Analytical Performance Comparison of BNP Scheduling Algorithms

Parallel computing is related to the application of many computers running in parallel to solve computationally intensive problems. One of the biggest issues in parallel computing is efficient task scheduling. In this paper, we survey the algorithms that allocate a parallel program represented by an edge-directed acyclic graph (DAG) to a set of homogenous processors with the objective of minimizing the completion time. We examine several such classes of algorithms and then compare the performance of a class of scheduling algorithms known as the bounded number of processors (BNP) scheduling algorithms. Comparison is based on various scheduling parameters such as makespan, speed up, processor utilization and scheduled length ratio. The main focus is given on measuring the impact of increasing the number of tasks and processors on the performance of these four BNP scheduling algorithms