--------------------------------------- C-tree -- The construction of buffered Steiner trees for Difficult Instances Author : Charles J. Alpert (IBM Austin Research Laboratory) C. N. Sze (Texas A&M University) Jiang Hu (Texas A&M University) Copyright (c) 2003 All rights reserved. http://ece.tamu.edu/~cnsze/GSRC/ctree.html Version : 1.0 Last updated : 2003 July 22 --------------------------------------- Important Notes International Business Machines Corporation (IBM) holds the patent (#6591411) of the C-tree techniques. Please refer to the patent page: NAME ctree - the C-tree implementation SYNOPSIS ctree -i [ -v ] [ -alpha ] [ -beta ] [ -clu_buf ] [ -D ] [ -o ] [ -vg_buf ] DESCRIPTION C-tree is a implementation for the construction of buffered Steiner trees which focus on the ``difficult instances'' which are characterized by a large number of sinks, large variation in sink criticalities nonuniform sink distribution, and varying polarity requirements. PARAMETERS [ -v ] verbose mode [ -alpha ] set the parameter alpha in the equation of crit(). This parameter is to control that how much amount of sinks will be treated as critical and all other non-critical sinks would have a similar value of crit(). Default = 2 [ -beta ] set the parameter beta in the equation of dist(). This parameter is to balance the importance of spatial distance and temporal distance. A bigger beta value indicates that spatial distance is more important. Default = 0.65 [ -clu_buf ] set the buffer resistance, capacitance and intrinsic delay for the achievable slack calculation in sink clustering step. Please note that this buffer information is not related to actual buffer insertion. [ -D ] set the diameter threshold D which is used by the K-center heuristic. The heuristic is controlled by D -- we keep finding a new seed which is furthest from all other old seeds until the minimum distance from the new seed to all other old seeds is less than or equal to D. Based on the distance equation, the maximum distance between two sinks is 2 and so, if only one cluster is needed, set "-D 2". Default = 0.1 -i input file to ctree is a .net file which contains all the pin informations about the net as well as the technology parameter. see http://ece.tamu.edu/~cnsze/GSRC/ctree.html for more information. [ -o ] if the output file is not specified, standard output will be used. ctree outputs the tree information (.out) including the physical location of all pins and Steiner points as well as their connections. It also outputs a (.log) log file which record the program activities, the CPU usage and the slack information. Default = result.out [ -vg_buf ] this option specifies the buffer library which is used by the final van Ginnekin style buffer insertion step. If the buffer library is not specified, the program will skip the buffer insertion step. REFERENCE C. J. Alpert, M. Hrkic, J. Hu, A. B. Kahng, J. Lillis, B. Liu, S. T. Quay, S. S. Sapatnekar, A. J. Sullivan, P. Villarrubia, ``Buffered Steiner trees for difficult instances'', IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, Volume: 21 Issue: 1 , January 2002, pp. 3-14 http://ece.tamu.edu/~cnsze/GSRC/ctree.html