Parallel Processing For Differencing and Histogramming
Abstract
Algorithms which take advantage of massively parallel and vector architectures are formulated for the calculation of the differences of times-of- arrival derived from a sequence of time samples. The application which generated the work is the classification of pulse trains which are detected by an electronic support measures receiver. Four algorithms are proposed, of which three have been implemented and tested on several machines. The results obtained through testing are compared with each other and with results for more conventional architectures. It is found that one of the machines, the massively parallel MASPAR MP-1 computer, is able to perform histogramming fastest through exploitation of its data-parallel architecture. Several suggestions are made which could significantly improve performance on all architectures. Parallel processing, Histograms, Time of arrival(Signals).
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1993
- Accession Number
- ADA277060
Entities
People
- Pinaki S. Ray
- Vaughan Clarkson