partitioning scheme one can achieve scalable performance on lightly coupled distributed memory parallel systems. In particular, we demonstrate performance enhancement achieved by using the latency tolerant “selective inversion" scheme in the sparse triangular substitution phase of the computation.

We describe computation migration, a new technique that is based on compile-time program transformations, for accessing remote data in a distributed-memory parallel system. In contrast with RPC-style access, where the access is performed remotely, and with data migration, where the data is moved so

This paper proposes a distributed strategy for monitoring the messages density within a distributed memory parallel system. Each processor keeps track of the communication delays with its contiguous neighbours; when this time exceeds a certain average value, the data requests are not sent to those

and data partition-ing scheme one can achieve scalable performance on lightly coupled distributed memory parallel systems. In particular, we demonstrate performance enhancement achieved by using the latency tolerant “selective inver-sion " scheme in the sparse triangular substitution phase

this paper, we give a brief overview of the domain decomposition method used for the parallelisation of UIFS and present some performance results. The paper concludes with an examination of the load balancing difficulties presented by the inhomogeneous problem space typical of multi-physical models such as fluidstructure interaction and metals casting. * Sponsored by the Engineering and Physical Science Research Council. 2. PARALLELISATION BY DOMAIN DECOMPOSITION 2.1 Objectives

. The inherent difficulties of benchmarking parallel computers and the motivation behind the development of the GENESIS benchmark suite introduce the paper. Some of the broader issues that must be taken into account with any rigorous benchmarking effort on a distributed memory system

parallel applications and porting of legacy codes to parallel architectures have not yet gained a broad acceptance and diffusion. Common objections claim difficulty of performance tuning, limitation of its application to regular, data parallel computations, and lack of robustness of parallelizing

This paper describes the Global Trees (GT) system that provides a multi-layered interface to a global address space view of distributed tree data structures, while providing scalable performance on distributed memory systems. The Global Trees system utilizes coarse-grained data movement to enhance

Executing multi-threaded programs on multiprocessors can exploit the inherent paral-lelism among threads. However I the load imbalance and frequent remote data access factors may degrade the performance in distributed-memory parallel systems. To cope with these problems, one of the solutions

Executing multi-threaded programs on multiprocessors can exploit the inherent parallelism among threads. However, the load imbalance and frequent remote data access factors may degrade the performance in distributed-memory parallel systems. To cope with these problems, one of the solutions