In deep submicron fabrication technology, transistors can now switch much faster, but wire resistances are now larger, and delay due to wires can exceed gate delay. Consequently, the interconnect delay is the dominant factor in the construction of wire routing in very large scale integrated (VLSI) circuits, which today, has feature dimensions in the nanometer range. Today, the state-of-the-art circuit design involves as much the engineering of the wires as the design of transistors. Hence, a successful VLSI design today depends heavily on a successful interconnect design. An effective approach for reducing the interconnect delay is buffer insertion (van Ginneken, 1990). In this method, a wire is divided into segments with a buffer inserted between the segments (Cong et al., 1996). Traditionally, buffer insertion is a post-layout optimization technique, implying that the routing paths are first found, and then buffers are inserted in these paths. However, today?s VLSI designs typically apply some form of design reuse utilizing pre-designed cells, or macro blocks.
展开▼
