Some experimental results on placement techniques

Abstract

Seven placement algorithms -one constructiveinitial placement algorithm and six iterative-improvement algorithms -were prograi0med and run on six problems ranging in size from 60 to 1300 modules. These problems included placing IC packs on a card, cards on a board and circuits on an LSI chip.It was found that the new force-directed pairwise relaxation algorithm was the best algorithm for the larger problems and was competitive with the other algorithms for the smaller problems.Other questions relating to placement strategies (such as, what are the advantages of constructiveinitial start vs. random start and what is the best transformation of the placement problem to an associated quadratic assignment problem) are also answered.

Mathematical methods for physical layout of printed circuit boards: an overview

“…Consequently, practically every known heuristic scheme, including cluster development (Areibi and Yang 2004;Hanan and Kurtzberg 1972a;Hanan et al 1976a;Magnuson 1977;Cox and Carroll 1980), knowledge based systems (Pannérec 2003), randomized local search algorithms such as simulated annealing (Sechen and Sangiovanni-Vincentelli 1986;Sechen 1988;Wong et al 1988;Wang et al 2000;Murata et al 1998), and genetic algorithms (Cohoon and Paris 1987;Shahookar and Mazumder 1990;Valenzuela and Wang 2002;Sait et al 2005;Areibi and Yang 2004), as well as combinations of these approaches (Zhang et al 2005) have been used to compute placements. Often, computed placements are improved by iterative heuristics based on component interchange (Magnuson 1977;Coté and Patel 1980).…”

“…This can be formulated as a linear integer optimization problem with either a linear (Akers 1981) or a quadratic objective function (Hanan and Kurtzberg 1972b;Hanan et al 1976a;Weismantel 1992).…”

An offer generation approach to SLA negotiation support in service oriented computing

“…As pointed out in Hanan et al, the placement problem can be solved with several methods that fall into three categories, the constructive initial-placement, the iterative placement-improvement and branch-and-bound. In further work, the authors in [17] have classified the methods into seven types that include one constructive algorithm and 6 iterative algorithms.…”

A new placement algorithm minimizing path delays

“…In this phase, the paths are regarded as nets, and iterative placement is performed using the FDPR method [8,9] after specific placement which is either random placement or initial placement. The target point is calculated as follows: where Pc is the target position of the cell h), PC~P is the cell's (c) favorable position of the path (p), and npec is the number of paths which go through the cell (c).…”