In: 11th International Conference on Systems Engineering, Las Vegas, July 9-11, 1996, pages 396-401. 1996.
Abstract: In many cases, the execution of a parallel program will be strongly influenced by data dependencies and processes running at the same time on the machine or network. These factors cannot be predicted. Therefore, powerful load balancing procedures are necessary in order to improve the run time behaviour. The disadvantage of this more efficient dynamic approach arises from a large communication or management overhead and delayed activi ties. In our contribution we intend to present an improved dynamic, decentral method for hierarchical automation appliances using idle times of the processor and the communication system for the realization of an efficient load balancing.