Carnegie Mellon University
file.pdf (932.6 kB)
Download file

Distributed Hartstone real-time benchmark suite

Download (932.6 kB)
journal contribution
posted on 2005-05-01, 00:00 authored by Clifford W. Mercer, Yutaka Ishikawa, Hideyuki Tokuda
Abstract: "In our definition of the Distributed Hartsone (DHS) benchmark, we extend the Hartstone real-time benchmark [14] to the distributed environment where communication and end-to-end scheduling become important. Traditional performance metrics for real-time systems do not tell the whole story, so we need a benchmark to gauge the performance for different areas of the operating system. Furthermore, we give a detailed description of the motivation and formulation of the benchmark to enable practitioners to implement the benchmark on a wide variety of systems. The benchmark is defined to be easily implemented on any system with basic real-time capability such as prioritized, preemptive scheduling and a real-time clock.Also, the tests are open-ended to allow comparisons on a wide variety of hardware. The benchmark has been implemented on the ARTS Kernel [9], a distributed real-time testbed we have developed in the ART Project at Carnegie Mellon University. We give the benchmark results from the ARTS Kernel for different types of processor and communication scheduling algorithms. These results verify the usefulness of the benchmark suite in distinguishing between performance characteristics of these algorithms."




Usage metrics