Optimal wiresizing for interconnects with multiple sources

Abstract

The optimal wiresizing problem for nets with multiple sources is studied under the distributed Elmore delay model. We decompose such a net into a source subtree (SST) and a set of loading subtrees (LSTs), and show the optimal wiresizing solution satisfies a number of interesting properties, including: the LST separability, the LST monotone property, the SST local monotone property and the general dominance property. Furthermore, we study the optimal wiresizing problem using a variable grid and reveal the bundled refinement property. These properties lead to efficient algorithms to compute the lower and upper bounds of the optimal solutions. Experiment results on nets from an Intel processor layout show an interconnect delay reduction of up to 35.9\% when compared to the minimum-width solution. In addition, the algorithm based on a variable grid yields a speedup of two orders of magnitude without loss of accuracy, when compared with the fixed grid based methods.