In this paper, a new fuzzy-clustering-based approach is proposed for two-way circuit partitioning. First, a circuit netlist is represented as an undirected edge-weighted graph by a tree net model. Furthermore, the fuzzy memberships and the clustering distance are introduced into the graph. Based on fuzzy c-means clustering, two groups of the fuzzy memberships will be assigned onto all of the vertices in the graph. Finally, according to these fuzzy memberships and the area information of the circuit netlist, the circuit netlist will be partitioned into two smaller netlists with area-balanced constraints. As a result, the proposed fuzzy clustering-based approach is implemented to obtain a better two-way partitioning with area-balanced constraints on circuit benchmarks.
展开▼
