GreenEr Mobile Systems by Cross LAyer Integrated energy Management
GreenEr Mobile Systems by Cross LAyer Integrated energy Management
ERA-Net: CHIST ERA
Disciplines
Computer Sciences (100%)
Keywords
-
Green IT,
Parallel Processing,
Energy Management,
Energy Optimization,
Multicore platforms,
Customizable Hardware
Personal computing currently faces a rapid trend from desktop machines towards mobile services, accessed via tablets, smart phones, and similar terminal devices. With respect to compute power, today`s handheld devices are similar to Cray-2 supercomputers from the 1980s. Due to higher computational load (e.g. via multimedia apps) and the variety of radio interfaces (such as WiFi, 3G, and LTE), modern terminals are getting more and more energy hungry. For instance, a single UMTS upload or a video recording process on today`s smart phones may consume as much as 1.5 Watts, i.e. roughly 50% of the maximal device power. In the near future, higher data rates and traffic, advanced media codecs, and graphics applications will ask for more energy than the battery can deliver. At the same time, the power density limit might lead to a significant share of "Dark Silicon" at 22nm CMOS and below. Obviously, disruptive energy optimizations are required that go well beyond traditional technologies like DVFS (dynamic voltage and frequency scaling) and power-down of temporarily unused components. The GEMSCLAIM project aims at novel approaches for reducing the "greed for energy", thereby improving the user experience and enabling new opportunities for mobile computing. The focus is on three novel approaches: (1) cross layer energy optimization, ranging from the compiler over the operating system down to the target HW platform, (2) efficient programming support for energy-optimized heterogeneous Multicore platforms based on energy-aware service level agreements (SLAs) and energy-sensitive tunable parameters, and (3) introducing energy awareness into Virtual Platforms for the purpose of dynamically customizing the HW architecture for energy optimization and online energy monitoring and accounting. GEMSCLAIM will provide new methodologies and tools in these domains and will quantify the potential energy savings via benchmarks and a HW platform prototype.
Mobile computing devices such as laptops, tablets and smartphones are becoming more widespread. The performance that these devices offer is increasing at a steady pace, with increasing number of processors powering contemporary smartphones. However, mobility comes at a price: energy is a scarce resource on these devices, especially with power-hungry media consumption constituting a major use case. Furthermore, mobile computing is not the sole field where energy is increasingly relevant - the tremendous increase in power consumption by performance-oriented servers has made power budgeting unavoidable in HPC as well. Several solutions have been proposed to address the energy problem on different levels. While system-level approaches have been explored to a certain extent, we believe there still exists a lot of room for improvement on the application level since no power management interface that can be considered complete, generic and easy to use has been proposed so far. GEMSCLAIM explored application-level energy saving opportunities through the specification, implementation and evaluation of OpenMPE an OpenMP extension for Energy. OpenMPE adds new directives and clauses to enable per-region customization of multiple optimization objectives and tunable parameters. GEMSCLAIM resulted in three major contributions: A novel API for application-level energy-aware programming, allowing programmers to expose energy saving opportunities through i) characterizing application behavior by providing a semantic program region structure, ii) setting per-code region multi-objective goals, and iii) exposing application-level tunable parameters. A compilation and runtime system supporting OpenMPE. The runtime system exploits opportunities exposed by programmers through several techniques such as dynamic voltage and frequency scaling, dynamic concurrency throttling (DCT), and application-level content adaptation. An empirical evaluation of the effectiveness of OpenMPE. A video codec reference implementation has been enriched with OpenMPE and benchmarked on a desktop and mobile platform. GEMSCLAIM thus contributed important research results that can lead to substantial improvement in saving energy for mobile systems.
- Universität Innsbruck - 100%
- Rainer Leupers, RWTH Aachen - Germany
- Dimitros Nikolopoulos, Queens University Belfast - Ireland
- Marius Marcu, Universitatea de Vest din Timisoara - Romania
Research Output
- 167 Citations
- 12 Publications
-
2012
Title A Multi-Objective Auto-Tuning Framework for Parallel Codes. Type Conference Proceeding Abstract Author Jordan H Conference SC '12, November 11-15, 2012 -
2012
Title A Multi-Objective Auto-Tuning Framework for Parallel Codes DOI 10.1109/sc.2012.7 Type Conference Proceeding Abstract Author Jordan H Pages 1-12 -
2014
Title Compiler multiversioning for automatic task granularity control DOI 10.1002/cpe.3302 Type Journal Article Author Thoman P Journal Concurrency and Computation: Practice and Experience Pages 2367-2385 -
2014
Title Multi-Objective Auto-Tuning with Insieme: Optimization and Trade-Off Analysis for Time, Energy and Resource Usage DOI 10.1007/978-3-319-09873-9_8 Type Book Chapter Author Gschwandtner P Publisher Springer Nature Pages 87-98 Link Publication -
2014
Title On the potential of significance-driven execution for energy-aware HPC DOI 10.1007/s00450-014-0265-9 Type Journal Article Author Gschwandtner P Journal Computer Science - Research and Development Pages 197-206 -
2015
Title Energy Prediction of OpenMP Applications Using Random Forest Modeling Approach DOI 10.1109/ipdpsw.2015.12 Type Conference Proceeding Abstract Author Benedict S Pages 1251-1260 -
2015
Title On the Quality of Implementation of the C++ 11 Thread Support Library DOI 10.1109/pdp.2015.33 Type Conference Proceeding Abstract Author Thoman P Pages 94-98 -
2015
Title Application-Level Energy Awareness for OpenMP DOI 10.1007/978-3-319-24595-9_16 Type Book Chapter Author Alessi F Publisher Springer Nature Pages 219-232 -
2015
Title Optimizing Task Parallelism with Library-Semantics-Aware Compilation DOI 10.1007/978-3-662-48096-0_19 Type Book Chapter Author Thoman P Publisher Springer Nature Pages 237-249 Link Publication -
2013
Title INSPIRE: The Insieme Parallel Intermediate Representation. Type Conference Proceeding Abstract Author Fahringer T Et Al Conference PACT 2013, September 7-11, 2013. -
2013
Title INSPIRE The Insieme Parallel Intermediate Representation DOI 10.1109/pact.2013.6618799 Type Conference Proceeding Abstract Author Jordan H Pages 7-17 -
2013
Title Adaptive Granularity Control in Task Parallel Programs Using Multiversioning DOI 10.1007/978-3-642-40047-6_19 Type Book Chapter Author Thoman P Publisher Springer Nature Pages 164-177