An efficient search space smoothing based netlist partitioning

Partitioning is important in many phases of modern VLSI hierarchical design. The search space smoothing concept has received much attention recently on solving combinational optimization since it improves local search quality. In this paper, we define a new net weight metric based on hypergraph model and present an alternate smoothing process. A size constraint partitioning approach, based on well-known Fiduccia-Mattheyses algorithm with search space smoothing, is also proposed, which not only inherits the efficiency, but also enhances the quality of partition. Experimental results on industry benchmark suites show an average of 10% reduction of cutsize with the same predefined cut ratio.