Parallel Bucket Sorting Algorithm

Bucket sorting algorithm achieves O(n) running time complexity in average. It is very fast compared to any comparison-based sorting algorithms that usually have a lower bound of Ω(n log n). Given an unsorted array of n positive integers, bucket sorting algorithm works by distributing the elements into m ordered buckets each contains zero or more elements. Each bucket is then sorted using either a different sorting algorithm or recursively using bucket sorting algorithm. At the end, elements are gathered from each bucket in order, so the input array is sorted. With the O(n) complexity, the bucket sorting algorithm can execute and finish sorting very quickly in sequential processing. However, this execution time can even be improved by parallelizing the sorting of each bucket. This paper will show how the algorithm can be converted to a parallel algorithm and be implemented and executed using OpenMP API. Furthermore, it is a scalable algorithm as its performance can be linearly improved with the increase of core number of multicore processors. Experiment results show the efficiency of the algorithm.