Computational & Technology Resources
an online resource for computational,
engineering & technology publications 

Computational Science, Engineering & Technology Series
ISSN 17593158 CSETS: 4
HIGH PERFORMANCE COMPUTING FOR COMPUTATIONAL MECHANICS Edited by: B.H.V. Topping, L. Lämmer
Chapter 10
Parallelisation of a Finite Element Program P. Devloo^{1} and E.C. da Silva^{2}
^{1}FEC/UNICAMP, São Paulo, Brazil P. Devloo, E.C. da Silva, "Parallelisation of a Finite Element Program", in B.H.V. Topping, L. Lämmer, (Editors), "High Performance Computing for Computational Mechanics", SaxeCoburg Publications, Stirlingshire, UK, Chapter 10, pp 173185, 2000. doi:10.4203/csets.4.10
Abstract
An object oriented finite element program is parallelized using
OOPAR, an object oriented environment for parallel computing. Within the
framework presented, the finite element meshes corresponding to subdomains of
the complete domain are read by the individual nodes. The submeshes communicate
between themselves to identify neighbouring nodes and an iterative
conjugate gradient method solves the global system of equations. No single processor
needs to allocate memory for the global grid, nor is there any need to
even allocate a solution or residual vector of global size. The issuing algorithm
is therefore suited for solving very large problems on a large number of nodes,
where none of the nodes is capable of holding the global mesh and/or solution.
purchase the fulltext of this chapter (price £20)
go to the previous chapter 
