Performance evaluation of task assignment algorithms in cube-based multiprocessor systems

The key advantage of massively parallel systems is to allow concurrent execution of workload characterized by computation units known as processes or tasks, which can be independent programs or partitioned modules of a single program. The scheduling problem is to maintain a balanced execution of all the tasks among the various available processors (nodes) in a multiprocessor system. Cube based multiprocessors have been widely used for executing parallel applications with proper utilization of interconnected nodes. In this paper a new dynamic scheduling scheme considering the topological properties of cube based networks is proposed. The algorithm is designed through the comprehensive study and analysis of various existing dynamic scheduling schemes designed for different types of multiprocessor networks. The main idea behind the design is to check the effectiveness of the proposed algorithm on various cube based multiprocessor systems by scheduling the divisible load efficiently. Two existing algorithms have also been implemented on similar systems. The comparative simulation study shows that the proposed dynamic scheme gives better performance in terms of task scheduling on cube based multiprocessor networks.