Power Profiling for Low Power Optimization Using New Computing Model

Power profiling helps to do power-aware synthesis and power optimization for digital system. Traditional profiling method involves inserting probes and feeding the program with a large amount of testing data to collect statistical information. The cost of this method is high so the use of which is limited. In this paper, a power profiling method that is based on a new computing model - multiple-dimensional computing model, is presented. It helps to collect information such as signal probability and transition density in a single run and the information is kept accurate enough. The method is easy to be implemented and incorporated into synthesis and optimization system