Work in bootstrapping energy consumption exists in academia; for example Pathak et al[0]. The changes in 10.10.4 could be an implementation of something along t

Work in bootstrapping energy consumption exists in academia; for example Pathak et al[0]. The changes in 10.10.4 could be an implementation of something along the same lines, by fitting a model to the measured power consumption. As comex notes in the comments on the article, there appears to be a table of weights for various features, presumably obtained by regression. [0]: http://research.microsoft.com/en-us/people/mzh/eurosys-2012.pdf http://research.microsoft.com/en-us/people/mzh/eurosys-2012....