Abstract :
Abstract — Parallel program represented by an edgedirected
acyclic graph (DAG) based on homogenous
processors. The objective is to minimize the execution time,
evaluate and compare the performance of the individual
algorithms and select the best algorithm. For this purpose
various tasks scheduling algorithms are available like BNP,
UNC, TDB and APB. All the models of algorithms are used
in various cases where BNP is used for bounded number of
processors i.e. in this case the number of processors is fixed,
UNC is for unbounded number of processors i.e. where the
number of processors is not fixed and it changes. For
mapping the tasks onto to processor in a parallel system,
DAG plays an important role. DAG represents the tasks and
their computational times along with the communications
times in between them. It makes easy representation of the
task in graphical fashion. The execution times of tasks on
multiple processors depicted using Gantt charts, help in
analyzing the starting and ending times of multiple tasks.
The performance factor of task scheduling algorithms is
calculated using various measures which includes the
Makespan, Average processor utilization, SpeedUp and the
Scheduled Length Ratio. All these factors help in effectively
measuring the performance of algorithms.
