Late Breaking Results: An Effective Legalization Algorithm for Heterogeneous FPGAs with Complex Constraints
TimeWednesday, December 8th6:00pm - 7:00pm PST
LocationLevel 2 - Lobby
Event Type
Late Breaking Results Poster
Networking Reception
Virtual Programs
Presented In-Person
DescriptionAn ideal legalization method should remove all overlaps and satisfy the design rules (clock, routability, location and chain constraints, etc.) while minimizing both the average and maximum movements. This paper first identifies a virtual range to cluster an instance with the location constraint and formulate minimum cost integer linear programming. We use an adaptive algorithm to deal with chain-aware legalization for better quality and runtime trade-offs. Finally, a cell refinement technique is proposed to improve the solution quality further. Experimental results show that our algorithm achieves the best average and maximum cell movements, smallest routed wirelength, and shortest routing runtime.