Dynamic circuit generation for solving specific problem instances of Boolean satisfiability

Optimization and query problems provide the best clear opportunity for configurable computing systems to achieve a significant performance advantage over ASICs. Programmable hardware can be optimized to solve a specific problem instance that only needs to be solved once, and the circuit can be thrown away after its single execution. This paper investigates the applicability of this technology to solving a specific query problem, known as Boolean Satisfiability. We provide a system for capturing the complete execution cost of this approach, by accounting for CAD tool execution time. The key to this approach is to circumvent the standard CAD tools and directly generate circuits at runtime. A set of example circuits is presented as part of the system evaluation, and a complete implementation on the Xilinx XC6216 FPGA is presented.