A Comparison of Different Parallel Techniques Applied to the Solution of the Navier-Stokes Equations

摘要

The increasingly widespread availability of multi-processor computers has resulted in interest to design software that can make a full use of them. This is specially important in research and engineering applications, which typically require significant computing power.rnIn this paper we present a parallel finite element solver for the Navier-Stokes equations, based on algebraic subgrid scale stabilization and using a generalized Newmark scheme for time iteration. This solver is implemented using two different parallel strategies, OpenMP and MPI, and used to solve a practical example. We are interested in the performance of each parallel implementation, which we will test by solving a practical example.