A Novel Approach for Task Scheduling in Parallel Computing using Priority Attributes

Abstract

The major objective of any task-scheduling algorithm in parallel computing is to minimize the overall scheduling length. In this research paper, we present a task-scheduling algorithm in parallel computing called the Critical Path and b-level Based Task Scheduling algorithm. This algorithm is based on a new attribute, which is called the CPB attribute, which is based on two well-known priority attributes: Critical Path and B-Level. This algorithm uses entry tasks as duplicates in all of the available processors. The major objective of any task-scheduling algorithm is to minimize the overall scheduling length. Here, an application program in task scheduling is represented by a directed acyclic graph. We consider all processors to be homogenous in nature and fully connected. The Critical Path and b-level Based Task Scheduling algorithm gives a minimum scheduling length when compared to heuristic task-scheduling algorithms like the HLFET, MCP, DLS, and ETF. We have also done an analysis of the proposed algorithm and heuristic algorithms based on other priority attributes: Speedup, Efficiency, Scheduling Length Ratio, Load Balancing, and Cost.