DEMM

A Dynamic Energy-Saving Mechanism for Multicore Memories

Akbar Sharifi, Wei Ding, Diana Guttman, Hui Zhao, Xulong Tang, Mahmut Kandemir, Chitaranjan Das

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

Since main memory system contributes to a large and increasing fraction of server/datacenter energy consumption, there have been several efforts to reduce its power and energy consumption. DVFS schemes have been used to reduce the memory power, but they come with a performance penalty. In this work, we propose DEMM, an OS-based, high performance DVFS mechanism that reduces memory power by dynamically scaling individual memory channel frequencies/voltages. Our strategy also involves clustering the running applications based on their sensitivities to memory latency, and assigning memory channels to the application clusters. We introduce a new metric called Discrete Misses per Kilo Cycle (DMPKC) to capture the performance sensitivities of the applications to memory frequency modulation. DEMM allows us to save power in the memory system with negligible impact on performance. We demonstrate around 25% savings in the memory system energy and 10% savings in the total system energy, with only a 4% loss in workload performance.

Original languageEnglish (US)
Title of host publicationProceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages210-220
Number of pages11
ISBN (Electronic)9781538627631
DOIs
StatePublished - Nov 13 2017
Event25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017 - Banff, Canada
Duration: Sep 20 2017Sep 22 2017

Publication series

NameProceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017

Other

Other25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017
CountryCanada
CityBanff
Period9/20/179/22/17

Fingerprint

Energy Saving
Energy conservation
Data storage equipment
Energy Consumption
Energy utilization
Frequency Modulation
Frequency modulation
Energy
Power Consumption
Workload
Penalty
Latency
Computer systems
Electric power utilization
Servers
Server
High Performance
Voltage
Clustering
Scaling

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications
  • Hardware and Architecture
  • Modeling and Simulation

Cite this

Sharifi, A., Ding, W., Guttman, D., Zhao, H., Tang, X., Kandemir, M., & Das, C. (2017). DEMM: A Dynamic Energy-Saving Mechanism for Multicore Memories. In Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017 (pp. 210-220). [8107447] (Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/MASCOTS.2017.16
Sharifi, Akbar ; Ding, Wei ; Guttman, Diana ; Zhao, Hui ; Tang, Xulong ; Kandemir, Mahmut ; Das, Chitaranjan. / DEMM : A Dynamic Energy-Saving Mechanism for Multicore Memories. Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 210-220 (Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017).
@inproceedings{2ac270adde364ac09da9dd2921534d82,
title = "DEMM: A Dynamic Energy-Saving Mechanism for Multicore Memories",
abstract = "Since main memory system contributes to a large and increasing fraction of server/datacenter energy consumption, there have been several efforts to reduce its power and energy consumption. DVFS schemes have been used to reduce the memory power, but they come with a performance penalty. In this work, we propose DEMM, an OS-based, high performance DVFS mechanism that reduces memory power by dynamically scaling individual memory channel frequencies/voltages. Our strategy also involves clustering the running applications based on their sensitivities to memory latency, and assigning memory channels to the application clusters. We introduce a new metric called Discrete Misses per Kilo Cycle (DMPKC) to capture the performance sensitivities of the applications to memory frequency modulation. DEMM allows us to save power in the memory system with negligible impact on performance. We demonstrate around 25{\%} savings in the memory system energy and 10{\%} savings in the total system energy, with only a 4{\%} loss in workload performance.",
author = "Akbar Sharifi and Wei Ding and Diana Guttman and Hui Zhao and Xulong Tang and Mahmut Kandemir and Chitaranjan Das",
year = "2017",
month = "11",
day = "13",
doi = "10.1109/MASCOTS.2017.16",
language = "English (US)",
series = "Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "210--220",
booktitle = "Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017",
address = "United States",

}

Sharifi, A, Ding, W, Guttman, D, Zhao, H, Tang, X, Kandemir, M & Das, C 2017, DEMM: A Dynamic Energy-Saving Mechanism for Multicore Memories. in Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017., 8107447, Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017, Institute of Electrical and Electronics Engineers Inc., pp. 210-220, 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017, Banff, Canada, 9/20/17. https://doi.org/10.1109/MASCOTS.2017.16

DEMM : A Dynamic Energy-Saving Mechanism for Multicore Memories. / Sharifi, Akbar; Ding, Wei; Guttman, Diana; Zhao, Hui; Tang, Xulong; Kandemir, Mahmut; Das, Chitaranjan.

Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017. Institute of Electrical and Electronics Engineers Inc., 2017. p. 210-220 8107447 (Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - DEMM

T2 - A Dynamic Energy-Saving Mechanism for Multicore Memories

AU - Sharifi, Akbar

AU - Ding, Wei

AU - Guttman, Diana

AU - Zhao, Hui

AU - Tang, Xulong

AU - Kandemir, Mahmut

AU - Das, Chitaranjan

PY - 2017/11/13

Y1 - 2017/11/13

N2 - Since main memory system contributes to a large and increasing fraction of server/datacenter energy consumption, there have been several efforts to reduce its power and energy consumption. DVFS schemes have been used to reduce the memory power, but they come with a performance penalty. In this work, we propose DEMM, an OS-based, high performance DVFS mechanism that reduces memory power by dynamically scaling individual memory channel frequencies/voltages. Our strategy also involves clustering the running applications based on their sensitivities to memory latency, and assigning memory channels to the application clusters. We introduce a new metric called Discrete Misses per Kilo Cycle (DMPKC) to capture the performance sensitivities of the applications to memory frequency modulation. DEMM allows us to save power in the memory system with negligible impact on performance. We demonstrate around 25% savings in the memory system energy and 10% savings in the total system energy, with only a 4% loss in workload performance.

AB - Since main memory system contributes to a large and increasing fraction of server/datacenter energy consumption, there have been several efforts to reduce its power and energy consumption. DVFS schemes have been used to reduce the memory power, but they come with a performance penalty. In this work, we propose DEMM, an OS-based, high performance DVFS mechanism that reduces memory power by dynamically scaling individual memory channel frequencies/voltages. Our strategy also involves clustering the running applications based on their sensitivities to memory latency, and assigning memory channels to the application clusters. We introduce a new metric called Discrete Misses per Kilo Cycle (DMPKC) to capture the performance sensitivities of the applications to memory frequency modulation. DEMM allows us to save power in the memory system with negligible impact on performance. We demonstrate around 25% savings in the memory system energy and 10% savings in the total system energy, with only a 4% loss in workload performance.

UR - http://www.scopus.com/inward/record.url?scp=85034014178&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85034014178&partnerID=8YFLogxK

U2 - 10.1109/MASCOTS.2017.16

DO - 10.1109/MASCOTS.2017.16

M3 - Conference contribution

T3 - Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017

SP - 210

EP - 220

BT - Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Sharifi A, Ding W, Guttman D, Zhao H, Tang X, Kandemir M et al. DEMM: A Dynamic Energy-Saving Mechanism for Multicore Memories. In Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017. Institute of Electrical and Electronics Engineers Inc. 2017. p. 210-220. 8107447. (Proceedings - 25th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, MASCOTS 2017). https://doi.org/10.1109/MASCOTS.2017.16