A Constraint-Driven Compact Model with Partition Strategy for Ordered Escape Routing

Zhaopo Liao,Sheqin Dong
DOI: https://doi.org/10.1145/3386263.3406942
2020-01-01
Abstract:Ordered escape routing(OER) is a key problem in PCB-board level routing. In this paper, a compact model driven by linear constraints for ordered escape routing problem is proposed. The model can solve the ordered escape routing with minimized wire length. Our algorithm builds an injective map between the solution space of model and the solution space of OER. Linear contraints are added to eliminate infeasible solutions. Without loss of solution feasibility and optimality, this model can prune the integer linear programming(ILP) variables(constraints) by 79.43%(74.05%) compared to previous work. To further accelerate the speed, a partition strategy based on the routing region of each net is proposed. This can divide the problem into several sub-problems. Experimental results show that, compared to former method which considers wire length, our compact model with partition can achieve up to 100 times speed up with solutions that have almost the same cost.
What problem does this paper attempt to address?