Areas with large inherent complexity, such as robotics, that use software both for control in realtime, and simulations, predicting future actions in simulated time, typically are gained by support for modeling autonomous entities. The Object-Orientation (OO) paradigm has proved itself to be an appropriate tool against complexity. A reason to
this seems to be that the OO-paradigm lays close to a common view upon entities of the real world. This means that the phase of modeling such entities into OO-terms is facilitated by the inherent structural properties of the OO-paradigm. However, modeling systems with autonomous active entities need further support than traditionally provided by OO because of the inherent properties of active entities concerning aspects, such as, concurrency and messagebased communication. First, there is a need for software development tools illuminating on the inherent concurrent property of autonomy. Second, there is a need for a basic supporting platform to map the software on that handles issues, such as, interruptions and context switches in time efficient ways. This, however, seems to be a critical matter, and a common issue of criticism against concurrent software often relates to the time it takes to handle, for instance, switches between contexts. We will here illuminate on matters of large scaled software modeling autonomous entities and provide investigations on performance regarding processes and threads on a level of operating system support.