Abstract—A good way of characterizing a parallel system is to consider the synchronization granularity or frequency of synchronization between processes in a system. The scientific applications of the parallel system consist of multiple processes running on different processors that communicate frequently. The performance evaluation of such systems mainly depends on how the processes are co scheduled. If the processes are not co scheduled properly, then the system will lead to severe performance penalties. The various co scheduling techniques available are First Come First Served, Gang Scheduling and Flexible Co Scheduling. First Come First Servead nd Gang Scheduling suffer from internal and external fragmentation. Flexible Co Scheduling saturates at heavy loads. The paper focuses on a new co scheduling algorithm, which concentrates on a detailed classification of the synchronization granularity, and the new algorithm gives better results under heavy loads.
Index Terms—First Come First Served, Flexible Co Scheduling, Gang Scheduling, Parallel Job Scheduling, Performance Metrics.
S. V. Sudha, working as Assistant Professor in the Department of Information Technology, Kalignar Karunanidhi Institute of Technology, Coimbatore 641 402, Tamil Nadu, India (e-mail: svsudha@rediffmail.com)
K. Thanushkodi, Principal, Akshaya College of Engineering and Technology, Coimbatore -642 109, Tamil Nadu, India.
Cite: S. V. Sudha and K. Thanushkodi, "Process Grain Size Based Scheduling of Parallel Jobs with Agile Algorithm," International Journal of Computer Theory and Engineering vol. 1, no. 5, pp. 647-658, 2009.
Copyright © 2008-2024. International Association of Computer Science and Information Technology. All rights reserved.
