Vector and Parallel Processing - Vecpar'98: Third International Conference Porto, Portugal, June 21-23, 1998 Selected Papers and Invited Talks

ISBN-13: 9783540662280 / Angielski / Miękka / 1999 / 712 str.

Jose M. L. M. Palma; Jack Dongarra; Vicente Hernandez

Opis

This volume constitutes the refereed post-conference proceedings of the Third International Conference on Vector and Parallel Processing (VECPAR'98), held in Porto, Portugal, in June 1998. The 41 revised full papers presented were carefully selected during two rounds of reviewing and revision. Also included are six invited papers and introductory chapter surveys. The papers are organized into sections on: eigenvalue problems and solutions of linear systems; computational fluid dynamics, structural analysis and mesh partitioning; computing in education; computer organization, programming and benchmarking; image analysis and synthesis; parallel database servers; and nonlinear problems.

Spis treści

1: Eigenvalue Problems and Solution of Linear Systems.- to “Eigenvalue Problems and Solution of Linear Systems”.- Some Unusual Eigenvalue Problems.- Multi-sweep Algorithms for the Symmetric Eigenproblem.- A Unified Approach to Parallel Block-Jacobi Methods for the Symmetric Eigenvalue Problem.- Calculation of Lambda Modes of a Nuclear Reactor: A Parallel Implementation Using the Implicitly Restarted Arnoldi Method.- Parallel Jacobi-Davidson for Solving Generalized Eigenvalue Problems.- Parallel Preconditioned Solvers for Large Sparse Hermitian Eigenproblems.- Solving Eigenvalue Problems on Networks of Processors.- Solving Large-Scale Eigenvalue Problems on Vector Parallel Processors.- Direct Linear Solvers for Vector and Parallel Computers.- Parallel Preconditioners for Solving Nonsymmetric Linear Systems.- Synchronous and Asynchronous Parallel Algorithms with Overlap for Almost Linear Systems.- The Parallel Problems Server: A Client-Server Model for Interactive Large Scale Scientific Computation.- 2: Computational Fluid Dynamics, Structural Analysis and Mesh Partioning Techniques.- to “Computational Fluid Dynamics, Structural Analysis and Mesh Partioning Techniques”.- Parallel Domain-Decomposition Preconditioning for Computational Fluid Dynamics.- Influence of the Discretization Scheme on the Parallel Efficiency of a Code for the Modelling of a Utility Boiler.- Parallel 3D Airflow Simulation on Workstation Clusters.- Parallel Turbulence Simulation: Resolving the Inertial Subrange of the Kolmogorov Spectrum.- The Study of a Parallel Algorithm Using the Laminar Backward-Facing Step Flow as a Test Case.- A Low Cost Distributed System for FEM Parallel Structural Analysis.- Dynamic Load Balancing in Crashworthiness Simulation.- Some Concepts of the Software Package FEAST.- Multilevel Mesh Partitioning for Optimising Aspect Ratio.- 3: Computing in Education.- Parallel and Distributed Computing in Education (Invited Talk).- 4: Computer Organisation, Programming and Benchmarking.- to “Computer Organisation, Programming and Benchmarking Introduction”.- Reconfigurable Systems: Past and Next 10 Years.- A Systolic Algorithm for the Factorisation of Matrices Arising in the Field of Hydrodynamics.- Automatic Detection of Parallel Program Performance Problems.- Behavioural Analysis Methodology Oriented to Configuration of Parallel, Real-Time and Embedded Systems.- Spatial Data Locality with Respect to Degree of Parallelism in Processor-and-Memory Hierarchies.- Partitioning Regular Domains on Modern Parallel Computers.- New Access Order to Reduce Inter-Vector Conflicts.- Registers Size Influence on Vector Architectures.- Limits of Instruction Level Parallelism with Data Value Speculation.- High Performance Cache Management for Parallel File Systems.- Using Synthetic Workloads for Parallel Task Scheduling Improvement Analysis.- Dynamic Routing Balancing in Parallel Computer Interconnection Networks.- Algorithm-Dependant Method to Determine the Optimal Number of Computers in Parallel Virtual Machines.- Low Cost Parallelizing: A Way to be Efficient.- A Performance Analysis of the SGI Origin2000.- An ISA Comparison Between Superscalar and Vector Processors.- 5: Image, Analysis and Synthesis.- to “Image, Analysis and Synthesis”.- High Performance Computing for Image Synthesis.- Parallel Implementations of Morphological Connected Operators Based on Irregular Data Structures.- 6: Parallel Database Servers.- The Design of an ODMG Compatible Parallel Object Database Server.- 7: Nonlinear Problems.- to “Nonlinear Problems”.- A Parallel N-Body Integrator Using MPI.- A Parallelisation Strategy for Power Systems Composite Reliability Evaluation.- High Performance Computing of an Industrial Problem in Tribology.- Parallel Grid Manipulations in Earth Science Calculations.- Simulating Magnetised Plasma with the Versatile Advection Code.- Parallel Genetic Algorithms for Hypercube Machines.