Implementation Issues of Computational Fluid Dynamics Algorithms on Parallel Computers
In this paper we compare efficiency of three versions of a parallel algorithm for finite element compressible fluid flow simulations on unstructured grids. The first version is based on the explicit model of parallel programming with message-passing paradigm, the second is the implicit model in which data-parallel programming is used, while in the last we propose a heterogeneous model which combines these two models.
Time discretization of the compressible Euler equations is organized with a linear, implicit version of the Taylor-Galerkin time scheme, with finite elements employed for space discretization of one step problems. The resulting nonsymmetric system of linear equations is solved iteratively with the preconditioned GMRES method.
The algorithm has been tested on two parallel computers HP SPP1600 and HP S2000 using a benchmark problem of 2D inviscid flow simulations - the ramp problem.
KeywordsDomain Decomposition Computing Node Nite Element Compressible Euler Equation Step Problem
Unable to display preview. Download preview PDF.
- 1.G.F. Carey, A. Bose, B. Davis, C. Harle, and R. McLay, Parallel computation of viscous ows, in Proceedings of the 1997 Simulation Multiconference. High Performance Computing’ 97, 6–10 April 1997, Atlanta, USA.Google Scholar
- 2.G. Mahinthakumar, F. Saied, and A.J. Valocchi, Comparison of some parallel Krylov solvers for large scale ground-water contaminant transport simulations, in Proceedings of the 1997 Simulation Multiconference. High Performance Computing’ 97, 6–10 April 1997, Atlanta, USA.Google Scholar
- 3.Finite Element Method in Large-Scale Computational Fluid Dynamics, International Journal for Numerical Methods in Fluids, 21, (1995).Google Scholar
- 4.Idelsohn, S.R., Onate, E., Dvorkin, E.N. (eds.), Computational Mechanics. New Trends and Applications, Proc. of Fourth World Congress on Computat.Mechanics, June 29–July 2, 1998, Buenos Aires, CIMNE (Barcelona), 1998.Google Scholar
- 5.Pŀażek, J., Banaś, K., Kitowski, J., and Boryczko, K., Exploiting two-level parallelism in FEM applications, in: B. Hertzberger and P. Sloot, (eds.), Proc.HPCN’97, April 28–30, 1997, Vienna, Lecture Notes in Computer Science, Vol. 1225 (Springer, Berlin, 1997) 272–281.Google Scholar
- 6.P. LeTallec, Domain decomposition method in computational mechanics, in: J.T. Oden, ed. (North Holland, Amsterdam, 1994).Google Scholar
- 7.Pŀażek, J., Banaś, K., and Kitowski, J., Finite element message-passing/DSM simulation algorithm for parallel computers, in: P. Sloot, M. Bubak and B. Hertzberger, (eds.), Proc.HPCN’98, April 21–23, 1998, Amsterdam, Lecture Notes in Computer Science, Vol. 1401 (Springer, Berlin, 1998) 878–880.Google Scholar