Abstract
Simulating complex systems may require to handle a huge number of entities, raising scalability issues. In this respect, GPGPU is a relevant approach. However, GPU programming is a very specific approach that limits both accessibility and re-usability of developed frameworks. We here present our approach for integrating GPU in TurtleKit, a multi-agent based simulation platform. Especially, we show how we keep the programming accessibility while gaining advantages of the GPU power. The paper also presents how this approach could be generalized and proposes a MABS design guideline dedicated to the GPU context.