ORCHESTRA: an asyncrhonous non-blocking distributed GVT algorithm

Abstract

Taking advantage of high computing capabilities of modern distributed architectures is fundamental to run large-scale simulation models based on the Parallel Discrete Event Simulation (PDES) paradigm. In particular, by exploiting clusters of modern multi-core architectures it is possible to efficiently overcome both the power and the memory wall. This is more the case when relying on the speculative Time Warp simulation protocol. Nevertheless, to ensure the correctness of the simulation, a form of coordination such as the GVT is fundamental. To increase the scalability of this mandatory synchronization, we present in this paper a coordination algorithm for clusters of share-everything multi-core simulation platoforms which is both wait-free and asynchronous. The nature of this protocol allows any computing node to carry on simulation activities while the global agreement is reached.