Subdivision of 3D space based on the graph partitioning for parallel ray tracing

Abstract

An approach for parallel ray tracing is to subdivide the 3D space into rectangular volumes and assign the object descriptions with their related computations in each volume to a different processor. The subdivision process is critical in reducing the interprocessor communication overhead, and maintaining the load balance among processors of a multicomputer. In this paper, after a brief overview of parallel ray tracing, a heuristic is proposed to subdivide the 3D space by converting the problem into a graph partitioning problem. The proposed algorithm tries to minimize the communication cost while maintaining a load balance among processors.