Obstacle-Aware Length-Matching Routing with Constraint Programming and Mixed Integer Linear Programming
碩士 === 元智大學 === 資訊工程學系 === 100 === In the design of high performance printed circuit boards (PCBs), length-matching routing is one of the most important tasks. Nets which belong to the same bus may be required to consider signal propagation delay. When the clock frequency increases, the signal propa...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Online Access: | http://ndltd.ncl.edu.tw/handle/11272810557572342649 |
| Summary: | 碩士 === 元智大學 === 資訊工程學系 === 100 === In the design of high performance printed circuit boards (PCBs), length-matching routing is one of the most important tasks. Nets which belong to the same bus may be required to consider signal propagation delay. When the clock frequency increases, the signal propagation delay problem will become more serious to affect the timing skew. The length-matching routing problem which contains multiple nets and multiple obstacles has been proven to be NP-hard. In this thesis, we propose an approach which is capable of transforming instances of the length-matching routing problem into constraint programming (CP) and into mixed integer linear programming (MILP) problems. As a result, commercial optimization tools can be used to find optimal and suboptimal solutions of those CP and MILP problems. Experimental results show that as high as 365x speed-up can be achieved when a parallel MILP solver is used to solve an instance of the length-matching routing problem on a computer containing 8 processor cores. Compared with a formulation method proposed previously in the literature, our MILP-based formulation method can effectively reduce the number of variables by 10% and the number of constraints by 67%.
|
|---|