Design and Implementation of a Parallel Iterative Linear System Solver with Dynamic Load Balancing
نویسنده
چکیده
This report describes the design and implementation of the parallel iterative linear system solver PAISS (Parallel Adaptive Iterative linear System Solver) for distributed memory multicomputers and workstation clusters. It is capable to apply a heterogeneous data distribution and dynamic load balancing within an iterative solver routine at matrix level. Sparse and dense matrices, as well as vectors are distributed heterogeneously according to the available performances of the processors, and redistribution is carried out at run time if the load of the processors changes. The concepts behind the chosen matrix data structures and the load measurement are discussed, and the dynamic load balancing algorithm is presented in details. Important aspects of the implementation are given, and the report completes with an overview of all implemented modules and data structures.
منابع مشابه
Parallel Dynamic Load-Balancing for the Solution of Transient CFD Problems Using Adaptive Tetrahedral Meshes
This paper considers a new parallel dynamic load-balancing algorithm which has been developed for use in conjunction with an unstructured tetrahedral parallel adaptive solver for transient ow problems. A brief description of the 3-d adaptivity algorithm is then followed by a discussion of the load-balancing problem. The practical load-balancing algorithm is then explained, followed by a discuss...
متن کاملSolving Large Nonsymmetric Sparse Linear Systems Using MCSPARSE
In this paper, the methods and implementation techniques used for the nonsymmetric sparse linear system solver, MCSPARSE, on the Cedar system are described. A novel reordering scheme (H*) upon which the solver is based is presented. The tradeoffs discussed include stability and fill-in control, hierarchical parallelism, and load balancing. Experimental results demonstrating the effectiveness of...
متن کاملParleda: a Library for Parallel Processing in Computational Geometry Applications
ParLeda is a software library that provides the basic primitives needed for parallel implementation of computational geometry applications. It can also be used in implementing a parallel application that uses geometric data structures. The parallel model that we use is based on a new heterogeneous parallel model named HBSP, which is based on BSP and is introduced here. ParLeda uses two main lib...
متن کاملLevenberg Marquardt ( LM ) Algorithm 1 –
1 – Introduction Parameter estimation for function optimization is a well established problem in computing, as there are countless applications in practice. For this work, we will focus specifically in implementing a distributed and parallel implementation of the Levenberg Marquardt algorithm, which is a well established numerical solver for function approximation given a limited data set. Para...
متن کاملA comparison of some dynamic load-balancing algorithms for a parallel adaptive flow solver
In this paper we contrast the performance of three different parallel dynamic load-balancing algorithms when used in conjunction with a particular parallel, adaptive, time-dependent, 3-d ow solver that has recently been developed at Leeds. An overview of this adaptive solver is given along with a description of a new dynamic load-balancing algorithm. The eeectiveness of this algorithm is then a...
متن کامل