Skip to Main Content
In the past decades, EDA tools were able to keep up with the increasing design sizes, enabled by Moore's Law, through smarter algorithms running on faster processors. Now that most key algorithms are close to n:Log??n??, implementation inefficiencies have been cleaned up and single-threaded performance gains have leveled off, parallel CAD is the way forward. Every EDA application will need to scale to dozens of processors. While several EDA tools have shown good scalability to many threads, some are not scaling beyond 12-16 threads and other tools have not even been replaced with more scalable equivalents. To reach massive parallelism, it is important for every EDA developer to master the following key skills described by the author.