Dynamics and Control, 8, 107–116 (1998) c  1998 Kluwer Academic Publishers, Boston. Manufactured in The Netherlands. Load Management in Loosely Coupled Multiprocessor Systems S. BATAINEH AND M. AL-IBRAHIM Dept. of Electrical Engineering, JUST at Irbid, P.O Box 3030 Irbid, Jordan Editor: E. P. Ryan Received September 22, 1995; Revised December 11, 1996 Abstract. Scheduling is a typical technique used to distribute the load in multiprocessor systems. Usually, the manager(dispatcher or operating system)schedule the tasks so that the average finish time is minimized. Constraints related to the characteristics of the load such as precedence relation, deadline time, etc. must be taken into consideration. With ever increasing applications of a new paradigm of divisible tasks in image processing and parallel processing, one must concentrate on the characteristics of the system such as processor speed, link speed, and processor interconnection topology when distributing the load. By exploiting queuing theory, we managed to find different bounds on the arrival rate(load) as a function of link speed, processor speed and the size of tasks. A flow control mechanism for different multiprocessor systems with different topologies is embedded in our analysis. Moreover, our model indicates to the design engineers, depending on the traffic intensity, which element(s)of a parallel system has to be upgraded or replaced to meet the new load. This, of course, has to be justified by cost consideration. Keywords: computation time, communication time, distributed system, divisible tasks, multiprocessor system, task scheduling, task assignment 1. Introduction Parallel processing systems are capable of solving the complex problems that outpace the ability of world’s fastest and most powerful single computers. The idea of exploiting parallelism to speed up processing, gives better utilization of the available resources and improve the throughput ranges from tightly coupled multiprocessor systems to distributed systems. Examples of the first type include ILLIAC IV computer [1], Twisted Torus [4] and Cray-1 [27]. A remote access strategy is used in distributed systems such as LAN and MAN to share resources. This, of course, made it necessary to develop networking and distributed operating systems such as Mach and Plan-9. Unlike sequential systems, parallel systems are more complex. Therefore, we need more intelligent operating systems that must be aware of the status of all processors and distribute the available tasks or subtasks in such away that the average finish time is minimized. In this paper we refer to the operating system by dispatcher or controller. In parallel systems, the software must also be fractioned and distributed to available processors such that the interaction among the processors is minimized. Another important issue is balancing the load among the available processors. This issue is of fundamental importance in enhancing the system performance [5]. Each aspect of parallel processing system is discussed in literature independently. For instance, some works discuss scheduling techniques that give the best finish time [5–11, 17–26]. Eager discussed the issue of speedup versus efficiency in parallel systems [2]. The