Isosurface Extraction on the CELL Processor

Abstract

In this paper we describe a parallel method of extracting isosurfaces from large volumetric datasets, adapted for implementation on the Cell Broadband Engine (CBE). The CBE is a new multicore microprocessor architecture designed to improve upon conventional processors in terms of memory latency, bandwidth and power. However, a different approach to algorithm design is needed in order to take full advantage of the CBE’s potential. Our method consists of dividing the volume into groups of slices and submitting each group to a different core in the CBE for parallel processing with a marching tetrahedra algorithm. We describe the algorithm’s adaptation for implementation on a Cell-based Blade server and demonstrate overall isosurface extraction speeds that are a significant improvement when compared to conventional CPUsand Graphics Processor Units (GPUs).