Measuring, Predicting, and Improving CGF Performance

摘要

There are at least two ways to improve the performance of a simulation: actively develop algorithmic improvements that make better use of available computing power, or passively wait for computing power to increase on its own. This paper closely examines these two approaches as they relate to computer generated forces. It begins by discussing quantitative measures of CGF performance. It then presents a model of the trend of CGF performance requirements, and shows that the passive approach of waiting for computing power to increase will not provide sufficient CGF performance improvements to meet the future needs of the simulation community. Thus, there is a need for algorithmic improvements to CGF. Finally, the paper concludes by examining some algorithmic methods that can be used to improve CGF performance.