Optimizing Performance of Scientific Visualization Software to Support Frontier-Class Computations

Report No. ARL-TR-7369
Authors: Richard C Angelini
Date/Pages: August 2015; 46 pages
Abstract: The Department of Defense (DOD) High-Performance Computing (HPC) Modernization Program has initiated a new class of funded research initiatives known as Frontier projects. These multiyear projects are devised to leverage using multidisciplinary teams to solve problems which are orders-of-magnitude larger than typical HPC-sized challenge projects. Current scientific visualization techniques are more than adequate for solving a vast majority of the datasets generated; however, these Frontier project datasets with anticipated mesh sizes of more than 10 billion cells offer a variety of challenges for data analysis applications. In this project, EnSight was evaluated against several Frontier-sized datasets to determine if this commercial software application could be optimized to process this data. EnSight has already been established as an essential application within the DOD for interactive data analysis of large HPC simulation results. This particular project focused on not only being able to produce results from very large Frontier-class computational data results but to also optimize overall performance to reduce the amount of time required to visualize those results.
Distribution: Approved for public release
  Download Report ( 1.010 MBytes )
If you are visually impaired or need a physical copy of this report, please visit and contact DTIC.
 

Last Update / Reviewed: August 1, 2015