ABSTRACT

Abstract

Nowadays, there has been an increasing demand for high-speed computation, parallel processing, and efficient algorithms to efficiently manage the rapid increase in the volume of data which gave rise to much research being carried out in concurrent applications and mechanisms to improve performances. In this paper, two parallel concurrency frameworks were developed and tested using the quick-sorting algorithm on a multiprocessor machine to compare their performances. System.nanoTime() method was used for benchmarking the two frameworks. The results show that the ServExecSort framework performs better than NaïveParallelSort on a larger array size while the NaïveParallelSortr exhibits better performance on smaller array elements. It is therefore concluded that a framework that uses separate thread pool and future to keep track of submitted tasks exhibits better performances than a framework that creates separate threads for each task which leads to large overheads and performance degradation.

Keywords: Algorithms, Computation, Efficiency, Framework, Performances, Mechanism.