Performance of Parallel Volume Graphics Algorithms on SGI Power Challenge

The shared-memory multiprocessor system recently has attracted a lot of attention because of its low cost and wide availability to the graphics researchers. In this paper, we show practical parallel solutions to the following volume graphics[4] problems: volume texture modeling, volume texture rendering, and voxelization for unstructured grid rendering. All these solutions process 2D and 3D arrays and employ P processors on the shared-memory system. We present results for parallel execution time, parallel memory requirement and overheads for the three algorithms described in this work. We also present results of performance prediction based on both analytical and experimental results.