Multi-threaded circuit simulation using OpenMP

Circuit-level simulation is a computationally-intensive problem that has proven to be particularly difficult to parallelize. While device evaluation can be performed in parallel in conventional circuit simulators, the execution overhead is high. We show that, by partitioning a circuit, OpenMP can be used to solve sub-circuits in different threads, without compromising accuracy. It is shown that execution time can be reduced proportionally to the number of threads.