Recent Advances in Parallel Virtual Machine and Message Passing Interface: 7th European PVM/MPI Users’ Group Meeting Balatonfüred, Hungary, September 10–13, 2000 Proceedings

Parallel Virtual Machine (pvm) And Message Passing Interface (mpi) Are The Most Frequently Used Tools For Programming According To The Message Passing Paradigm, Which Is Considered One Of The Best Ways To Develop Parallel Applications. This Volume Comprises 42 Revised Contributions Presented At The Seventh European Pvm/mpi Users’ Group Meeting, Which Was Held In Balatonfr Ed, Hungary, 10 13 September 2000. The Conference Was Organized By The Laboratory Of Parallel And Distributed Systems Of The Computer And Automation Research Institute Of The Hungarian Academy Of Sciences. This Conference Was Previously Held In Barcelona, Spain (1999), Liverpool, Uk (1998) And Cracow, Poland (1997). The First Three Conferences Were Devoted To Pvm And Were Held At The Technische Universit T M Nchen, Germany (1996), Ecole Normale Superieure Lyon, France (1995), And University Of Rome, Italy (1994). This Conference Has Become A Forum For Users And Developers Of Pvm, Mpi, And Other Message Passing Environments. Interaction Between Those Groups Has Proved To Be Very Useful For Developing New Ideas In Parallel Computing And For Applying Existing Ideas To New Practical Fields. The Main Topics Of The Meeting Were Evaluation And Performance Of Pvm And Mpi, Extensions And Improvements To Pvm And Mpi, Algorithms Using The Message Passing Paradigm, And Applications In Science And Engineering Based On Message Passing. The Conference Included Four Tutorials And Five Invited Talks On Advances In Mpi, Cluster Computing, Network Computing, Grid Computing, And Sgi Parallel Computers And Programming Systems. Invited Speakers -- Pvm And Mpi: What Else Is Needed For Cluster Computing? -- Managing Your Workforce On A Computational Grid -- Isolating And Interfacing The Components Of A Parallel Computing Environment -- Symbolic Computing With Beowulf-class Pc Clusters -- High Speed Networks For Clusters, The Bip-myrinet Experience -- Evaluation And Performance -- A Benchmark For Mpi Derived Datatypes -- Working With Mpi Benchmarking Suites On Ccnuma Architectures -- Performance Measurements On Dynamite/dpvm -- Validation Of Dimemas Communication Model For Mpi Collective Operations -- Automatic Performance Analysis Of Master/worker Pvm Applications With Kpi -- Mpi Optimization For Smp Based Clusters Interconnected With Sci -- Algorithms -- Parallel, Recursive Computation Of Global Stability Charts For Liquid Bridges -- Handling Graphs According To A Coarse Grained Approach: Experiments With Pvm And Mpi -- Adaptive Multigrid Methods In Mpi --^ Multiple Parallel Local Searches In Global Optimization -- Towards Standard Nested Parallelism -- Pipeline Algorithms On Mpi: Optimal Mapping Of The Path Planing Problem -- Use Of Pvm For Map Image Restoration: A Parallel Implementation Of The Artur Algorithm -- Parallel Algorithms For The Least-squares Finite Element Solution Of The Neutron Transport Equation -- Extensions And Improvements -- Gamma And Mpi/gamma On Gigabit Ethernet -- Distributed Checkpointing Mechanism For A Parallel File System -- Thread Communication Over Mpi -- A Simple, Fault Tolerant Naming Space For The Harness Metacomputing System -- Runtime Checking Of Datatype Signatures In Mpi -- Implementation Issues -- A Scalable Process-management Environment For Parallel Programs -- Single Sided Communications In Multi-protocol Mpi -- Mpi-2 Process Creation & Management Implementation For Nt Clusters -- Composition Of Message Passing Applications On-demand -- Heterogeneous Distributed Systems --^ An Architecture Of Stampi: Mpi Library On A Cluster Of Parallel Computers -- Integrating Mpi Components Into Metacomputing Applications -- Tools -- Pvmaple: A Distributed Approach To Cooperative Work Of Maple Processes -- Cis - A Monitoring System For Pc Clusters -- Monito: A Communication Monitoring Tool For A Pvm-linux Environment -- Interoperability Of Ocm-based On-line Tools -- Parallel Program Model For Distributed Systems -- Translation Of A High-level Graphical Code To Message-passing Primitives In The Grade Programming Environment -- The Transition From A Pvm Program Simulator To A Heterogeneous System Simulator: The Hesse Project -- Comparison Of Different Approaches To Trace Pvm Program Execution -- Applications In Science And Engineering -- Scalable Cfd Computations Using Message-passing And Distributed Shared Memory Algorithms -- Parallelization Of Neural Networks Using Pvm -- Parallel Dsir Text Indexing System: Using Multiple Master/slave Concept --^ Improving Optimistic Pdes In Pvm Environments -- Use Of Parallel Computers In Neurocomputing -- A Distributed Computing Environment For Genetic Programming Using Mpi -- Experiments With Parallel Monte Carlo Simulation For Pricing Options Using Pvm -- Time Independent 3d Quantum Reactive Scattering On Mimd Parallel Computers -- Ft-mpi: Fault Tolerant Mpi, Supporting Dynamic Applications In A Dynamic World -- Acct: Automatic Collective Communications Tuning. Jack Dongarra, Peter Kacsuk, Norbert Podhorszki (eds.). Includes Bibliographical References And Index. PVM and MPI: What Else Is Needed for Cluster Computing?....Pages 1-2 Managing Your Workforce on a Computational Grid....Pages 3-4 Isolating and Interfacing the Components of a Parallel Computing Environment....Pages 5-6 Symbolic Computing with Beowulf-Class PC Clusters....Pages 7-8 High Speed Networks for Clusters, the BIP-Myrinet Experience....Pages 9-9 A Benchmark