Parallel DRC by Functional Decomposition of DRC Decks

Introduction

A new version of Savage DRC system taps the full power of multiprocessor computers by introducing simultaneous execution of DRC commands from a DRC deck. A DRC deck is a set of DRC rules to be performed for a given layout. For typical DRC decks, the described approach alone delivers high processor utilization and speed-up. Further improvements are possible using parallelization by decomposition of IC layout data and exploiting its hierarchy.

This article describes the approach used in the current implementation. The description provides guidelines for writing efficiently parallelizable DRC decks.