Mark IIIfp hypercube concurrent processor architecture

  • Authors:
  • J. Tuazon;J. Peterson;M. Pniel

  • Affiliations:
  • Jet Propulsion Laboratory, Pasadena, CA and Professor of Electrical Engineering, California State University, Fullerton, CA;Jet Propulsion Laboratory, Pasadena, CA and Member of the Technical Staff, Concurrent Processor Systems Group, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA;Jet Propulsion Laboratory, Pasadena, CA and Member of the Technical Staff, Concurrent Processor Systems Group, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA

  • Venue:
  • C3P Proceedings of the third conference on Hypercube concurrent computers and applications: Architecture, software, computer systems, and general issues - Volume 1
  • Year:
  • 1988
  • The cosmic cube

    Communications of the ACM - Special section on computer architecture

Quantified Score

Hi-index 0.00

Visualization

Abstract

The Mark lllfp Hypercube is a new generation of hypercube concurrent processor system developed at JPL/Caltech, with peak performance of 5 Mips, 14 Mflops per node, and a peak communication rate of 6 Mbytes per second. Each node utilizes two Motorola MC68020 microprocessors, an MC68882 scalar floating- point coprocessor, and a Weitek 8000 floating-point chip set. One of the MC68020 processors serves as the application and computational processor, the other is dedicated to communication. The three processors are interconnected through a common system bus and share 4 Mbytes of dynamic memory. Each processor has his own fast local memory, which increases parallelism, minimizes shared memory referencing, reduces memory contentions and improves system performance.