Toward an evaluation infrastructure for power and energy optimizations

Execution-driven simulators are often used for power/energy and performance evaluation. Simula-tors can provide semantic details but they provide insufcient speed and accuracy for compiler and OS re-search. Physical measurement is fast and objective but lacks a semantic connection between the mea-surement result and the evaluated program. The objective of our research is to bring together the ad-vantages of simulation and physical measurement to build an infrastructure for power and energy optimiza-tion. Power and energy behavior is obtained through physical measurement. Simulation is used for ob-serving the connection between power and energy behavior and the evaluated program. Our prelimi-nary results demonstrate the ability of this infrastruc-ture to capture detailed power behavior of any region of a program. To simplify the power/energy evalua-tion of programs with long execution times and over-come the limitation of physical devices, we propose using the SimPoints methodology developed by re-searchers at UC San Diego to nd representative slices of a program. Through simulation, we vali-date the feasibility of the SimPoint idea in simplifying power/energy evaluation. We expect that this infrastruc-ture will help researchers in OS/compiler power/energy optimization to evaluate their optimizations more ef-ciently and observe more optimization opportuni-ties. 1