Home         Authors   Papers   Year of conference   Themes   Organizations        To MES conference

Constraint Driven VLSI Floorplanning by Non-Linear Optimization
Authors
	Lokinskiy S.N.
	Marchenko A.M.
	Popov E.A.
	Savelyev V.I.
	Yakimov M.Yu.
Date of publication
	2016
Abstract
	In this paper, we present an effective algorithm for VLSI floorplanning. Given any topology or technology constraints between either rectangles or complex rectilinear polygons, representing floorplan objects, we formulate floorplanning problem as constrained NLP problem with area-like cost function. This approach can handle any consistent topological/technological constraints, such as soft/hard/preplaced blocks represented by rectilinear shapes, routing objects, such as pins, vias, routing segments, their relations and positions. Unlike linear programming, the proposed approach can be deal with non-convex cost function, which is common for VLSI floorplanning tasks, and is flexible to add new topological/technological constraints in the form of specific analytical functions without changing the optimization algorithm. The optimization algorithm itself is based on well-known penalty functions method where penalty functions corresponds to topological/technological constraints. The algorithm replaces an original non-linear constrained optimization problem by a series of unconstrained non-linear problems whose solutions iteratively converge to the solution of the original constrained problem. In the paper we have shown empirically, that our method needs not much iteration and finds the reasonable sub-optimum solution in reasonable run time.
Keywords
	floorplan, packing, non-linear programming, sizing, cost function, penalty function, non-convex function, smooth approximation.
Library reference
	Lokinskiy S.N., Marchenko A.M., Popov E.A., Savelyev V.I., Yakimov M.Yu. Constraint Driven VLSI Floorplanning by Non-Linear Optimization // Problems of Perspective Micro- and Nanoelectronic Systems Development - 2016. Proceedings / edited by A. Stempkovsky, Moscow, IPPM RAS, 2016. Part 1. P. 172-178.
URL of paper
	http://www.mes-conference.ru/data/year2016/pdf/D154.pdf