Integer Linear Programming-based ECO Approach with Wire Length Consideration

Abstract

This paper identifies an integer linear programming-based approach with regards to the spare cells selection as the problem of resource allocation. The objective aims to search the spare cell assignment with minimization of total wire length. A novel wire length estimation method is proposed to accelerate the runtime of the proposed approach. In the proposed method, only the virtual node, i.e. the location of buffer insertion (middle point in this work) of the interconnection and the spare cells are taken into account for acquiring the spare cell assignment with the minimization of total engineering change order in length. If the timing violations still appear, the procedure will insert buffer again by performing the ILP formulations. The circuit characteristics, including the demand of each engineering change order and the supply of spare cell resource, could be successfully integrated into the ILP formulations. Experimental results showed that the proposed integer linear programming-based approach significantly minimized the total wire length.