Using Hartstone Uniprocessor Benchmark in a real-time systems course

Demonstrates the usability of the application-oriented Hartstone Uniprocessor Benchmark (HUB) to support teaching in fixed-priority scheduling analysis. Our implementation of the benchmark in C runs on different real-time operating systems (e.g. LynxOS) and also on standard desktop systems (Windows NT). We have created some helpful tools around the Hartstone program that may be used for a performance evaluation and a rapid analysis of real-time applications. They give the possibility to understand the real-time responsiveness of applications using both real-time and standard operating systems in an uncomplicated way. We distinguish three different methods to compare performance and to check the theoretical results of scheduling analysis. The first is based on finding breakdown utilization points of process sets. The second method inspects the special overload behavior beyond the breakdown utilization point. This observation reveals a very interesting behavior of the system under overload conditions. Finally, we evaluate the performance based on a simulation of a real-world real-time application.