Massively Parallel Dataflow Computer Dedicated to Real-Time Image Processing

Issue title: Massively Parallel Computing

Guest editors: Giacomo R. SechiGuest Editor

Article type: Research Article

Authors: Zavidovique, Bertrand^a | Sérot, Jocelyn^b | Ouénot, Georges M.^{c; *}

Affiliations: [a] IEF/ETCA, Université Paris Sud, Institut d'Electronique Fondamentale, Bât. 220-91405 Orsay, France | [b] LASMEA/ETCA, URA 1793 CNRS, Université Blaise Pascal, 63177 Aubiere, France | [c] UMSI-CNRS, BP 133, 91403 Orsay Cedex, France

Correspondence: [*] Correspondence to: Georges M. Quénot.

Abstract: This paper describes a massively parallel dataflow computer dedicated to real-time image processing. Two types of data-driven processing elements are embedded into a regular cubic mesh: 1024 low level custom dataflow processors (50Mops/s each) and 36 T800 transputers. The programming environment relies on an extensive use of the functional programming (FP) concept. An image processing algorithm first expressed in an FP-like functional syntax is converted into a dataflow graph, the nodes of which are real-time operators to be implemented on physical processors by direct mapping onto the array. The concepts and the architecture, both hardware and software, are explained in detail. We conclude with an example of a complete implementation: a connected components labeling working in real time.

DOI: 10.3233/ICA-1997-4102

Journal: Integrated Computer-Aided Engineering, vol. 4, no. 1, pp. 9-29, 1997