Issue No. 11 - November (1993 vol. 4)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/71.250102
<p>Proposes techniques for mapping application algorithms onto a class of hierarchicallystructured parallel computing systems. Multiprocessors of this type are capable ofefficiently solving a variety of scientific problems because they can efficiently implementboth local and global operations for data in a two-dimensional array format. Among theset of candidate application domains, low-level and intermediate-level image processingand computer vision (IPCV) are characterized by high-performance requirements.Emphasis is given to IPCV algorithms. The importance of the mapping techniques stemsfrom the fact that the current technology cannot be used to build cost-effective andefficient systems composed of very large numbers of processors, so the performance ofvarious systems of lower cost should be investigated. Both analytical and simulationresults prove the effectiveness and efficiency of the proposed mapping techniques.</p>
Index Termsmapping algorithms; hierarchical systems; parallel computing systems; image processing;computer vision; high-performance requirements; mapping techniques; processassignment; scheduling; computer vision; image processing; multiprocessorinterconnection networks; parallel architectures
S. Ziavras, "Efficient Mapping Algorithms for a Class of Hierarchical Systems," in IEEE Transactions on Parallel & Distributed Systems, vol. 4, no. , pp. 1230-1245, 1993.