A class of min-cut placement algorithms

In this paper we present a class of min-cut placement algorithms for solving some assignment problems related to the physical implementation of electrical circuits. We discuss the need for abandoning classical objective functions based upon distance, and introduce new objective functions based upon "signals cut." The number of signals cut by a line c is a lower bound on the number of routing tracks which must cross c in routing the circuit. Three specific objective functions are introduced and the relationship between one of these and a classical distance measure based upon half-perimeter is presented. Two min-cut placement algorithms are presented. They are referred to as Ouadrature and Slice/Bisection. The concepts of a block and cut line are introduced. These two entities are the major constructs in developing any new min-cut placement algorithm. Most of the concepts presented have been implemented, and some experimental results are given.