ABSTRACT
In this paper we describe a methodology to predict the tightly clustered (shorter) connections in the final optimal placement of an arbitrary netlist. This new methodology is based on classifying the connections in the netlist into several groups based on their topological characteristics. The most important characteristic that helps in this classification of connections is the presence of multiple paths between two nodes in a netlist. We show that this new methodology consistently results in identifying shorter connections much better than previous models and is independent of the placement approach used. In fact, on average, the cumulative length of the identified shorter connections is 36% (with Simulated Annealing), 33% (with Capo8.8) and 33% (with Dragon) less than those identified by a current mutual contraction model [10]. The model on average identifies 27% more connections than the minimum required for coarsening the netlist and reducing its size in half.
- W. E. Donath, "Placement and average interconnection lengths of computer logic," in IEEE Transactions on Circuits and Systems, vol. CAS-26, pp. 272--277, April 1979.Google Scholar
- W. E. Donath, "Wire length distributions for placements of computer logic," in IBM Journal Research and Development, vol. 25, pp. 152--155, May 1981.Google ScholarDigital Library
- M. Pedram and B. Preas, "Interconnection length estimation for optimized standard cell layouts," in Proceedings of International Conference on Computer-Aided Design, 1989, pp. 390--393.Google Scholar
- H. T. Heineken and W. Maly, "Standard cell interconnect length prediction from structural circuit attributes," in Proceedings of Custom Integrated Circuits Conference, 1996, pp. 167--170.Google ScholarCross Ref
- A. E. Caldwell, A. B. Kahng, S. Mantik, I. L. Markov and A. Zelikovsky, "On wire length estimations for row-based placement," in IEEE Transactions on Computer-Aided Design, vol. 18, pp. 1265--1278, September 1999. Google ScholarDigital Library
- A. Alvandpour, P. Larsson-Edefors and C. Svensson, "GLMC: Interconnect length estimation by growth-limited multifold clustering," in Proceedings of IEEE International Symposium on Circuits and Systems, 2000, pp. 465--468.Google Scholar
- S. Bodapati and F.N. Najm, "Pre-layout estimation of individual wire lengths," in ACM International Workshop on System-Level Interconnect Prediction, 2000, pp. 93--98. Google ScholarDigital Library
- J. A. Davis, V.K. De and J.D.Meindl, "A stochastic wire-length distribution for gigascale integration (GSI) -- Part I: Derivation and validation," in IEEE Transactionson Electronic Devices, vol. 45, pp. 580--589, March 1998.Google ScholarCross Ref
- D. Stroobandt and J. V. Campenhout, "Accurate interconnection length estimations for predictions early in the design cycle," in VLSI Design, vol. 10, pp. 1--20, June 1999.Google ScholarCross Ref
- B. Hu and M. Marek-Sadowska, "Wire length prediction based clustering and its application in placement," in ACM/IEEE Design Automation Conference, 2003, pp. 800--805. Google ScholarDigital Library
- IBM Placement Benchmarks {Online}. Available: http://er.cs.ucla.edu/benchmarks/ibm-place/Google Scholar
- Capo Placement Tool {Online}. Available: http://vlsicad.eecs.umich.edu/BK/PDtools/tar.gz/Placement-bin/Google Scholar
- Dragon Placement Tool {Online}. Available: http://er.cs.ucla.edu/Dragon/Google Scholar
Index Terms
- A priori prediction of tightly clustered connections based on heuristic classification trees
Recommendations
Wire length prediction in constraint driven placement
SLIP '03: Proceedings of the 2003 international workshop on System-level interconnect predictionExperiments show that lengths of individual wires are different for different placement algorithms. To achieve accurate wire length prediction, some knowledge of a placer's details is necessary. We postulate that wire length prediction should be coupled ...
Wire length prediction based clustering and its application in placement
DAC '03: Proceedings of the 40th annual Design Automation ConferenceIn this paper we introduce a metric to evaluate proximity of connected elements in a netlist. Compared to connectivity [8] and edge separability [4], our metric is capable of predicting short connections more accurately. We show that the proposed metric ...
MP-Trees: A Packing-Based Macro Placement Algorithm for Modern Mixed-Size Designs
In this paper, we present a new multipacking-tree (MP-tree) representation for macro placements to handle modern mixed-size designs with large macros and high chip utilization rates. Based on binary trees, the MP-tree is very efficient, effective, and ...
Comments