Predictive Estimation for Distribution of Interconnects

This paper proposes a new predictive method for distribution of interconnects density function (DIDF) without any need to design layout information. Parameters such as number of cells, Rent´s factors, Gate Pitch, and so on are used as model input. For any specific wire length the model predicts a mean and variance for number of interconnects with that length. It also predicts the interconnect pattern density (IPD) which shows the congestion of interconnects. The derived model shows a worst case of 16% deviation compared to previous works.