Energy and peak-current per-cycle estimation at RTL

We present novel macromodel ing techniques for estimating the energy dissi pated and peak current drawn in a logic cir cuit for every input vector pair we call this the energy per cycle and peak current per cycle respec tively The macromodels are based on clas sifying the input vector pairs on the basis of their Hamming distances and using a di erent equation based macromodel for every Hamming distance The variables of our macromodel are the zero delay transition counts at three logic levels inside the circuit We present an auto matic characterization process by which such macromodels can be constructed The energy per cycle macromodel provides a transient en ergy waveform and can also be used to esti mate the moving average energy over any time window whereas peak current per cycle macro model provides peak current which can be used for studying IR drop problems Some key fea tures of this technique are the models are compact linear in the number of inputs they can be used for any input sequence and the characterization is automatic and requires no user intervention These approaches have been implemented and models have been built and tested for many circuits The average er rors observed in estimating the energy per cycle and peak current per cycle are under The energy per cycle model can also be used to This research was supported in part by the National Science Foundation NSF MIP and by the Semiconductor Research Corporation SRC DJ and TJ with technical mentorship from Texas Instruments IBM and Intel measure the long term average power with an observed error of under on average