Reduced Parallel Multigrid : Analysis
نویسنده
چکیده
The most common technique for the parallelization of multigrid methods is grid partitioning. For such methods Brandt and Diskin have suggested the use of a variant of segmental reenement in order to reduce the amount of inter{processor communication. A parallel multigrid method with this technique avoids all communication on the nest grid levels. This article will examine some features of this class of algorithms as compared to standard parallel multigrid methods. In particular, the communication pattern will be analysed in detail.
منابع مشابه
A geometric data structure for parallel finite elements and the application to multigrid methods with block smoothing
We present a parallel data structure which is directly linked to geometric quantities of an underlying mesh and which is well adapted to the requirements of a general finite element realization. In addition, we define an abstract linear algebra model which supports multigrid methods (extending our previous work in Comp. Vis. Sci. 1 (1997) 27-40). Finally, we apply the parallel multigrid precond...
متن کاملParallel Performance of Coupled Multigrid Methods on the Parallel Performance of Coupled Multigrid Methods for the Solution of the Incompressible Navier{stokes Equations
This paper presents a detailed numerical study of the parallel performance of coupled multigrid V{, F{, and W{cycle for the solution of the incompressible Navier{Stokes equations. The results show the considerable innuence of the number of processors on a good choice of components and parameters of the algorithms. Some general suggestions concerning this choice can be derived from the tests.
متن کاملLoad Balancing for Adaptive Multigrid Methods
This paper presents two algorithms solving the load balancing problem arising in a data parallel implementation of multigrid methods on unstructured, locally refined meshes. The differences between additive and multiplicative multigrid and their influence on the load balancing procedure are discussed in detail. The quality of the proposed algorithms is assessed by numerical experiments on sever...
متن کاملThe Parallelization of an Adaptive Multigrid Eigenvalue Solver with LPARX
We have developed a parallel adaptive eigenvalue solver and applied it to a model problem in theoretical materials science. Our method combines adaptive mesh reenement techniques with a novel multigrid eigenvalue algorithm. By exploiting adaptivity, we have reduced computation time and memory consumption by more than two orders of magnitude. We have implemented our solver using the LPARX parall...
متن کاملThree-Dimensional High-Lift Analysis Using a Parallel Unstructured Multigrid Solver
A directional implicit unstructured agglomeration multigrid solver is ported to shared and distributed memory massively parallel machines using the explicit domain-decomposition and message-passing approach. Because the algorithm operates on local implicit lines in the unstructured mesh, special care is required in partitioning the problem for parallel computing. A weighted partitioning strateg...
متن کامل