We have previously developed a data structure, called parallel heap, which is the first heap-based data structure to have efficiently implemented a scalable parallel priority queue on an exclusive-read and exclusive-write PRAM [4], Employing p processors, a parallel heap allows deletions of O(p) highest- priority items and insertion of O(p) new items each in O(logn) time, where n is the size of the parallel heap. The number of processors, p, is optimally scalable up to n. A simplified and practical version of parallel heap has been reported in [l] which forms the theoretical basis for much of the current work. However, significant practical improvements were needed before the parallel heap could be implemented efficiently on a commercial parallel computer.