Uma abordagem estática para recomendar estruturas de dados Java para melhorar o consumo de energia

Content Provider	Semantic Scholar
Author	Neto, Araújo De, José Benito Fernandes
Copyright Year	2016
Abstract	Java developers have a wide repertoire of collections at their disposal. These collections implement well-known abstractions such as lists, sets, queues, and maps. In addition, the Java language has a rich framework of collections that can be used in parallel by multiple threads running without compromising the correctness of the program. Previous work has shown that these collections have different characteristics in terms of performance, scalability, concurrency control, and energy consumption. In particular, Pinto et al. (2016) investigated the impact of using different thread safe collections on the energy consumption of an application. Among the findings, this study found that different operations of the same implementation have different characteristics, in terms of energy consumption, and the same applies to different implementations of the same abstraction. Since both dimensions, the operations and the collection implementation, impact energy, this work aims to improve the energy efficiency of concurrent applications that do intense use of collections by automatically recommending potentially more efficient collection implementations in a given context. Our approach leverages static analysis to collect information about how applications use thread-safe collections from Java and, based on the energy usage profiles of the operations of these collections, make informed recommendations that potentially save energy. We have developed CECOtool, a tool that implements this approach. Usage of the tool does not require the application under analysis to be executed. To evaluate the approach, we used well-known benchmarks based on two real-world applications. Using this approach we observed a reduction in energy consumption of up to 4.37%. To the best of our knowledge, this is the first work aiming to reduce the energy consumption of an application without the need to run it.
File Format	PDF HTM / HTML
Alternate Webpage(s)	https://repositorio.ufpe.br/bitstream/123456789/24526/1/DISSERTA%C3%87%C3%83O%20Jos%C3%A9%20Benito%20Fernandes%20de%20Ara%C3%BAjo%20Neto.pdf
Language	English
Access Restriction	Open
Content Type	Text
Resource Type	Article