Technische Universität Chemnitz Sonderforschungsbereich 393 Numerische Simulation auf massiv parallelen Rechnern
نویسندگان
چکیده
The characteristics of irregular algorithms make a parallel implementation difficult, especially for PC clusters or clusters of SMPs. These characteristics may include an unpredictable access behavior to dynamically changing data structures or strong irregular coupling of computations. Problems are an unknown load distribution and expensive irregular communication patterns for data accesses and redistributions. Thus the parallel implementation of irregular algorithms on distributed memory machines and clusters requires a special organizational mechanism for a dynamic load balance while keeping the communication and administration overhead low. In this paper we propose task pool teams for implementing irregular algorithms on clusters of PCs or SMPs. A task pool team combines multithreaded programming using task pools on single nodes with explicit message passing between different nodes. The dynamic load balance mechanism of task pools is generalized to a dynamic load balance scheme for all distributed nodes. We have implemented and compared several versions for task pool teams. As application example, we use the hierarchical radiosity algorithm, which is based on dynamically growing quadtree data structures annotated by varying interaction lists expressing the irregular coupling between the quadtrees. Experiments are performed on a PC cluster and a cluster of SMPs.
منابع مشابه
Technische Universität Chemnitz Sonderforschungsbereich 393 Numerische Simulation auf massiv parallelen Rechnern
The paper is concerned with algorithms for transforming hexahedral finite element meshes into tetrahedral meshes without introducing new nodes. Known algorithms use only the topological structure of the hexahedral mesh but no geometry information. The paper provides another algorithm which is then extented such that quality criteria for the splitting of faces are respected.
متن کاملNumerische Simulation Auf Massiv Parallelen Rechnern Preprint-reihe Des Chemnitzer Sfb 393
We study classical control problems like pole assignment, stabilization, linear quadratic control and H1 control from a numerical analysis point of view. We present several examples that show the di culties with classical approaches and suggest reformulations of the problems in a more general framework. We also discuss some new algorithmic approaches. AMS subject classi cation. 65F15, 93B40, 93...
متن کاملTechnische Universität Chemnitz Sonderforschungsbereich 393 Numerische Simulation auf massiv parallelen Rechnern
In this document we present the architectural design we develop for the system software, which integrates our new architecture PCI–SCI Bridge with VIA support into the LINUX operating system. The development of this system software is a part of a bigger project, the end goal of which is to build a complete interconnect solution from the hardware to communication libraries and to prove in practi...
متن کاملTechnische Universit at ChemnitzSonderforschungsbereich 393
The recursive inverse eigenvalue problem for matrices is studied, where for each leading principle submatrix an eigenvalue and associated left and right eigenvectors are assigned. Existence and uniqueness results as well as explicit formulas are proven, and applications to nonnegative matrices, Zmatrices, M -matrices, symmetric matrices, Stieltjes matrices and inverse M -matrices are considered...
متن کامل