A Method to Evaluate the Performance of a Multiprocessor Machine based on Data Flow Principles

In this paper we present a method to model a static data pow oriented multiprocessor system. This methodology of modelling can be used to examine the machine behaviour for executing a program according to three scheduling strategies, viz., static, dynamic and quasi-dynamic policies. The processing elements (PEs) of the machine go through different states in order to complete tasks they are allotted. Hence, the time taken by the machine to execute a program is directly dependent on the time spent by the PEs in various states during the execution of tasks. We adopt a “state diagram approach to model the machine. This modeling scheme can be used for a class of machines, which have similar execution paradigm. By introducing “wait states in the state diagram of a PE at appropriate places, we capture the delays that are incurred by the PE waiting on events; the events during the execution of a program being those of wait for availability of inputs, access to global memory, response from the scheduling unit and access to communication media. The communication media are modelled as queueing networks and the delay introduced by the wait state (of a PE for accessing a medium) is specified by the queueing delay of the corresponding network model. The novelty of the state diagram approach is that it facilitates faster simulated execution of programs on the machine as compared to that of conventional simulation languages. This is because, the properties of the machine are described by a set of polynomials