Possibilities of Workstation Clusters for Parallel Finite–Element Analysis

Horie, Takeshi; Kuramae, Hiroyuki

doi:10.1111/0885-9507.00051

Cited by 8 publications

(3 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, the domain decomposition method [5,6] is adopted as a solver of parallel FEM because large granularity is required to achieve higher computational performance for PC cluster. MPI [7], which is de facto standard of message-passing library, is used for parallel programming.…”

Section: Simulation Methodsmentioning

confidence: 99%

See 1 more Smart Citation

FRP Fracture Simulation Using Parallel Processing on a PC Cluster

Shimamura

Yamamura

Todoroki

2007

KEM

View full text Add to dashboard Cite

Abstract. Recently, applications of integrated large composite structures have been attempted to many structures of vehicles. In order to improve the cost performance and reliability, it is necessary to judge the structural integrity of composite structures. Fracture simulation techniques using FEM have been developed for the purpose. Since a number of iterations of finite element analysis are required in the fracture simulation, the simulation techniques consume many memory resources and much calculation time. In this study, a personal computer cluster (PC cluster) and the domain decomposition method were incorporated into a fracture simulation system. Calculations using a Windows PC cluster were carried out to confirm the efficiency of the proposed simulation system. As a result, it is concluded that adopting the domain decomposition method and the computer cluster is remarkably efficient to reduce calculation time.

show abstract

Section: Simulation Methodsmentioning

confidence: 99%

“…Recent advance on personal computer and computer network technology enables us to construct a personal computer cluster (PC cluster) as a parallel computing system [5,6]. A PC cluster consists of several PCs and Ethernet.…”

Section: Introductionmentioning

confidence: 99%

FRP Fracture Simulation Using Parallel Processing on a PC Cluster

Shimamura

Yamamura

Todoroki

2007

KEM

View full text Add to dashboard Cite

show abstract

“…Moreover, this model satisfies the important directive for the efficient handling of ethernet networks, according to which messages should be organized into groups and passed in one step: since the slaves communicate only through the master, each slave in need to communicate with several other slaves sends all its messages to the master in one communication step; the master groups together the results for each slave and sends them to their destination. Some characteristic papers that have successfully used the master/slave model to develop distributed FEA programs for clusters of networked computers are given in the reference list (Hudli and Pidaparti 1994;Chadha and Baugh 1996;Feriani and Genna 1996;Horie and Kuramae 1997).…”

Section: The Master/slave Model For Distributed Computingmentioning

confidence: 99%

Enhancing the performance of the FETI method with preconditioning ¶techniques implemented on clusters of networked computers

Charmpis¹,

Papadrakakis²

2002

Computational Mechanics

View full text Add to dashboard Cite

The FETI domain decomposition method for solving large-scale problems in computational structural mechanics involves the solution of an interface problem, which is handled by a Preconditioned Conjugate Projected Gradient (PCPG) algorithm. Two preconditioners are widely used to accelerate the convergence of the iterative PCPG algorithm: the optimal Dirichlet preconditioner and the economical lumped preconditioner. The Dirichlet preconditioner is computationally more efficient than the lumped preconditioner for ill-conditioned problems, but needs additional storage for the stiffness matrices of the subdomains' internal degrees of freedom (d.o.f.). In this study a new set of PCPG preconditioners is presented by providing approximate expressions to the inverse iteration matrix of the PCPG algorithm. The resulting approximate Dirichlet preconditioners are obtained by using instead of the whole stiffness matrix of the internal d.o.f. in each subdomain the following alternatives: a diagonal scaling matrix, a SSOR type matrix or an incomplete Cholesky factorization matrix. The computational behavior and performance of the proposed PCPG preconditioners is evaluated using an implementation of the FETI method on a cluster of ethernetnetworked PCs running the message passing software PVM. It is demonstrated that the FETI method equipped with the approximate Dirichlet preconditioners leads for a number of large-scale problems to faster and less storage demanding overall solutions than with either Dirichlet or lumped preconditioner. IntroductionThe method of Finite Element Tearing and Interconnecting (FETI) is an efficient and well-documented parallel solution approach for large-scale structural problems (Farhat and Roux 1994a;Bitzarakis et al. 1997). According to the FETI method the domain is partitioned into a set of totally disconnected subdomains and the global problem is replaced by subdomain equations of equilibrium subject to the compatibility of the subdomain displacements across the subdomain interfaces. The resulting interface problem, which is in general indefinite due to the presence of floating subdomains, is handled by a Preconditioned Conjugate Projected Gradient (PCPG) algorithm.A platform for parallel computing that is continuously gaining popularity over the last years is the cluster of computers, which consists of several computers networked by a Local Area Network (LAN). Such clusters provide program developers and users with a familiar and costeffective environment for high-performance computing. The cluster's computers are usually interconnected, however, via inexpensive but slow LANs (the most commonly available is the classical ethernet network), which are usually not dedicated to specific users and may suffer heavy communication loads. Thus, to deal with the low communication speed that most LANs can offer, Finite Element Analysis (FEA) algorithms require specialized implementations exploiting the characteristics of such networks. In this work the FETI method is applied on a cluster of ethernet-network...

show abstract