Accession Number : ADP001201

Title :   Software Metrics for Performance Analysis of Parallel Hardware,

Corporate Author : HUGHES RESEARCH LABS MALIBU CA

Personal Author(s) : Etchells,R. David ; Nudd,Graham R.

Report Date : JUN 1983

Pagination or Media Count : 11

Abstract : The state of the art of parallel processing is characterized by an extraordinary proliferation of architectures. To date, little work has been done to quantify the relative performance capabilities of this range of architectures, especially from the viewpoint of general Image Understanding processing requirements. This paper discusses performance evaluation of parallel hardware. A set of software metrics is proposed, based on the processing requirements of common IU systems. The intent of the paper is to serve as a point of departure for further work and discussion. (Author)

Descriptors :   *Parallel processing, *Performance tests, *Test and evaluation, Image processing, Algorithms, State of the art, Parallel processors, Computer programs, Taxonomy, Shape, Threshold effects, Convolution, Sorting, Histograms, Correlation techniques, Lines(Geometry), Points(Mathematics)

Distribution Statement : APPROVED FOR PUBLIC RELEASE