Performance prediction model for distributed applications on multicore clusters

Abstract

Distributed processing offers a way of successfully dealing with computationally demanding applications such as scientific problems. Over the years, researchers have investigated ways to predict the performance of parallel algorithms. Amdahl’s law states that the speedup of any parallel program has an upper bound which is determined by the amount of time spent on the sequential fraction of the program, no matter how small and regardless of the number of processing nodes used. This research discusses some of the short comings of this law in the current age. We propose a theoretical model for predicting the behavior of a distributed algorithm given the network restrictions of the cluster used. The paper focuses on the impact of latency and bandwidth which affect the cost of interprocessor communication and the number of processing nodes used to predict the performance. The model shows good accuracy in comparison to Amdahl’s law.