Performance evaluation of rank sort parallelism

Abstract

Parallelism performance is a crucial part for testing algorithmic design in various platforms (not only on multiprocessor but also on multi-computer processor systems). In this paper we demonstrate a comparative parallelism performance outcome from Rank sort algorithm and then compare the parallel and sequential approach of the Rank sort. The divide and conquer approach is used to control a large problem into pieces (called tasks) and send the tasks to some inter-process/multi-computer processor systems for doing some computations and return back the results to the master computer. The master computer then combines all the tasks from working computer and displays the final result. In this research the execution time, speedup time and efficiency matrices are used to find the barrier of Rank sort parallelism. Full Text Link: http://doi.org/10.1109/ISTT.2012.6481596