Refactoring multicore applications towards energy efficiency

Abstract

Great strides have been made to increase the energy efficiency of hardware, data center facilities, and network infrastructure. However, in any computer system, it is software that directs much of the activity of the hardware. Moreover, multicore processors have become ubiquitous, mainly because their multiple benefits, especially enhanced performance for multi-threaded and computeintensive applications. Nonetheless, there are few studies addressing the topic of restructuring multicore applications to consume less energy and even fewer that leverage developer expertise to achieve that goal. In this thesis we present a brief background study for refactoring multicore applications in order to improve both performance and energy consumption. The idea consists in proposing a catalog of refactorings targeting some languages of the JVM platform.

Energy and run-time performance practices in software engineering

“…Developing software to be energy-efficient is a demanding and challenging task because of the lack of tooling support, practices, and guidelines [134,115,93]. Researchers have carried out studies on different aspects and granularity of software artifacts to determine the energy consumption of data structures [136,63,124,127,172], different programming languages [4,143,87], multi-threaded applications [135,132,131], coding practices [166,89,153,152,138], and so on. Although there is a large and consistent body of 1.2.…”