Abstract: Circuit design is an area of obvious importance, and a great deal of activity. Due to the increasing manufacturing costs of cutting-edge circuits, there is a strong need for better design tools—this allows for improvements in circuit performance without changes to manufacturing technology.
The focus of this dissertation is on the physical design of integrated circuits; there is an emphasis on circuit placement. By improving the quality of a circuit placement, we directly impact the speed, power, and size of the device.
Key research contributions of this work are the following. First, a novel k-way partitioning method has been developed; this allows for large values of k, and results in improvements in the quality of circuit placements without increases to the run times for the tool. Second, the interaction between cut directions in bisection based placement has been made plain; what had been pursued previously in an ad hoc manner now has a firm theoretic foundation. Using our techniques, a sophisticated placement tool from the University of Michigan and UCLA obtained a 6% improvement in performance. Third, prior research on Steiner minimal trees has been extended to a routing model that more accurately represents “realistic” circuit applications.
A unifying theme for this work is that all decisions in circuit design are inter-related. Where previously the design flow was decomposed into independent steps, we show that by careful analysis of the problem, the interactions between steps can be controlled and optimized.
Publication Year: 2003
Publication Date: 2003-01-01
Language: en
Type: article
Access and Citation
Cited By Count: 6
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot