Berkeley Lab - Scientific Computing Seminar

Date:
Friday, December 5, 2008
Time:
1:00pm-2:00pm  
Location:
Building 50A, 5132 Conference Room
Seminar Speaker:
Azzam Haidar
University of Toulouse, INPT-ENSEEIHT, France
Title:
Parallel Hybrid Solver for Large 3D Problems
Abstract:
Large-scale scientific simulations are nowadays fully integrated in many scientific and industrial applications. Many of these simulations rely on modelisations based on PDEs that lead to the solution of huge linear or nonlinear systems of equations involving millions of unknowns. In that context the use of large high performance computers in conjunction with advanced fully parallel and scalable numerical techniques is mandatory to efficiently tackle these problems.

In this talk we consider the parallel scalability of variants of an algebraic additive Schwarz preconditioner for the solution of large 3D problems in a non-overlapping domain decomposition framework. To alleviate the computational cost, both in terms of memory and floating-point complexity, we investigate variants based on a sparse approximation or on mixed 32- and 64-bit calculation. The robustness and the scalability of the preconditioners are investigated through extensive parallel experiments on up to two thousand processors.

We represent also an approach based on two levels of parallelism, that offers the flexibility to combine the numerical and the parallel implementation scalabilities. The combination of the two levels of parallelism enables an optimal usage of the computing resource while preserving attractive numerical performance. Consequently such a numerical technique appears as a promising candidate for intensive simulations on massively parallel platforms.

The robustness and parallel numerical performance of the solver is reported on large challenging linear systems arising from the structural mechanics and seismic modeling applications.

(This is joint work with Luc Giraud.)
Sponsor of Seminar:
Sherry Li
Scientific Computing

Contact Esmond G. Ng EGNg@lbl.gov