high performance computing on graphics processing units: hgpu.org

One of the principle challenges of Constraint Automata composition is the rapid growth of the state space and the diffficulty inherent in processing very large state spaces both in terms of space as well as computation time. We show that the method outlined here goes some way in tackling both these issues by making it possible to process the composition in parallel using GPGPU programming. We also show how, using the methods put forth to make the GPGPU parallelization possible, it is possible to distribute the composition of Constraint Automata across many nodes.