PORTAL: Power- and Thermal- Aware Computing
sustainable computing, designers of computer systems, ranging from
small mobile devices to massive
data centers, emphasize obtaining a sustainable level of environmental
and societal costs as the first-order
design principle. These cost may result from the computer systems
manufacturing, operation, and disposal.
Among all the factors in sustainable computing that contribute to
system operational costs, power
and thermal dissipation are fundamental in modern computer-controlled
systems. To understand the relationship between
power & thermal dissipation and software applications, hardware
manufacturers have provided capabilities
such as dynamic voltage and frequency scaling (DVFS) to permit a
in the power dissipation. However, these capabilities may often have a
negative and (sometimes)
unpredictable effect upon the runtime performance of the software
Due to inherent
physical uncertainties and environmental dynamics as well as the
power and thermal dissipation in a system, empirical measurement study
is crucial for understanding
the system behavior. For instance, precise power and temperature
measurements are required to fully
evaluate the effectiveness of power-saving or temperature-reducing
software design. Given this fundamental gap in software design for
sustainable computing, we propose the development of an experimental
infrastructure POweR and Thermal Aware computing Laboratory –
that captures the challenges and complexities of power and
thermal–aware analysis of modern computer
systems, including software power behavior analysis on different
platforms and thermal effects under
real–time system constrains. In PORTAL, we will develop software
profiling techniques to assist developers in better understanding and
optimizing the power dissipation associated with source
code, under different hardware configurations. We will also formally
introduce the software applications′ real–time requirements
into dynamic thermal management (DTM).
In combining theoretical results and experimental properties of modern
computer systems′ power dissipation
and thermal effects, PORTAL is expected to enable a broad range of
research activities in the
investigators′ groups and in the broad research community.
Uniquely offering the capability of finegrained
instrumenting, measuring, controlling, and correlating power and
thermal dissipation, PORTAL
will enable diverse research topics in investigators′ group,
which include but not limit to the following:
(1) system software power estimation and optimization based on
mathematically power models and fieldexperiments;
(2) real-time guarantees under thermal constrains according to control
The basis of power–aware design is accurate, verbose, and
real-time power estimation.
A better understanding of the power dissipation of a system will enable
more energy saving opportunities
. Despite years of research efforts on computer system power
estimation, most existing approaches
either do not expose sufficient information to end–users and
software developers or lack the
consideration of platform-dependent factors. With more detailed
information on power dissipation of workload, software developers will
be enabled to leverage the algorithms
and implementations to fulfill performance and power dissipation
requirements. Thus, there
exists a gap between the software design and runtime software
dissipation. To bridge the gap, our major
objective is exposing more software power dissipation information to
operating systems, end users, and developers
to enable greater energy-efficient design.
The advent and ubiquity of hardware technology such as DVFS has enabled
software system designers to address temperature constraints via DTM.
Typical DTM techniques
involve determining opportune intervals, with respect to system
performance constraints, during which
the system′s execution may be reduced or stalled. The intervals
of reduced execution permit the system
to dissipate heat to the environment. For real–time systems, DTM
techniques must ensure that system
temporal constraints are not violated. Most prior work on DTM for
real–time systems has been developed under
the assumption that operating environment is static. However, this
assumption is unlikely to hold in
many real–world scenarios. Thus, we propose to address this
objective by developing a framework for thermal–aware control of
hard–real–time systems which includes methodology,
analysis, and tools required to design and implement real–time
systems that behave predictably under unexpected thermal changes in the
Hui Chen (graduated, now at Amazon)
Pradeep Hettiarachchi (graduated, now at GM)
Youhuizi Li (graduated, now at Hangdiang University)
- Corey Tessler and Nathan Fisher, BUNDLE: Real-Time Multi-Threaded Scheduling to Reduce Cache Contention, IEEE Real-Time Systems Symposium (RTSS), Porto, Portugal, Nov. 29-Dec. 2, 2016.
- Congfeng Jiang, Dongyang Ou, Yumei Wang, Xindong You, Jilin Zhang, Jian Wan, Bing Luo and Weisong Shi, Energy Efficiency Comparison of Hypervisors, in Proceedings of the 7th International Green and Sustainable Computing (IGSC), Hangzhou, China. Nov. 7-9, 2016.
- Bo Peng, Nathan Fisher, and Thidapat Chantem, MILP-based Deadline Assignment for End-to-End Flows in Distributed Real-Time Systems, Proceedings of the 24th International Conference on Real-Time Networks and Systems (RTNS), Brest, France, Oct 19-21, 2016,
- Bing Luo, Shinan Wang, Weisong Shi and Yanfei He, eCope: Workload-aware Elastic Customization for Power Efficiency of High-End Servers, IEEE Transactions on Cloud Computing, Vol. 4, No. 2, April-June 2016.
- Guosai Wang, Shuhao Wang, Bing Luo, Weisong Shi, Yinghang Zhu, Wenjun Yang, Dianming Hu, Longbo Huang, Xin Jin, Wei Xu, Increasing Large-Scale Data Center Capacity by Statistical Power Control, in Proceedings of EuroSys 2016. London UK, April 18-21, 2016.
