D. Logan

Large-scale computations on a scalar, vector and parallel ‘supercomputer’

Abstract We discuss two experimental parallel computer systems 1CAP-1 and 1CAP-2 which can be applied to the entire spectrum of scientific and engineering applications. These systems achieve ‘supercomputer’ levels of performance by spreading large scale computations across multiple cooperating processors—several with vector capabilities. We outline system hardware and software, and discuss our programming strategy for migrating codes from a conventional sequential system to a parallel one. The performance of a variety of applications programs is analyzed to demonstrate the merits of this approach. Finally, we discuss 1CAP-3, an extension to this computing system, which has been recently assembled.

ICAP/3090: parallel processing for large-scale scientific and engineering problems

Described is the lCAP/3090 (for loosely coupled array of processors) parallel processing system. General parallel processing performance issues that determine the success of all multiple-instruction/multiple-data-stream parallel computing systems are examined in the context of large-scale scientific and engineering problems. Experiments with previous lCAP parallel pro-cessing systems that have made possible the present design of lCAP/3090 are also described.

Solution of Large Scale Engineering Problems Using a Loosely Coupled Array of Processors

Our interest in parallel processing systems has been motivated by the need to solve large scale computational problems occuring in theoretical chemistry, biophysics, and more recently in engineering. We have been experimenting with a loosely coupled system, that uses a master processor to control a group of independent slave processors. This architecture was designed to meet the following objectives: the initial system 1) should be classifiable as a “supercomputer”, 2) be easily extended to higher computing speeds, 3) be flexible and fault tolerant, 4) permit easy migration of large codes to parallel execution 5) be relatively inexpensive.

Large Scale Computations on the Loosely Coupled Array of Processors

An experimental parallel computer system which is expected to achieve supercomputing performance across the entire spectrum of scientific and engineering applications is described. This system allows the execution of single large scale scientific and engineering applications on multiple processors. The system hardwares and softwares will be briefly described, as well as the programming strategies used to migrate codes from sequential to parallel. The validity of this approach to solving large scale problems is verified by analyzing the performance results of a variety of application programs. The type of scientific/engineering applications which may be investigated using this type of system is demonstrated by discussing one of our applications in biochemistry; namely the statistical and quantum mechanical study of DNA. Finally, ongoing and future extensions to this system are presented.