Actual systems supporting dynamic load distribution for parallel programs such as Dynamo [3] manage load distribution by directing the assignment of newly generated subproblems to processes of the parallel program (non-preemptive load distribution). This so called task farming paradigm is only useful for a relatively small number of applications. Most numerical algorithms and simulation program...

Many high speed networks have been developed that may be suitable for parallel computing on clusters of workstations. This paper compares three different networks: FastEthernet, ATM, and Myrinet. We have implemented the Panda portability layer on all three networks, using the same host machines and as much the same software as possible. We compare the latency and throughput for Panda’s point-to...

Cellular computing is a special approach to massive parallelism. While the direct implementation of really large cellular computers is limited by technological constraints, good results can be obtained by simulation. This paper presents an example for such a simulation. By using a cluster of interconnected workstations the time consumed when simulating fairly large fields can be shortened. The ...

A parallel SAR processor is presented in this paper. The target configuration is a cluster of UNIX workstations, available in most user sites. This fact allows to obtain an increased computing performance without the need of dedicated hardware investment.

Networks of workstations are becoming a popular platform for parallel computing. Asynchronous Transfer Mode (ATM) network connections can improve the communication characteristics of these parallel clusters. This paper introduces a new class of topologies for constructing multicomputers called Small ATM Switch based Interconnection Networks (SASIN). The Round Table Network (one kind of SASIN) i...

—A natural approach for scheduling tasks to a workstation cluster is to employ the multiple machines in the cluster to schedule the task graphs so that the cluster manifests itself as a “self-scheduled” platform. A few parallel approaches have been devised for scheduling task graphs using a parallel machine such as an Intel Paragon but they are not suitable for a cluster of workstations environ...

The paper investigates and compares skeleton-based Eden implementations of different FFT-algorithms on workstation clusters with distributed memory. Our experiments show that the basic divide-andconquer versions suffer from an inherent input distribution and result collection problem. Advanced approaches like calculating FFT using a parallel map-and-transpose skeleton provide more flexibility t...

We describe the PCN programming system, focusing on those features designed to improve the productivity of scientists and engineers using parallel supercomputers. These features include a simple notation for the concise speciication of concurrent algorithms, the ability to incorporate existing Fortran and C code into parallel applications , facilities for reusing parallel program components, a ...

We describe the design and implementation of the Distributed Threads System (DTS), a programming environment for the paralleliza-tion of irregular and highly data-dependent algorithms. DTS extends the support for fork/join parallel programming from shared memory threads to a distributed memory environment. It is currently implemented on top of PVM, adding an asynchronous RPC abstraction and tur...

In this paper, we describe an extension of Common Lisp which allows the deeni-tion of parallel programs within that functional and object-oriented language. In particular, the extensions are the introducing of active objects, sending synchronous and asynchronous messages between them, automatic and manual distribution of active objects to object spaces, and transparent object managing. With the...