Performance analysis of parallel sorting algorithms Using Message Passing Interface (MPI)

A large number of complex applications which can include database, data mining, as well as image processing need to involve a number of computations, simulations, iterations and accuracy. These types of high-end applications may well be solved making use of parallel processing a situation where the problem is divided into a set of independent sub-problems and solving each one of the sub-problems simultaneously. Message Passing Interface (MPI) is a powerful and standard interface with a library of subroutines and functions to write parallel programs. Sorting plays a very important role in many applications. In this paper some parallel algorithms namely parallel quick sort, parallel merge sort and parallel hybrid Merge-Quicksort are discussed. The algorithms are implemented using MPI with C programming language on multi-core system with Linux platform. Parallel programs are tested with large data sets. The overall performance with respect to computing time of each parallel algorithm is analyzed. The comparison of parallel and corresponding sequential algorithms is also performed.