To Thread or Not to Thread

EDA applications have traditionally pushed the performance limits of computing hardware, but raw performance gains from technology improvements have been diminishing over recent design generations. In fact, the push for reduced power consumption and implementation of multiple- core microprocessors has in some cases resulted in a reduction of single-core performance. Thus, parallelization appears to be the most viable option for achieving significant performance improvements over the foreseeable future and has become a focus in the design automation field. Modern microprocessor contains billions of transistors, each consisting of multiple overlapping geometric shapes in the layout. A network of electrical paths joins together the terminals of these transistors, and each path can consist of multiple geometric shapes.