SEC: a simple and effective netlist clustering

In this paper, we propose a simple and effective clustering method called SEC. The relatively strongly connected vertices tend to stay together even though a vertex set is partitioned several times with different initial solutions by using Fiduccia-Mattheyses algorithm called FM. These vertices are formed into a cluster in the proposed clustering method. As the proposed clustering scheme SEC uses FM algorithm, SEC also has the same merits as FM algorithm, i.e., easy to implement and fast. After clusters are formed, we run two-phase FM algorithm. Compared with FM, the proposed algorithm yields, on average, 31.2% improvement in terms of min-cut bisection