Reconfiguration Schemes for Fault-Tolerant Processor Arrays

Abstract

This project addressed several aspects of the problem of designing highly-reliable dynamically reconfigurable processor arrays. The proposed work focused mainly on reconfiguration schemes required to implement fault-tolerant processor arrays. According to the original statement of work, the following complementary objectives were pursued: (1) a methodology for the design and evaluation of processor-switched arrays. (2) a methodology for the design and evaluation of multi-level hierarchically reconfigurable processor arrays. (3) a methodology for the design of fault-tolerant interconnection routers for processor arrays with decentralized routing control, and (4) algorithm reconfiguration strategies which, together with hardware reconfiguration schemes, can be used to achieve graceful degradation in processor arrays. The emphasis of the proposed research was on the development of optimal reconfiguration schemes for each of the above objectives by using mathematical and simulation tools. For this purpose, evaluation methods and adequate measures were also studied and developed. These measures include not only reliability but also joint measures of performance, hardware area and reliability.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 15, 1992
Accession Number
ADA257833

Entities

People

  • Jose A. Fortes

Organizations

  • Purdue University

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Computer Programming
  • Computer Programs
  • Computers
  • Electrical Engineering
  • Integer Programming
  • Linear Programming
  • Mathematical Models
  • Numbers
  • Operating Systems
  • Parallel Computing
  • Parallel Processing
  • Parallel Processors
  • Processing Equipment
  • Reliability
  • Simplex Method
  • Three Dimensional
  • Two Dimensional

Fields of Study

  • Engineering

Readers

  • Parallel and Distributed Computing.
  • Systems Analysis and Design