REGRESSION-MODELS FOR BEHAVIORAL POWER ESTIMATION

Behavioral power estimation is required to help the designer in making important architectural choices. In this work Re propose an accurate and general behavioral power modeling approach especially suited for s ynthesis-based design flows making use of a library of hard macros imp lementing behavioral operators. Power dissipation models are pre-chara cterized and back-annotated in a preliminary step. Accurate informatio n on the power dissipation of the used macros can then be collected du ring behavioral simulation of the synthesized circuit. Our characteriz ation and modeling methodology is based on the theory of linear regres sion. Optimal linear power models are obtained with methods of least s quares fitting and their generalization to a recursive procedure calle d tree regression. The regression models can be used for pattern-based dynamic power simulation and for probabilistic static pourer estimati on as well. Our behavioral simulator is integrated within PPP, a multi level simulation engine for power estimation fully compatible with Ver ilog XL.