Handling routability in floorplan design with twin binary trees

  • Authors:
  • Steve T. W. Lai;Evangeline F. Y. Young;Chris C. N. Chu

  • Affiliations:
  • Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong;Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong;Department of Electrical and Computer Engineering, Iowa State University, IA 50011-3060, USA

  • Venue:
  • Integration, the VLSI Journal
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

As technology moves into the deep-submicron era, the complexities of VLSI circuits grow rapidly. Interconnect optimization has become an important concern. Most routability-driven floorplanners [H.M. Chen, H. Zhou, F.Y. Young, D.F. Wong, H.H. Yang, N. Sherwani, Integrated floorplanning and interconnect planning, in: Proceedings of IEEE International Conference on Computer-Aided Design, 1999, pp. 354-357; S. Krishnamoorthy, J. Lou, H.S. Sheng, Estimating routing congestion using probabilistic analysis, in: Proceedings of International Symposium on Physical Design, 2001, pp. 112-117; M. Wang, M. Sarrafzadeh, Modeling and minimization of routing congestion, in: IEEE Asia and South Pacific Design Automation Conference, 2000, pp. 185-190] use grid-based approach that divides a floorplan into grids as in global routing to estimate congestion by the expected number of nets passing through each grid. This approach is direct and accurate, but not efficient enough when dealing with complex circuits containing many nets. In this paper, an efficient and innovative interconnect-driven floorplanner using twin binary trees (TBT) representation [B. Yao, H. Chen, C.K. Cheng, R. Graham, Revisiting floorplan representations, in: Proceedings of International Symposium on Physical Design, 2001, pp. 138-143; E.F.Y. Young, C.C.N. Chu, Z.C. Shen, Twin binary sequences: a non-redundant representation for general non-slicing floorplan, in: Proceedings of International Symposium on Physical Design, 2002, pp. 196-201] is proposed. The estimations are based on the wire densities (number of wires passing through per unit length) on the half-perimeter boundaries of different regions in a floorplan. These regions are defined naturally by the TBT representation. Buffer planning is also considered by deciding if buffers can be inserted successfully for each net. In order to increase the efficiency of our floorplanner, a fast algorithm for the least common ancestor (LCA) problem in Bender and Farach-Colton [The LCA problem revisited, in: Latin American Theoretical INformatics, 2000, pp. 88-94] is used to compute wire density, and a table look-up approach is used to obtain the buffer insertion information. Experimental results show that our floorplanner can reduce the number of unroutable wires. The performance is comparable with other interconnect-driven floorplanners that perform global routing-like operations directly to estimate routability, but our estimation method is much faster and is scalable for large complex circuits.