Leakage power reduction is extremely important in the design of scaled CMOS logic circuits. The dominant leakage mechanism of such circuits is the gate leakage. This paper first describes a fast leakage estimation technique based on biasing states for both gate leakage and sub-threshold leakage. Next, it describes a leakage reduction method based on the selective insertion of control points. Simulations on a set of examples show that this method results in the average leakage being 28.7% of the leakage of the baseline circuit whose inputs have already been subjected to the minimum leakage vector (MLV).
展开▼