Search-based placement methods have long been used for placing integrated circuits targeting the field programmable gate array (FPGA) and standard cell design styles. Such methods offer the potential for high-quality solutions but often come at the cost of long run-times compared to alternative methods.
This dissertation examines strategies for enhancing local search heuristics---and in particular, simulated annealing---through the application of directed moves. These moves help to guide a search-based optimizer by focusing efforts on states which are most likely to yield productive improvement, effectively pruning the size of the search space.
The engineering theory and implementation details of directed moves are discussed in the context of both field programmable gate array and standard cell designs. This work explores the ways in which such moves can be used to improve the quality of FPGA placements, improve the robustness of floorplan repair and legalization methods for mixed-size standard cell designs, and enhance the quality of detailed placement for standard cell circuits. The analysis presented herein confirms the validity and efficacy of directed moves, and supports the use of such heuristics within various optimization frameworks.
Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:OWTU.10012/4528 |
Date | January 2009 |
Creators | Vorwerk, Kristofer |
Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
Language | English |
Detected Language | English |
Type | Thesis or Dissertation |
Page generated in 0.002 seconds