The OpenAB (open agent benchmark) initiative is a community driven project to provide models and procedures for benchmarking of multi agent systems on parallel and distributed computing systems. The purpose of the initiative is to promote best use of benchmarks to ensure that they are used to give insight into both simulator and model performance under a number of scaling characteristics which go beyond just scaling of population sizes. The advantage of this approach is twofold. Firstly it provides a set of common models which simulator designers can use to optimise performance by considering how models scale. Secondly it gives modellers insight into the scalability of models under different conditions to enable them to make informed decisions about the suitability of simulators and hardware architectures.
The initiative originates from a discussion session at the 3rd Annual PADABS (Parallel and Distributed Agent Based Simulation) conference held in Vienna Austria.
The OpenAB initiative is open to all and is governed by a set of bylaws which ensure that peer reviewing of proposed benchmarks is fair. If you are interested in joining the reviewing committee or contributing to the project then please get in touch.