A partitioning-based approach for the orientation and rotation assignments of macro cells

Abstract

A unified partitioning based algorithm for the orientation and rotation assignments of macro cells is proposed to minimize total wire length in a macro cell placement. For the orientation and rotation assignments of macro cells, we transform the orientation problem and the rotation problem into a constrained graph bisection problem and a constrained graph quadrisection problem, respectively. Furthermore, a unified fuzzy graph clustering is proposed to solve the two constrained partitioning problems at the same time. The partitioning results of the constrained graph bisection and the constrained graph quadrisection will lend to the orientation and rotation assignments for a macro cell placement. As a result, the proposed partitioning based approach has tested some macro cell placements for the orientation and rotation assignments. The experimental results show that the partitioning based approach obtains better wire reductions on these tested placements.