TurboSAT: Gradient-Guided Boolean Satisfiability Accelerated on GPU-CPU Hybrid System

TurboSAT represents a breakthrough in the field of Boolean satisfiability (SAT) by utilizing a hybrid GPU-CPU system to significantly accelerate the solving process. Traditional SAT solvers have struggled with scalability due to their reliance on sequential conflict-driven search algorithms. By reformulating the SAT problem as a binarized matrix-matrix multiplication layer, TurboSAT combines the strengths of parallel differentiable optimization with sequential search techniques. This innovative approach has resulted in runtime speedups exceeding 200 times compared to state-of-the-art CPU-based solvers. The implications of TurboSAT are profound, as it not only enhances the efficiency of SAT solving but also opens new avenues for research and application in logical reasoning, potentially impacting various fields that rely on complex decision-making processes.