Parallelizing Visualization Algorithms for Interactive Computational Analysis

Computational Engineering International, Inc. (CEI) proposes (in Phase I) to investigate and (in Phase II) develop a visualization environment which will enable scientists and engineers to effectively investigate very large data sets that are currently or will be produced through computational analysis in the High Performance Computing (HPC) community. The lack of an effective software visualization environment for these large data sets has the potential to negatively impact the productivity of computational analysts and to seriously retard the further development of computational methods. The visualization environment proposed can be made effective for HPC users by greatly increasing the performance of extraction and display algorithms, taking advantage of coarse and fine grain parallelization through task and domain decomposition, and by providing an environment which will encourage cooperation among HPC teams. The main goal of Phase I efforts will be to identify and test methods by which a visualization environment may be made more effective for "HPC size" results data sets than any visualization environment which exists today. The tools and methods identified in Phase I of the project will be fully implemented in Phase II into a commercially released software package. The release of a commercially available visualization tool specifically targeted for the postprocessing of large datasets will greatly enhance the ability of scientists and engineers to more quickly and effectively interpret computational results data generated in the HPC community. The results of Phase II work will be incorporated into CEI's existing product and immediately made commercially available.