Polymorphous Computer Architectures (PCA) Technology Transition to Joint Semi Automated Forces (JSAF)
Abstract
This effort investigated the feasibility of applying the University of North Carolina's OneSAF Objective System Graphics Processing Unit (GPU) algorithms to the Joint Semi Automated Forces (JSAF) code, and its civilian derivative, Culture. These codes are primarily used for situational understanding in the U.S. Joint Forces Commands (JFCOM) Urban Resolve experiments. Urban Resolve experiments include urban battle spaces, red and blue forces, a broad mix of sensor platforms, and very large numbers of civilian entities. Line-of-sight (LOS) and route planning calculations are both extensively used, and this effort explored the opportunity of exploiting GPUs to accelerate them. It was determined that both LOS and route planning can be significant performance bottlenecks in JSAF. They are not such bottlenecks that order-of-magnitude improvements can be hoped for. However, investigation determined that factors of two improvements in computational speed are feasible. Given the relatively low cost of GPUs when compared to commodity Linux computing systems, they would represent a cost-effective system upgrade for Joint Experimentation. As a result of this research project, JFCOM has submitted a proposal to the DoD High Performance Computing Modernization Program for a GPU-enhanced cluster.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 2006
- Accession Number
- ADA456918
Entities
People
- Dan M. Davis
- Gene Wagenbreth
- Robert F. Lucas
Organizations
- University of Southern California