Title :
Multi-Way Partitioning Via Spacefilling Curves and Dynamic Programming
Author :
Alpert, C.J. ; Kahng, A.B.
Author_Institution :
UCLA Computer Science Department, Los Angeles, CA
Abstract :
Spectral geometric embeddings of a circuit netlist can lead to fast, high quality multi-way partitioning solutions. Furthermore, it has been shown that d-dimensional spectral embeddings (d > 1) are a more powerful tool than single-eigenvector embeddings (d = 1) for multi-way partitioning [2] [4]. However, previous methods cannot fully utilize information from the spectral embedding while optimizing netlist-dependent objectives. This work introduces a new multi-way circuit partitioning algorithm called DP-RP. We begin with a d-dimensional spectral embedding from which a 1-dimensional ordering of the modules is obtained using a spacefilling curve. The 1-dimensional ordering retains useful information from the multi-dimensional embedding while allowing application of efficientalgorithms. We show that for a new Restricted Partitioning formulation, dynamic programming efficiently finds optimal solutions in terms of Scaled Cost [4] and can transparently handle user-specified cluster size constraints. For 2-way ratio cut partitioning, DP-RP yields an average of 45% improvement over KP [4] and EIG1 [6] and 48% improvement over KC [2].
Keywords :
Circuits; Clustering algorithms; Computer science; Cost function; Dynamic programming; Laplace equations; Optimization methods; Partitioning algorithms; Size measurement; Springs;
Conference_Titel :
Design Automation, 1994. 31st Conference on
Print_ISBN :
0-89791-653-0
DOI :
10.1109/DAC.1994.204183