Flattened Parallel DEVS Simulations on GPU Architectures

Discrete Event System Specification (DEVS) is a modeling and simulation of discrete event systems formalism. Most DEVS-based simulators are implemented as sequential programs. However, simulating large-scale complex models in a sequential simulator is impractical (if possible), as simulations may take a long time to execute. A usual technique to speed up simulations is the parallel execution of the simulator. Most parallel discrete-event simulation efforts focus on logical process approaches, resulting in complex simulation architectures. Recent parallelizing efforts lean towards executing the simulators in multicore architectures. Despite promising results, they are limited to the amount of CPU processing cores. In this work, we propose an algorithm to accelerate the execution of DEVS simulations on Graphical Processing Units (GPU) architectures. We show different case studies where the proposed algorithm achieved speedups of up to 12.29 and 16.53 compared to a sequential version.