high performance computing on graphics processing units: hgpu.org

This paper presents a series of efficient, data parallel algorithms for simulating agent based models. These include methods for handling environment updates, agent interactions and replication. One of the most important techniques presented in this work is a novel stochastic allocator which enables parallel agent replication in O(1) average time. These techniques can be easily implemented on a modern day graphics processing unit (GPU) resulting in a substantial performance increase. We believe that our system is the first ever completely GPU based agent simulation framework.