- Quan Zhang and Weisong Shi, Energy-Efficient Workload Placement in Enterprise Datacenters, IEEE Computer Magazine, Vol. 29, No. 2, pp. 46-52, 2016.
- Shinan Wang, Bing Luo, Weisong Shi and Devesh Tiwari, Application Configuration Selection for Energy-Efficient Execution on Multicore Systems, Journal of Parallel and Distributed Computing (JPDC), Vol. 87, January 2016, Pp. 43-54.
- Corey Tessler and Nathan Fisher (2016). Scheduling Multi-Thread Tasks to Reduce Intra-Task Cache Contention. Work-in-Progress Session of IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS), 2016.
Hettiarachchi and Nathan Fisher (2016). A Generalized Design Framework
for Adaptive Real-Time System Resiliency. ACM/SIGAPP Symposium On
Applied Computing. Rome, Italy. (to be submitted)
- Younes Chandarli, Nathan Fisher, and Damien
Time Analysis for Thermal-Aware Real Time Systems Under Fixed Priority Scheduling. IEEE Computer Society
Symposium on Object/Component/Service-Oriented Realtime Distributed Computing
(ISORC). Auckland, New Zealand, April 13-17, 2015.
- Grace Metri, Weisong Shi and Monica Brockmeyer, Energy-Efficiency Comparison of Mobile Platforms and Applications: A Quantitative Approach, in
Proceedings of the Sixteenth Workshop on Mobile Computing Systems and
Applications (ACM HotMobile 2015), Sante Fe, Feb. 12-13, 2015.
Fisher, Masud Ahmed, and Pradeep M. Hettiarachchi. Open Problems in Multi-Modal
Scheduling Theory for Thermal-Resilient Multicore Systems.
Real-Time Scheduling Open Problems Seminar (RTSOPS), Madrid Spain, July 2014.
- Youhuizi Li, Bing Luo, Hui Chen and Weisong Shi, One Charge for One Week: Hype or Reality?,
in Proceedings of the 2014 International Green Computing Conference (IGCC),
Dallas, TX USA, November 3-5, 2014.
Paolillo, Joel Goossens, Nathan Fisher, and Pradeep
Minimization for Parallel Real-Time Systems with Malleable Jobs and Homogeneous
Frequencies. IEEE International Conference on Embedded and
Real-Time Computing Systems and Applications. Chongqing, China. August 20-22, 2014.
Mashayekhy, Mahyar Movahed Nejad, Daniel Grosu, Quan Zhang and Weisong
Shi, Energy-aware Scheduling of MapReduce Jobs for Big Data
Applications, accepted by IEEE Transactions on Parallel and Distributed Systems. September 2014.
Metri, Weisong Shi, Monica Brockmeyer and Abhishek Agrawal, BatteryExtender: An Adaptive User-Guided
Tool for Power Management of
Mobile Devices, in Proceedings of the
2014 ACM International Joint Conference on
Pervasive and Ubiquitous Computing (UbiComp 2014), Seattle, USA, Sep.
13-17, 2014. Best Paper Nominee Award.
- Quan Zhang, Grace Metri, Sudharsan Raghavan and Weisong Shi, RESCUE: An Energy-Aware Scheduler for Cloud Environments,
accepted by Sustainable Computing: Informatics and Systems special
issue on Energy-Aware Resource Management and Scheduling (EARMS),
- Youhuizi Li, Hui Chen and Weisong Shi, Power Behavior Analysis of Mobile
Applications using Bugu, accepted by Sustainable
Systems, July 2014.
Hettiarachchi, Nathan Fisher and Le-Yi Wang, Achieving Thermal
Resiliency for Multicore Hard-Real-Time Systems, Proceedings of the Euromicro Conference on
Real-Time Systems (ECRTS), 2013.
Wang, Weisong Shi, and Bing Luo. CPT: An Energy
Efficiency Model for Multicore Computer Systems, in Proceedings of the 5th of Workshop on
Energy-Efficient Design, in conjuction with ACM/IEEE ISCA-39,
Tel-Aviv, Israel, June 23, 2013.
- Pradeep Hettiarachchi, Nathan Fisher, Masud Ahmed, Le-Yi Wang, Shinan
Wang, and Weisong Shi (2014). A Design and Analysis Framework for
Thermal-Resilient Hard-Real-Time Systems. ACM
Transactions on Embedded Computer Systems. 13 (5s),
Wang, Youhuizi Li, Weisong Shi, Lingjun Fan, and Abhishek
Function-Level Power Analysis using Automatic Instrumentation, in Proceedings of the 3rd International
Conference on Energy-aware Computing (ICEAC), December 3-5,
Chen, Bing Luo, and Weisong Shi.
Anole: A Case for Energy-aware Mobile Application Design, in Proceedings of the 1st International
Workshop on Power-Aware Systems and Architectures. Pittsburgh, PA, USA.September