for MPI Derived Datatypes....Pages 10-17 Working with MPI Benchmarking Suites on ccNUMA Architectures....Pages 18-26 Performance Measurements on Dynamite/DPVM....Pages 27-38 Validation of Dimemas Communication Model for MPI Collective Operations....Pages 39-46 Automatic Performance Analysis of Master/Worker PVM Applications with Kpi....Pages 47-55 MPI Optimization for SMP Based Clusters Interconnected with SCI....Pages 56-63 Parallel, Recursive Computation of Global Stability Charts for Liquid Bridges....Pages 64-71 Handling Graphs According to a Coarse Grained Approach: Experiments with PVM and MPI....Pages 72-79 Adaptive Multigrid Methods in MPI....Pages 80-87 Multiple Parallel Local Searches in Global Optimization....Pages 88-95 Towards Standard Nested Parallelism....Pages 96-103 Pipeline Algorithms on MPI: Optimal Mapping of the Path Planing Problem....Pages 104-112 Use of PVM for MAP Image Restoration: A Parallel Implementation of the ARTUR Algorithm....Pages 113-120 Parallel Algorithms for the Least-Squares Finite Element Solution of the Neutron Transport Equation....Pages 121-128 GAMMA and MPI/GAMMA on Gigabit Ethernet....Pages 129-136 Distributed Checkpointing Mechanism for a Parallel File System....Pages 137-144 Thread Communication over MPI....Pages 145-151 A Simple, Fault Tolerant Naming Space for the HARNESS Metacomputing System....Pages 152-159 Runtime Checking of Datatype Signatures in MPI....Pages 160-167 A Scalable Process-Management Environment for Parallel Programs....Pages 168-175 Single Sided Communications in Multi-protocol MPI....Pages 176-183 MPI-2 Process Creation & Management Implementation for NT Clusters....Pages 184-191 Composition of Message Passing Applications On-Demand....Pages 192-199 An Architecture of Stampi: MPI Library on a Cluster of Parallel Computers....Pages 200-207 Integrating MPI Components into Metacomputing Applications....Pages 208-215 PVMaple: A Distributed Approach to Cooperative Work of Maple Processes....Pages 216-224 CIS - A Monitoring System for PC Clusters....Pages 225-232 Monito: A Communication Monitoring Tool for a PVM-Linux Environment....Pages 233-241 Interoperability of OCM-Based On-Line Tools....Pages 242-249 Parallel Program Model for Distributed Systems....Pages 250-257 Translation of a High-Level Graphical Code to Message-Passing Primitives in the GRADE Programming Environment....Pages 258-265 The Transition from a PVM Program Simulator to a Heterogeneous System Simulator: The HeSSE Project....Pages 266-273 Comparison of Different Approaches to Trace PVM Program Execution....Pages 274-281 Scalable CFD Computations Using Message-Passing and Distributed Shared Memory Algorithms....Pages 282-288 Parallelization of Neural Networks Using PVM....Pages 289-296 Parallel DSIR Text Indexing System: Using Multiple Master/Slave Concept....Pages 297-303 Improving Optimistic PDES in PVM Environments....Pages 304-312 Use of Parallel Computers in Neurocomputing....Pages 313-321 A Distributed Computing Environment for Genetic Programming Using MPI....Pages 322-329 Experiments with Parallel Monte Carlo Simulation for Pricing Options Using PVM....Pages 330-337 Time Independent 3D Quantum Reactive Scattering on MIMD Parallel Computers....Pages 338-345 FT-MPI: Fault Tolerant MPI, Supporting Dynamic Applications in a Dynamic World....Pages 346-353 ACCT: Automatic Collective Communications Tuning....Pages 354-361 Parallel Virtual Machine (PVM) and Message Passing Interface (MPI) are the most frequently used tools for programming according to the message passing paradigm, which is considered one of the best ways to develop parallel applications. This volume comprises 42 revised contributions presented at the Seventh European PVM/MPI Users{u2019} Group Meeting, which was held in Balatonfr ed, Hungary, 10 13 September 2000. The conference was organized by the Laboratory of Parallel and Distributed Systems of the Computer and Automation Research Institute of the Hungarian Academy of Sciences. This conference was previously held in Barcelona, Spain (1999), Liverpool, UK (1998) and Cracow, Poland (1997). The first three conferences were devoted to PVM and were held at the Technische Universit t M nchen, Germany (1996), Ecole Normale Superieure Lyon, France (1995), and University of Rome, Italy (1994). This conference has become a forum for users and developers of PVM, MPI, and other message passing environments. Interaction between those groups has proved to be very useful for developing new ideas in parallel computing and for applying existing ideas to new practical fields. The main topics of the meeting were evaluation and performance of PVM and MPI, extensions and improvements to PVM and MPI, algorithms using the message passing paradigm, and applications in science and engineering based on message passing. The conference included four tutorials and five invited talks on advances in MPI, cluster computing, network computing, grid computing, and SGI parallel computers and programming systems This book constitutes the refereed proceedings of the 7th European PVM/MPI Users' Group Meeting held in Balatonfred, Hungary in September 2000. The 35 revised papers and seven posters presented were carefully reviewed and selected for inclusion in the book. The book offers topical sections on evaluation and performance, algorithms, extensions and improvements, implementation, heterogenous distributed systems, tools, and applications in science and engineering. All current aspects of the parallel virtual machine approach and the message passing interface paradigm are addressed