A Case for Simple SAT Solvers

As SAT becomes more popular due to its ability to handle large real-world problems, progress in efficiency appears to have slowed down over the past few years. On the other hand, we now have access to many sophisticated implementations of SAT solvers, sometimes boasting large amounts of code. Although low-level optimizations can help, we argue that the SAT algorithm itself offers opportunities for more significant improvements. Specifically, we start with a no-frills solver implemented in less than 550 lines of code, and show that by focusing on the central aspects of the solver, higher performance can be achieved over some best existing solvers on a large set of benchmarks. This provides motivation for further research into these more important aspects of SAT algorithms, which we hope will lead to future significant advances in SAT.