Gary Ross Ricard

Merging multiple lists on hierarchical-memory multiprocessors

An efficient multiprocessor algorithm to merge m, m ⩾ 2, sorted lists containing N elements is described. In a hierarchical-memory multiprocessor, the proposed method substantially reduces the data access costs in comparison with traditional schemes that successively merge the lists two at a time. The proposed method uses a novel partitioning algorithm to split up the m lists into range-disjoint parts of equal size. Each of the parts can then be merged concurrently and independently by a different processor using an m-way merge. The time complexity to find a partition is O(m log(fNm), where f, 0 < f < 1, is a fraction that determines the size of a partition. Implementation results for multiway merging using the partitioning algorithm on a Sequent Symmetry multiprocessor are described.