Parallel Multipipe Rendering for Very Large Isosurface Visualization
In exploratory scientific visualization, isosurfaces are typically created with an explicit polygonal representation for the surface using a technique such as Marching Cubes. For even moderate data sets, Marching Cubes can generate an extraordinary number of polygons, which take time to construct and to render. To address the rendering bottleneck, we have developed a multipipe strategy for parallel rendering using a combination of CPUs and parallel graphics adaptors. The multipipe system uses multiple graphics adapters in parallel, the so called SGI Onyx2 Reality Monster. In this paper, we discuss the issues of using the multiple pipes in a Sort-Last fashion which out performs a single graphics adaptor for a surprisingly low number of polygons.
KeywordsGraphic Hardware Single Graphic Processor Element Marching Cube Parallel Rendering
Unable to display preview. Download preview PDF.
- 1.Kurt Akeley. RealityEngine graphics. Computer Graphics, 27:109–116, August 1993. ACM Siggraph ‘83 Conference Proceedings.Google Scholar
- 3.David A. Ellsworth. A new algorithm for interactive graphics on multicomputers. IEEE Computer Graphics and Applications, 14(4), July 1994.Google Scholar
- 4.Kwan-Lui Ma et al. Parallel volume renderer using binary-swap image composition. IEEE Computer Graphics and Applications, 14(4), July 1994.Google Scholar
- 5.Steve Molnar et al. A sorting classification of parallel rendering. IEEE Computer Graphics and Applications, 14(4), July 1994.Google Scholar
- 7.C. Hansen, M. Krogh, and W. White. Massively parallel visualization: Parallel rendering. In Proceedings of SIAM Parallel Computation Conference, February 1995.Google Scholar
- 10.John Montrym, Daniel Baum, David Dignam, and Christopher Migdal Infinitereality: A real-time graphics system. Computer Graphics, 26:293–302, August 1997. ACM Siggraph ‘87 Conference Proceedings.Google Scholar
- 11.F. Ortega, C. Hansen, and J. Ahrens. Fast data parallel polygon rendering. In Proceedings of Supercomputing ‘83, pages 709–718, November 1993.Google Scholar