Energy efficient scheduling for multithreaded programs on general-purpose processors

Abstract

This paper investigates the effectiveness of Time Aggregation Scheduler (TAS) for commodity platforms from the view point of energy saving. TAS can aggregate the execution of runnable sibling threads, and decrease the number of internal events that have negative impact on energy consumption in executing multithreaded applications. We compare energy consumption in running multithreaded applications of DaCapo Java benchmarks when using Linux default Completely Fair Scheduler, and TAS. The experimental results indicate that TAS can reduce the amount of system level energy consumption, as well as the quantity of internal events and the execution time.