A continuing problem in entity-level, intelligent agent simulations has been one of efficiently, effectively and expediently aggregating smaller units like squads and platoons into larger ones like companies and battalions and then de-aggregating them again at appropriate times. This paper reviews the goals and issues of the aggregation/de-aggregation (A/DA) problem and then lays out some solutions based on High Performance Computing, computational science and lessons learned from advanced techniques, such as adaptive simulation meshes. Experience has shown and logic dictates that aggregation is a more straightforward operation than is de-aggregation. A/DA of collective units is required for future, large-scale simulations, e.g. Sentient World Simulation. Understanding how to distribute the smaller units and how to represent the impacts of the simulation on these segments has largely eluded the M&S community for years. This problem is made more complex by the existence of significant amounts of "legacy code" and this paper gives examples of a successful approach to working with such code in an HPC environment. Three workable solutions are enabled by HPC: simulating all levels continuously while displaying only the designated unit level, simulating smaller entities' behavior with reduced behavioral resolution to save compute resources, and foregoing all lower level simulation by simulating only the top-level designated. This last method requires laying down the lower-level entities using doctrine, status, and terrain to achieve realistic disposition. This paper will investigate the processes, impacts, and performance of all three methods. Entity migration across various compute nodes in cluster computers and germane HPC examples from similar computational approaches will be described. The approach applies methods, shown to be effective in on-going research in the physical sciences, to problems facing the DoD M&S community. Performance analyses are anticipated, as are user evaluations by operators, controllers, and analysts.