A performance driven probabilistic resource allocation algorithm for analog routers

Abstract

The phase of resource allocation is always ignored in analog routers, since the small scale of analog integrated circuits does not require it. However, in this paper, we introduce a novel resource allocation algorithm, for the traditional greedy strategy without resource allocation cannot meet many performance requirements of analog integrated circuits at all. This algorithm is based on the probabilistic model, considering the important performance requirements, such as variable wire widths, matching, and sensitivity, of analog integrated circuits. Factors about performance are involved in the routing phase in advance, by contrast to the traditional approaches that pay only attention to the wire congestion. Experimental results demonstrate that the routability of nets, especially critical nets, is improved, and the wire congestion in different metal layers is manipulated well, with the guarantee of much significant performance.