In this paper we present a parallel global routing algorithm for general cell layout. The algorithm applies a hierarchical decomposition strategy that recursively divides routing problems into simple, independent subproblems for parallel processing. The solution of each subproblem is based on integer programming and network flow optimization. The algorithm is implemented on a shared-memory machine and experiment results on different examples show relative speedup between 4 and 5 for 8 processors. The speedup is achieved without compromising the quality of the routing results.

This paper presents new upper bounds for channel routing of multiterminal nets, which answers the long-standing open question whether or not multiterminal problems really require channels two times wider than 2-terminal problems. We transform any multiterminal problem of density d into a so-called extended simple channel routing problem (ESCRP) of density 3d /2 + O√ d log d) . We then descibe routing algorithms for solving ESCRPs in three different models. The channel width w is ≤ 3d/2 +O(√d log d) in the knock-knee and unit-vertical-overlap models, and w ≤ 3d/2 + O√d log d) + O(ƒ) in the Manhattan model, where f is the flux of the problem. In all three cases, we improve the best-known upper bounds.

We give an Ο(n3·log n) time and Ο(n3) space algorithm for the continuous homotopic one layer routing problem. The main contribution is an extension of the sweep paradigm to a universal cover space of the plane.

This paper presents a new algorithm for channel routing of multiterminal nets. We first transform any multiterminal problem of density d to a socalled extended simple channel routing problem (ESCRP ) of density 3d/2+O(√dlog d), which will then be solved with channel width w ≤3d/2+O(√dlog d) in the knock-knee model. The same strategy can be used for routing in the other two models: The channel width is w ≤ 3d/2+O(√dlog d)+O(f) in the Manhattan model, where f is the flux of the problem, and w ≤ 3d/2+O(√dlogd) in the unit-vertical-overlap model. In all three cases we improve the best known upper bounds.