ESPRIT ’90 pp 227-241 | Cite as

Parallel Associative Development Machine as a Vehicle for Artificial Intelligence

  • Patricia Guichard-Jary


This ESPRIT Project, started in 1986, is a cooperative effort between THOMSON-CSF (prime contractor — France), GEC (United Kingdom), and FIRST (Greece). PROLOGIA (France) and NSL (France) are also involved as sub-contractors.

Its objective is to develop a machine for symbolic and time critical applications, such as image recognition, natural language, and parallel expert systems. PADMAVATI is a MIMD machine connected to a SUN host. It is composed of 16 Inmos T800 transputers, with a 16Mbyte DRAM. A Content Addressable Memory with a capacity of 48kbytes (up to 680kbytes) is also attached to the memory bus of each transputer and is especially usefull for symbolic languages. A dynamic Network and a static ring connecting the 16 transputers provide non local communications. Symbolic languages, such as Lisp, Prolog, extended with parallelism and associativity mechanisms, have been ported on this architecture.

In spite of the fact that PADMAVATI is a transputer based architecture, it has innovative features that no other current transputer machine has, some of which will be incorporated in the next generation transputer architecture which INMOS plan to release in two years time. Thus, this project can considered as an early prototype for the next generation.


Associative Memory Message Passing Processing Node Virtual Channel Symbolic Language 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Bodeveix J.P, “LOGARITHM: Un modèle de Prolog Parallèle. Son implémentation sur transputers”, Thèse Docteur en Sciences. Université Paris-Sud. Janvier 1989Google Scholar
  2. [2]
    Bruynooghe M., Pereira L.M, “Deduction Revision by Intelligent Backtracking” Implementation of Prolog, ed. Campbell, 1984Google Scholar
  3. [3]
    “Prolog syntax: draft 5.0 and comments”, Report BSI PS/239 (or ISO N9), march 1988Google Scholar
  4. [4]
    Chailloux J.,Devin M., Hullot J. “LeLisp a portable and efficient Lisp system” Proc. of the 1984 ACM symposium on Lisp and Functional Programming, Austin Taxas, August 1984Google Scholar
  5. Chailloux J., “Manuel de Reference LeLisp 15.2”,3ème Edition, Novembre 1986Google Scholar
  6. [6]
    Howe D., “Padmavati CAM chip functional specification”, Internal report REP-14, July 1989Google Scholar
  7. [7]
    INMOS, The Transputer Data Book, First edition 1989Google Scholar
  8. “Prolog built in predicates”, Report ISO N28Google Scholar
  9. [9]
    Jary P., De Joybert X., “Selection de Clauses en Prolog”. Actes du 8ème Séminaire Programmation en Logique, Trégastel, 24–26 mai 1989Google Scholar
  10. [10]
    Knott G.D., “Hashing Functions”, The computer Journal, vol 18, no 1, pages 38 – 44, 1975MathSciNetGoogle Scholar
  11. [11]
    Kohonen T., “Content Addressable Memories”, Springer series in Information sciences, 1980Google Scholar
  12. [12]
    Nicole D., Lloyd E., Ward J. “Switching Networks for Transputer Links”. 1989Google Scholar
  13. [13]
    Pountain D. Virtual Channels: The Next Generation of Transputers - BYTE, April 1990Google Scholar
  14. [14]
    Warren D.H, “Implementing Prolog”, D.A.I research reports no 39, 40, May 1977Google Scholar
  15. [15]
    Warren D.H, “An abstract Prolog Instruction Set”, Technical Note no309, SRI International, Menlo Park, 1983Google Scholar

Copyright information

© ECSC, EEC, EAEC, Brussels and Luxembourg 1990

Authors and Affiliations

  • Patricia Guichard-Jary
    • 1
  1. 1.THOMSON-CSF Division CIMSA-SINTRAColombesFrance

Personalised recommendations