Energy Consumption Measurement of C/C++ Programs Using Clang Tooling

While in the previous century computer manufacturers and software developers primary and single goal was to produce very fast computers and software systems, in this century this has changed: the widespread use of nonwired but powerful computer devices is making battery consumption/lifetime the bottleneck for both manufacturers and software developers. Unfortunately there is no software engineering discipline providing techniques and tools to help software developers to analyze, understand and optimize the energy consumption of their software! As a consequence, if a developer notices that his/her software is responsible for a large battery drain, he/she gets no support from the language/compiler he/she is using. The hardware manufacturers have already realized this concern and much work in terms of optimizing energy consumption by optimizing the hardware has been done. Unfortunately, the programming language and software engineering communities have not yet completely realize that bottleneck, and as consequence, there is little support for software developers to reason about energy consumption of their software. Although is the hardware that consumes energy, the software can greatly influence such consumption [Bener et al. 2014], very much like a driver that operates a car influences its fuel consumption. In this paper we introduce an automated instrumentation-based method to measure the process level energy consumption for C/C++ programs. The source code is compiled by our Clang tooling based compiler to produce an instrumented code. The generated executable will measure the energy con-