A control-theoretic approach to dynamic voltage scheduling

Varma, Ankush; Ganesh, Brinda; Sen, Mainak; Choudhury, Suchismita Roy; Srinivasan, Lakshmi; Jacob, Bruce

doi:10.1145/951710.951744

Cited by 67 publications

(30 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Varm [8] proposed the control theory method applied to dynamic voltage scheduling, and achieved good results. This paper puts forward SMBPP strategy referencing the basic idea: linear combination of the forecast value of the historical idle period, the actual value of the idle period and the idle period of the task equipment utilization rate is used to predict the next idle period length.…”

Section: Smbpp For Dpmmentioning

confidence: 99%

System Message Based Predict Policy for Dynamic Power Management

Yang

et al. 2017

Proceedings of the 3rd International Conference on Wireless Communication and Sensor Networks (WCSN 2016)

View full text Add to dashboard Cite

Abstract-According to the traditional dynamic power management random policy can't also pointed out that moment of decision and state transition, dynamic power management improvement algorithm, SMBPP(System message based predict policy) is presented in this paper. Firstly, the existing dynamic power management policy algorithm which neglects the application characteristics of the workload is introduced. Secondly, based on the information of the system, the equipment utilization rate of the task is established, and the distribution is updated according to the actual interval time. Thirdly, the predict policy based on task equipment utilization is proposed, and the influence of the strategy parameters on the system sensitivity is analyzed.. Finally, the algorithm is used in wind power bearing state monitoring device. Experimental results show that with the performance constrain, the algorithm pointed out moment of decision and state transition, more stable and more effectively reduces the power consumption.Keywords-DPM; equipment utilization rate; predict policy. I INTRODUCTIONWith the development of semiconductor technology and the improvement of embedded device performance, power consumption on the system performance constraints becomes focus of embedded system research [1]. At this stage, low-power system design in general policy including dynamic power management (DPM) and dynamic voltage scaling (DVS), usually as a kind of DPM.DPM policy in general can be divided into three categories: (a) Timeout policy, the basic idea is in the load after a timeout threshold will load the device into a low power state, including fixed timeout threshold method and adaptive timeout threshold policy [2]. (b) Predict policy, its essence is to forecast on the length of the idle time before making a decision, if the predicted value is large enough directly to PMC (power manageable component) to switch to the corresponding dormancy mode [3]. (c) Stochastic Markov policy, by establishing a Markov model to describe the stochastic behavior of the system device operation request and service, the DPM decision problem as a controlled Markov chain, thus more accurate choice decision time and equipment state transitions [4].Timeout policy has the characteristics of simple principle and applied widely, but it is a kind of exploratory policy. Predict policy, once the threshold of inaccurate prediction, it may be counterproductive to increase power consumption. The study of stochastic Markov policy is very active in recent years. It can be divided into the following categories according to the establishment of the DPM decision model: based on modified Markov stochastic process level DPM model; based on continuous time Markov decision process DPM model; based on discrete time Markov decision process DPM model. DPM model of semi-Markov decision process based on time index, although the parameters of historical events are added, its essence is a kind of random timeout policy. Wu Qi theoretically proves the optimal strategies of DPM is determini...

show abstract

Section: Smbpp For Dpmmentioning

confidence: 99%

System Message Based Predict Policy for Dynamic Power Management

Yang

et al. 2017

Proceedings of the 3rd International Conference on Wireless Communication and Sensor Networks (WCSN 2016)

View full text Add to dashboard Cite

show abstract

“…Varma et al in [147] choose ProportionalIntegral-Derivative (PID) controllers to determine voltage of systems dealing with the workload not accurately known in advance and interpreted the meaning of the discrete PID equation terms with regards to dynamic workloads. They proposed a heuristic to predict the future system load based on the workload rate of change, leading to significant energy reduction.…”

Section: Related Work 47mentioning

confidence: 99%

Dynamic Resource Allocation in Embedded, High-Performance and Cloud Computing

Indrusiak

Dziurzański

Singh

2016

Dynamic Resource Allocation in Embedded, High-Performance and Cloud Computing

View full text Add to dashboard Cite

The "River Publishers Series in Information Science and Technology" covers research which ushers the 21st Century into an Internet and multimedia era. Multimedia means the theory and application of filtering, coding, estimating, analyzing, detecting and recognizing, synthesizing, classifying, recording, and reproducing signals by digital and/or analog devices or techniques, while the scope of "signal" includes audio, video, speech, image, musical, multimedia, data/content, geophysical, sonar/radar, bio/medical, sensation, etc. Networking suggests transportation of such multimedia contents among nodes in communication and/or computer networks, to facilitate the ultimate Internet.Theory, technologies, protocols and standards, applications/services, practice and implementation of wired/wireless networking are all within the scope of this series. Based on network and communication science, we further extend the scope for 21st Century life through the knowledge in robotics, machine learning, embedded systems, cognitive science, pattern recognition, quantum/biological/molecular computation and information processing, biology, ecology, social science and economics, user behaviors and interface, and applications to health and society advance.Books published in the series include research monographs, edited volumes, handbooks and textbooks. The books provide professionals, researchers, educators, and advanced students in the field with an invaluable insight into the latest research and developments.

show abstract

“…[38], and controltheoretic approaches [35]. J. Li et al [26] proposed a run-time voltage/frequency and core-scaling scheduling algorithm that minimizes the power consumption of general-purpose chip multiprocessors within a performance constraint.…”

Section: Related Workmentioning

confidence: 99%

Coordinated energy management in heterogeneous processors

Paul

Ravi

Manne

et al. 2013

Proceedings of the International Conference on High Performance Computing, Networking, Storage and Analysis

View full text Add to dashboard Cite

This paper examines energy management in a heterogeneous processor consisting of an integrated CPU-GPU for highperformance computing (HPC) applications. Energy management for HPC applications is challenged by their uncompromising performance requirements and complicated by the need for coordinating energy management across distinct core types -a new and less understood problem.We examine the intra-node CPU-GPU frequency sensitivity of HPC applications on tightly coupled CPU-GPU architectures as the first step in understanding power and performance optimization for a heterogeneous multi-node HPC system. The insights from this analysis form the basis of a coordinated energy management scheme, called DynaCo, for integrated CPU-GPU architectures. We implement DynaCo on a modern heterogeneous processor and compare its performance to a state-of-the-art power-and performance-management algorithm. DynaCo improves measured average energy-delay squared (ED^2) product by up to 30% with less than 2% average performance loss across several exascale and other HPC workloads.

show abstract

A control-theoretic approach to dynamic voltage scheduling

Cited by 67 publications

References 7 publications

System Message Based Predict Policy for Dynamic Power Management

System Message Based Predict Policy for Dynamic Power Management

Dynamic Resource Allocation in Embedded, High-Performance and Cloud Computing

Coordinated energy management in heterogeneous processors

Contact Info

Product

Resources

About