Utilization of VLSI for Creating an Active Data Base of 3-D Geometric Models

Abstract

Parallelism of geometric computation can be achieved by distributing the computation efforts according to essentially three different strategies, based on functional, spatial and structural division, respectively (Mantyla 1983). The conventional and already commercialized way to introduce parallel computation for viewing 3-D geometric models is employing functional parallelism as a pipeline for performing different sequential transformation phases of the 3-D viewing operation (Clark 1981). This approach limits the number of parallel activities to the number of separable functional computational modules. A second approach for parallelism is the division of the modeling space into separable volume elements, which can be processed independently using a suitable data structure like an octree(Kronlof 1985). The logical component structure of a model gives a third distribution strategy. Then each processor answers only to the computational needs of its assigned objects.