Sorting has been a fundamental research area of computer algorithms, so that sorting algorithms on the sequential processing model are well-developed, but those on the parallel processing models are not the same, because various approaches for the processing model itself are undertaken. But recent rapid progresses in VLSI technology make it possible to realize the supercomputer of parallel processing model.
Especially some hypercube machines are realized and widely spread, the studies on parallel algorithms for hypercube (or binary n-cube) become more important issues in multicomputer researches.
In this thesis, a new parallel sorting algorithm(hypersort) on hypercube machines is proposed for cost optimality. At first, some important properties of hypercube including a computational model are studied and conventional sequential sorting algorithms are summarized for a node processor(computer). Secondly, four analogical parallel sorting algorithms on hypercube are implemented and compared using test-data-sets. Finally, a new parallel sorting algorithm on hypercube(hypersort) is proposed and compared with previous analogical parallel sorting algorithms.
