Enhancement and analysis of a simple and efficient VLSI model

To estimate the delay and area of a VLSI circuit before its implementation demands a cost and delay model which should be simple enough to allow a quick and therefore cost-effective paper and pencil analysis. However, common VLSI models are either to complicated resulting in long simulation times or do not consider wire-delays, neglecting their importance in sub-micron and nanotechnologies. Paul and Seidel introduced such a technologyindependent model in 1998 [Paul, Wolfgang J.; Seidel, Peter-Michael: On the Complexity of Booth Recoding, 1998]. Based on this model, we will introduce a new VLSI cost and delay model. It consists of two parts – a basic technology-independent part based on reasonable enhancements of the Paul and Seidel model and a technology- specific part, where a specific technology can be implied to get the effective circuit cost and delay. We analyze the accuracy by comparing the cost and performance of synthesized adder circuits with the results derived from the model. The comparison reveals that the model closely matches the cost and delay of the standard-cell and freely placeable element-based implementations in the majority of the tested circuits. Therefore, it could enhance existing conventional VLSI models while retaining their simplicity.