| |
| Ignoring some cell overlaps, global placement computes the bestposition for each cell to minimize some cost metric (e.g., totalwirelength, density overflow). It is a crucial step in very largescale integration(VLSI) physical design, since it affectsroutability, performance, and power consumption of a circuit. Inthis paper, we propose an Augmented Lagrangian method to solve theVLSI global placement. In this method, a cautiousdynamic density weight increasing strategy is used to balance thewirelength and density constraint. We incorporated our method intoNTUplace3's global placement framework,and tested it on the IBM mixed-size benchmark circuits.Experimental results show that it obtains high-quality results in areasonable running time. |
|
| Keywords:nonlinear optimization; VLSI; global placement; augmented lagrangian method |
|
