Power Consumption Characterization of Synthetic Benchmarks in Multicores Jonathan Mura˜ na 1(B ) , Sergio Nesmachnow 1 , Santiago Iturriaga 1 , and Andrei Tchernykh 2 1 Universidad de la Rep´ ublica, Montevideo, Uruguay {jmurana,sergion,siturria}@fing.edu.uy 2 CICESE Research Center, Ensenada, Baja California, Mexico chernykh@cicese.mx Abstract. This article presents an empirical evaluation of power con- sumption of synthetic benchmarks in multicore computing systems. The study aims at providing an insight of the main power consumption char- acteristics of diﬀerent applications when executing over current high performance computing servers. Three types of applications are studied executing individually and simultaneously on the same server. Intel and AMD architectures are studied in an experimental setting for evaluating the overall power consumption of each application. The main results indi- cate the power consumption behavior has a strong dependency with the type of application. An additional performance analysis shows that the best load of the server regarding energy eﬃciency depends on the type of the applications, with eﬃciency decreasing in heavily loaded situa- tions. These results allow characterizing applications according to power consumption, eﬃciency, and resource sharing, and provide useful infor- mation for resource management and scheduling policies. Keywords: Green computing · Energy eﬃciency · Multicores Computing eﬃciency 1 Introduction Nowadays, data centers are key infrastructures for executing industrial and sci- entiﬁc applications. Data centers have become highly popular for providing stor- age, computing power, middleware software, and others information technology (IT) utilities, available to researchers with ubiquitous access [3]. However, their energy eﬃciency has become a major concern in recent years, having a signif- icant impact on monetary cost, environment, and guarantees for service-level agreements (SLA) [4]. The main sources of power consumption in data centers are the computa- tional resources and the cooling system [13]. When focusing on power consump- tion due to resource utilization, several techniques for hardware and software optimization can be applied to improve energy eﬃciency. Software characteri- zation techniques [1] are used to determine features that are useful to analyze c  Springer International Publishing AG 2018 E. Mocskos and S. Nesmachnow (Eds.): CARLA 2017, CCIS 796, pp. 21–37, 2018. https://doi.org/10.1007/978-3-319-73353-1_2