar or vector computers is the distribution of data (and hence workload) to the processors and the non-uniform memory hierarchy associated with the processors, particularly the non-uniform costs associated with on-processor and off-processor memory references. Accessing data stored in a remote processor requires computing resources an order of magnitude greater than accessing data locally in a processor. This distribution of data motivates the development of alternatives to traditional algorithms and data structures designed for shared-memory computers, which must now account for distributed-memory architectures. Data structures as well as data decomposition and data communication algorithms designed for distributed-memory computers are presented in the context of high level language constructs from High Performance Fortran. The discussion relies primarily on abstract features of the hardware and software environment and should be applicable, in principle, to a variety of distributed-memory system. The actual implementation is carried out on a Connection Machine CM-5 system with high performance communication functions.

Original languageEnglish
Pages (from-to)95-111
Number of pages17
JournalComputer Methods in Applied Mechanics and Engineering
Volume119
Issue number1-2
DOIs
Publication statusPublished - 1994
Externally publishedYes

Fingerprint

incompressible flow
distributed memory
Incompressible flow
central processing units
finite element method
Finite element method
memory (computers)
Data storage equipment
data structures
Connection Machine
communication
high level languages
Data structures
parallel computers
incompressible fluids
Memory architecture
High level languages
hierarchies
Communication
resources

ASJC Scopus subject areas

  • Computer Science Applications
  • Computational Mechanics
  • Engineering(all)

Cite this

Implementation of implicit finite element methods for incompressible flows on the CM-5. / Kennedy, J. G.; Behr, M.; Kalro, V.; Tezduyar, Tayfun E.

In: Computer Methods in Applied Mechanics and Engineering, Vol. 119, No. 1-2, 1994, p. 95-111.

Research output: Contribution to journalArticle

@article{e3fae3445ed34f178c704c29f0e35ecd,
title = "Implementation of implicit finite element methods for incompressible flows on the CM-5",
abstract = "A parallel implementation of an implicit finite element formulation for incompressible fluids on a distributed-memory massively parallel computer is presented. The dominant issue that distinguishes the implementation of finite element problems on distributed-memory computers from that on traditional shared-memory scalar or vector computers is the distribution of data (and hence workload) to the processors and the non-uniform memory hierarchy associated with the processors, particularly the non-uniform costs associated with on-processor and off-processor memory references. Accessing data stored in a remote processor requires computing resources an order of magnitude greater than accessing data locally in a processor. This distribution of data motivates the development of alternatives to traditional algorithms and data structures designed for shared-memory computers, which must now account for distributed-memory architectures. Data structures as well as data decomposition and data communication algorithms designed for distributed-memory computers are presented in the context of high level language constructs from High Performance Fortran. The discussion relies primarily on abstract features of the hardware and software environment and should be applicable, in principle, to a variety of distributed-memory system. The actual implementation is carried out on a Connection Machine CM-5 system with high performance communication functions.",
author = "Kennedy, {J. G.} and M. Behr and V. Kalro and Tezduyar, {Tayfun E.}",
year = "1994",
doi = "10.1016/0045-7825(94)00078-6",
language = "English",
volume = "119",
pages = "95--111",
journal = "Computer Methods in Applied Mechanics and Engineering",
issn = "0374-2830",
publisher = "Elsevier",
number = "1-2",

}

TY - JOUR

T1 - Implementation of implicit finite element methods for incompressible flows on the CM-5

AU - Kennedy, J. G.

AU - Behr, M.

AU - Kalro, V.

AU - Tezduyar, Tayfun E.

PY - 1994

Y1 - 1994

N2 - A parallel implementation of an implicit finite element formulation for incompressible fluids on a distributed-memory massively parallel computer is presented. The dominant issue that distinguishes the implementation of finite element problems on distributed-memory computers from that on traditional shared-memory scalar or vector computers is the distribution of data (and hence workload) to the processors and the non-uniform memory hierarchy associated with the processors, particularly the non-uniform costs associated with on-processor and off-processor memory references. Accessing data stored in a remote processor requires computing resources an order of magnitude greater than accessing data locally in a processor. This distribution of data motivates the development of alternatives to traditional algorithms and data structures designed for shared-memory computers, which must now account for distributed-memory architectures. Data structures as well as data decomposition and data communication algorithms designed for distributed-memory computers are presented in the context of high level language constructs from High Performance Fortran. The discussion relies primarily on abstract features of the hardware and software environment and should be applicable, in principle, to a variety of distributed-memory system. The actual implementation is carried out on a Connection Machine CM-5 system with high performance communication functions.

AB - A parallel implementation of an implicit finite element formulation for incompressible fluids on a distributed-memory massively parallel computer is presented. The dominant issue that distinguishes the implementation of finite element problems on distributed-memory computers from that on traditional shared-memory scalar or vector computers is the distribution of data (and hence workload) to the processors and the non-uniform memory hierarchy associated with the processors, particularly the non-uniform costs associated with on-processor and off-processor memory references. Accessing data stored in a remote processor requires computing resources an order of magnitude greater than accessing data locally in a processor. This distribution of data motivates the development of alternatives to traditional algorithms and data structures designed for shared-memory computers, which must now account for distributed-memory architectures. Data structures as well as data decomposition and data communication algorithms designed for distributed-memory computers are presented in the context of high level language constructs from High Performance Fortran. The discussion relies primarily on abstract features of the hardware and software environment and should be applicable, in principle, to a variety of distributed-memory system. The actual implementation is carried out on a Connection Machine CM-5 system with high performance communication functions.

UR - http://www.scopus.com/inward/record.url?scp=0028534248&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0028534248&partnerID=8YFLogxK

U2 - 10.1016/0045-7825(94)00078-6

DO - 10.1016/0045-7825(94)00078-6

M3 - Article

AN - SCOPUS:0028534248

VL - 119

SP - 95

EP - 111

JO - Computer Methods in Applied Mechanics and Engineering

JF - Computer Methods in Applied Mechanics and Engineering

SN - 0374-2830

IS - 1-2

ER -