Automatic jog insertion for 2D mask compaction: a global optimization perspective

Abstract

A novel approach is presented to global optimization in 2D symbolic layout compaction based on 'branch and bound' optimization, including automatic overconstraint resolution and jog insertion. The main characteristics are: an efficient generation in 2D of a nearly irredundant set of simple X and Y, diagonal and user constraints; incremental event-driven longest path calculation with positive cycle detection and identification simultaneously in the X and Y constraint graphs; jog insertion based on LP information from both graphs; and incremental update of constraints based on local information within the graphs whenever a diagonal constraint is relaxed or a jog is inserted. As both dimensions are processed concurrently from a global perspective, the method shows promising results.<>