Per-task energy metering and accounting in the multicore era

Energy has become arguably the most expensive resource in a computing system. As multi-core processors are the preferred processing platform across different computing domains, measuring the energy usage draws vast attention. In this thesis, for the first time, we formalize the need for per-task energy measurement in multicore by establishing a two-fold concept: per-task energy metering and sensible energy accounting. The former, for a task running in a multi-core system, provides estimates on the actual energy consumption corresponding to its resource usage. The latter provides estimates on the energy the task would have consumed running in isolation with a given fraction of the shared resources. We have shown how these two concepts can be applied to the main components of a computing system: the processor and the memory system