Abstract
We propose an architecture-level power estimation model for a configurable multiplier. The model can be automatically scaled with respect to different configurations of multiplier. Large amount of characterization effort is reduced. The user can directly feed the input sequences of a configurable multiplier to this model and derive the estimated average power consumption. This model can assist the designer to optimize the power consumption of multiplication operation at architecture-level abstraction. . The proposed model can efficiently estimate the power consumption at the early design stage, and yields average estimation errors around 32%. The standard deviation of the average error is around 14%.