PARASOL (1996-1999)
- Test data from the PARASOL project is
here.
Project summary
An industrial requirement is to be able to efficiently solve large sparse
linear equation systems. Many codes in CFD and CMD use either direct solvers
or iterative solvers with preconditioning - both to solve
large coarse matrices.
Simple iterative methods such as CG and SOR do not work for stability reasons.
Solving such large systems in reasonable time requires the use of MPP systems.
Therefore, message passing versions of direct
sparse solvers are needed. Up to now, however, no generally available library of
scalable parallel sparse matrix solvers or preconditioners exists.
The aim of the PARASOL project (EU ESPRIT IV LTR Project No. 20160)
was to develop new parallel algorithms for direct solution of sparse systems of
equations and for their preconditioning. These algorithms were implemented
as portable prototypes and integrated into the PARASOL library.
A further industrial requirement was that the PARASOL library
must contain interfaces to domain decomposition algorithms, utilities for data
sorting, symbolic factorization and mapping algorithms.
Industrial developers played a key role in the project and were active in the specification of the library as well as in the evaluation of
the prototype solvers.
Objectives
Partners