This paper presents a hierarchical floorplanning approach for macrocell layouts which is based on the bottom-up clustering, shape function computation, and top-down floorplan optimization with integrated global routing and pin assignment. This approach provides means for specifying and techniques for satisfying a wide range of constraints (physical, topological, timing) and is, therefore, able to generate floorplans for a number of different layout styles. A systematic and efficient optimization procedure during the selection of suitable floorplan patterns that integrates floorplanning, global routing and pin assignment, a new pin assignment technique based on linear assignment and driven by the global routing solution and floorplan topology, and an effective timing-driven floorplanning scheme are among the other novel features of the floorplanner. These techniques have been incorporated in BEAR-FP, a macrocell layout system developed at the University of California, Berkeley. Results on various placement and floorplanning benchmarks are quite good.
A sequential dual simplex algorithm for the linear assignment problem
