PORTING INDUSTRIAL CODES AND DEVELOPING SPARSE LINEAR SOLVERS ON PARALLEL COMPUTERS

We address the main issues when porting existing codes from serial to parallel computers and when developing portable parallel software on M IMD multiprocessors (shared memory, virtual shared memory, and distrib uted memory multiprocessors, and networks of computers). We discuss th e use of numerical libraries as a way of developing portable and effic ient parallel code. We illustrate this by using examples from our expe rience in porting industrial codes and in designing parallel numerical libraries. We report in some detail on the parallelization of scienti fic applications coming from Centre National d'Etudes Spatiales and fr om Aerospatiale, and we illustrate how it is possible to develop porta ble and efficient numerical software by considering the parallel solut ion of sparse linear systems of equations.