An efficient load balancing method for parallel ray tracing on heterogeneous workstation networks

摘要

In recent years, research on ray tracing has mostly concentrated on speeding up the algorithm by means of parallelisation. These approaches are usually based on the use of multicomputer and all the analysis of load balancing have been carried out by these architectures. But not the advances in high performance network allow us to think of another type of parallel platform. These architectures present new problems that hinder load balancing. In a network, the workstations connected to it usually have different computational powe4r. This situation causes that some previous proposals of load balancing are not valid now. This paper examines the effectiveness of load balancing strategies for ray tracing on heterogeneous workstations networks. After this analysis we present an efficient load balancing strategy for this type of architectures that bears in mind the computational power of the connected stations to the network, and it reduces the communication needs.