The performance impact of different master nodes on parallel loop self-scheduling schemes for rule-based expert systems

Abstract

The technique of parallel loop self-scheduling has been successfully applied to auto-parallelize rule-based expert systems previously. In a heterogeneous system, different compute nodes have different computer powers. Therefore, we have to choose a node to run the master process before running an application. In this paper, we focus on how different master nodes influence the performances of different self-scheduling schemes. In addition, we will investigate how the file system influences the performance. Experimental results give users the good guidelines on how to choose the master node, the self-scheduling scheme, and the file system for storing the results. © 2011 Springer-Verlag.