A Minimum Area VLSI (Very Large Scale Integrated Architecture for O(LOGN) Time Sorting.
Abstract
A generalization of a known class of parallel sorting algorithms is presented, together with a new architecture to execute them. A VLSI implementation is also proposed, and its area-time performance is discussed. It is shown that an algorithm in the class is executable in 0(logn) time by a chip occupying O(n2) area. The design is a typical instance of a 'hybrid architecture', resulting from the combination of well-known VLSI arrays as the orthogonal-trees and the cube-connected-cycles; it is also the first known to meet the AT21 = omega(n2log2n) lower bound for sorters of n words of length (1 + epsilon) and working in minimum 0(logn) time. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1983
- Accession Number
- ADA142414
Entities
People
- Franco P. Preparata
- G. Bilardi
Organizations
- University of Illinois Urbana–Champaign