Fast Multicore FastHenry and a Tetrahedral Modeling Method for Inductance Extraction of Complex 3D Geometries

FastHenry is a powerful numerical engine with which to calculate inductance in superconducting structures, but modern high-end multilayer fabrication processes result in dense calculation problems for which it was not optimized. We identify these shortcomings for typical calculation problems and present algorithmic improvements and multicore parallelization to increase computational efficiency. We attain performance increases of one to two orders of magnitude for models of real circuit layouts. We also show that these multicore methods deliver similar performance increases when applied to an engine that uses tetrahedral elements.