Designing a Parallel Collaborative SAT Solver

The combinatorial optimization problems are difficult to set up within a parallel context: a search-space is built during the computation and is explored in an irregular way. Moreover, the size of the search-space grows exponentially according to the size of the problem. Since the current processor industry is multiplying the number of cores in their product instead of increasing their frequency, setting up a parallel scheme to combinatorial problem remains needed. In this paper we focus on the SAT problem, which is central in theory of the Computer Science. We show different ways to solve SAT in parallel and we explain the advantages and the shortages of them. Then we explain how to design a parallel SAT solver that is able to keep the very efficient sequential techniques while parallelizing the work among the available cores. Obviously, the key-feature for the SAT community is to parallelize a sequential SAT solver since last sequential improvements are embedded in it. To answer these problems we propose a parallel SAT solver called MTSS (for MultiThreaded SAT Solver).