Wire density driven top-down global placement for CMP variation control

In this paper, we present a top-down global placement algorithm considering wire density uniformity for CMP variation control. The proposed algorithm is based on top-down recursive bisection framework. Wire weight balancing constraint is employed into bisection to consider wire density uniformity. A multilevel hypergraph partitioning satisfying balancing constraints on not only cell area but also wire weight is performed to acquire more uniform wire distribution. Empirical wire weight model is used to estimate wire density distribution before each bisection of a placement bin. Experimental results show that our algorithm improves ROOSTER with more uniform wire distribution by 3.1% on average and limited increment of wire length by 3.0%.