The design of general-purpose dynamic load-balancing tools for parallel applications is more challenging than the design of static partitioning tools. Both algorithmic and software engineering issues arise. The authors have addressed many of these issues in the design of the Zoltan dynamic load-balancing library. Zoltan has an object-oriented interface that makes it easy to use and provides separation between the application and the load-balancing algorithms. It contains a suite of dynamic load-balancing algorithms, including both geometric and graph-based algorithms. Its design makes it valuable both as a partitioning tool for a variety of applications and as a research test-bed for new algorithmic development. In this paper, the authors describe Zoltan's design and demonstrate its use in an unstructured-mesh finite element application.

There are many scientific applications for which the computational load varies throughout the execution and causes uneven distribution of workload during run-time. One such class of applications is Adaptive Mesh Refinement (AMR) applications. AMR is a type of multiscale algorithm that achieves high resolution in localized regions of dynamic, multidimensional numerical simulations. A typical AMR application may require enormous computing resources, which usually cannot be satisfied by a single-processor machine, thereby requiring parallel and distributed systems. One of the key issues related to AMR is dynamic load balancing (DLB), which allows large-scale adaptive applications to run efficiently on parallel and distributed systems. In investigating DLB schemes, we first complete a detailed analysis of structured AMR (SAMR) applications, identifying the unique characteristics that impose severe challenges on DLB schemes. The results indicate that most of the available DLB schemes are not appropriate for SAMR applications due to their unique adaptive characteristics. Thus, we propose a novel dynamic load balancing scheme for SAMR applications on parallel systems (denoted as parallel DLB). It integrates a grid-splitting technique with direct grid movements, for which the objective is to reduce the parallel execution time. Further, our experiment shows that simply moving a DLB scheme designed for parallel systems to distributed systems will introduce significant overhead. Therefore, we propose a framework for dynamic load balancing on distributed systems (denoted as distributed DLB). It takes into consideration: (1) heterogeneity of processors, (2) heterogeneity of networks, (3) shared nature of networks, and (4) adaptive characteristics of the applications. For SAMR applications, the distributed DLB incorporates the proposed parallel DLB during the load balancing process. Both parallel DLB and distributed DLB were implemented in the ENZO code, a parallel implementation of SAMR in astrophysics and cosmology. Experiments show that the proposed DLB schemes can significantly improve the performance of SAMR applications on both parallel and distributed systems in terms of the total execution time and the quality of load balancing.

This millennium will see the increased use of parallel computing technologies at all levels of mainstream computing. Most computer hardware will use these technologies to achieve higher computing speeds, high speed access to very large distributed databases and greater flexibility through heterogeneous computing. These developments can be expected to result in the extended use of all types of parallel computers in virtually all areas of human endeavour. Compute-intensive problems in emerging areas such as financial modelling and multimedia systems, in addition to traditional application areas of parallel computing such as scientific computing and simulation, will stimulate the developments. Parallel computing as a field of scientific research and development will move from a niche concentrating on solving compute-intensive scientific and engineering problems to become one of the fundamental computing technologies.This book gives a retrospective view of what has been achieved in the parallel computing field during the past three decades, as well as a prospective view of expected future developments./a

This volume contains papers presented at the NATO sponsored Advanced Research Workshop on "Software for Parallel Computation" held at the University of Calabria, Cosenza, Italy, from June 22 to June 26, 1992. The purpose of the workshop was to evaluate the current state-of-the-art of the software for parallel computation, identify the main factors inhibiting practical applications of parallel computers and suggest possible remedies. In particular it focused on parallel software, programming tools, and practical experience of using parallel computers for solving demanding problems. Critical issues relative to the practical use of parallel computing included: portability, reusability and debugging, parallelization of sequential programs, construction of parallel algorithms, and performance of parallel programs and systems. In addition to NATO, the principal sponsor, the following organizations provided a generous support for the workshop: CERFACS, France, C.I.R.A., Italy, C.N.R., Italy, University of Calabria, Italy, ALENIA, Italy, The Boeing Company, U.S.A., CISE, Italy, ENEL - D.S.R., Italy, Alliant Computer Systems, Bull RN Sud, Italy, Convex Computer, Digital Equipment Corporation, Rewlett Packard, Meiko Scientific, U.K., PARSYTEC Computer, Germany, TELMAT Informatique, France, Thinking Machines Corporation.

This volume gives an overview of the state-of-the-art with respect to the development of all types of parallel computers and their application to a wide range of problem areas. The international conference on parallel computing ParCo97 (Parallel Computing 97) was held in Bonn, Germany from 19 to 22 September 1997. The first conference in this biannual series was held in 1983 in Berlin. Further conferences were held in Leiden (The Netherlands), London (UK), Grenoble (France) and Gent (Belgium). From the outset the aim with the ParCo (Parallel Computing) conferences was to promote the application of parallel computers to solve real life problems. In the case of ParCo97 a new milestone was reached in that more than half of the papers and posters presented were concerned with application aspects. This fact reflects the coming of age of parallel computing. Some 200 papers were submitted to the Program Committee by authors from all over the world. The final programme consisted of four invited papers, 71 contributed scientific/industrial papers and 45 posters. In addition a panel discussion on Parallel Computing and the Evolution of Cyberspace was held. During and after the conference all final contributions were refereed. Only those papers and posters accepted during this final screening process are included in this volume. The practical emphasis of the conference was accentuated by an industrial exhibition where companies demonstrated the newest developments in parallel processing equipment and software. Speakers from participating companies presented papers in industrial sessions in which new developments in parallel computing were reported.