Net Clustering Based Macrocell Placement

  • Authors:
  • Stelian Alupoaei;Srinivas Katkoori

  • Affiliations:
  • University of South Florida, Department of Computer Science and Engineering, 4202 East Fowler Avenue, ENB 118, Tampa, FL;University of South Florida, Department of Computer Science and Engineering, 4202 East Fowler Avenue, ENB 118, Tampa, FL

  • Venue:
  • ASP-DAC '02 Proceedings of the 2002 Asia and South Pacific Design Automation Conference
  • Year:
  • 2002

Quantified Score

Hi-index 0.02

Visualization

Abstract

Given an RTL (Register-Transfer-Level) netlist, a net dependency graph with weighted edges is built. Each node in the graph represents a net and an edge exists between two nodes if the two nets represented by the nodes share one or more macrocells. Clusters of nets are then formed by clique partitioning. A net cluster level floorplan is derived by simulated annealing to define the regions where the nets in each cluster must be routed. The macrocell placement is formulated as a force-directed problem where the terminals of a net are free to move under the influence of forces in the quest for optimal length of the net. A new type of rejection force is introduced in order to obtain a feasible placement. In comparison with the placements generated by CADENCE Silicon Ensemble, we obtained an average total wire length reduction of 22.8% and an average longest wire length reduction of 33% with an average area penalty of only 1.1%.