Cluster refinement for block placement

We propose an iterative optimization approach for mixedmacro-cell and standard-cell placement, which minimizes the chipsize and interconnection wire length at the same time. We present abranch-and-bound algorithm which efficiently searches for the optimalsolution by evaluating all of the possible configurations on theselected cluster to minimize the gap distance between the ceilingand the floor. A virtual grid and permutation order are generateddynamically to eliminate redundant branches, which was the causeof much higher complexity in other approaches. Experimentalresults on the MCNC benchmark circuits show that the algorithmachieves very competitive results to manual design.

[1]  Hidetoshi Onodera,et al.  Branch-and-bound placement for building block layout , 1991, 28th ACM/IEEE Design Automation Conference.

[2]  H. Murata,et al.  Rectangle-packing-based module placement , 1995, ICCAD 1995.

[3]  Chung-Kuan Cheng,et al.  Module Placement Based on Resistive Network Optimization , 1984, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.

[4]  Martin D. F. Wong,et al.  A New Algorithm for Floorplan Design , 1986, 23rd ACM/IEEE Design Automation Conference.

[5]  Tsu-Chang Lee,et al.  A Bounded 2D Contour Searching Algorithm for Floorplan Design with Arbitrarily Shaped Rectilinear and Soft Modules , 1993, 30th ACM/IEEE Design Automation Conference.

[6]  Alberto L. Sangiovanni-Vincentelli,et al.  'Zone-refining' techniques for IC layout compaction , 1990, IEEE Trans. Comput. Aided Des. Integr. Circuits Syst..

[7]  Ernest S. Kuh,et al.  Simultaneous Floor Planning and Global Routing for Hierarchical Building-Block Layout , 1987, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.

[8]  Takashi Kambe,et al.  Hybrid floorplanning based on partial clustering and module restructuring , 1996, Proceedings of International Conference on Computer Aided Design.