Modeling Performance and Power Consumption of Energy-Aware Software on a Highly Adaptive Computing System (A04)

• Subject Area: Computer Architecture, Embedded and Massively Parallel Systems
• Term: from 2011 to 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 164481002

In order to achieve high energy efficiency in an adaptive system, it is essential to understand the effect of the interactions between hardware and software on the overall performance and energy consumption of the system. With the focus on those interactions, this project advances the methodologies for predicting the performance and energy costs of a system to suit the vision of the HAEC architecture. For our simulation-based approach, we create system component models based on a sophisticated energy measurement environment and utilize a discrete-event simulation environment comprising models of individual parts of the HAEC Box design, to predict the behavior of the executed software. In Phase III, this project will focus on the impact of adaptivity decisions based on the dynamic state of the HAEC Box. We will enhance the previously developed simulator to execute HAEC software concurrently to the simulation of the HAEC Box. This runtime simulation environment allows us to study the impact of adaptivity decisions in a greater context.

• DFG Programme: Collaborative Research Centres
• Subproject of: SFB 912: HAEC - Highly Adaptive Energy-Efficient Computing
• Applicant Institution: Technische Universität Dresden
• Project Head: Professor Dr. Wolfgang E. Nagel