I/O Aware Power Shifting

Lee Savoie, David K Lowenthal, Bronis R De Supinski, Tanzima Islam, Kathryn Mohror, Barry Rountree, Martin Schulz

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

5 Citations (Scopus)

Abstract

Power limits on future high-performance computing (HPC) systems will constrain applications. However, HPC applications do not consume constant power over their lifetimes. Thus, applications assigned a fixed power bound may be forced to slow down during high-power computation phases, but may not consume their full power allocation during low-power I/O phases. This paper explores algorithms that leverage application semantics - phase frequency, duration and power needs - to shift unused power from applications in I/O phases to applications in computation phases, thus improving system-wide performance. We design novel techniques that include explicit staggering of applications to improve power shifting. Compared to executing without power shifting, our algorithms can improve average performance by up to 8% or improve performance of a single, high-priority application by up to 32%.

Original languageEnglish (US)
Title of host publicationProceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages740-749
Number of pages10
ISBN (Electronic)9781509021406
DOIs
StatePublished - Jul 18 2016
Event30th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2016 - Chicago, United States
Duration: May 23 2016May 27 2016

Other

Other30th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2016
CountryUnited States
CityChicago
Period5/23/165/27/16

Fingerprint

Semantics

Keywords

  • HPC
  • Performance
  • Power

ASJC Scopus subject areas

  • Computer Networks and Communications

Cite this

Savoie, L., Lowenthal, D. K., Supinski, B. R. D., Islam, T., Mohror, K., Rountree, B., & Schulz, M. (2016). I/O Aware Power Shifting. In Proceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016 (pp. 740-749). [7516070] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IPDPS.2016.15

I/O Aware Power Shifting. / Savoie, Lee; Lowenthal, David K; Supinski, Bronis R De; Islam, Tanzima; Mohror, Kathryn; Rountree, Barry; Schulz, Martin.

Proceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016. Institute of Electrical and Electronics Engineers Inc., 2016. p. 740-749 7516070.

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

Savoie, L, Lowenthal, DK, Supinski, BRD, Islam, T, Mohror, K, Rountree, B & Schulz, M 2016, I/O Aware Power Shifting. in Proceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016., 7516070, Institute of Electrical and Electronics Engineers Inc., pp. 740-749, 30th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2016, Chicago, United States, 5/23/16. https://doi.org/10.1109/IPDPS.2016.15
Savoie L, Lowenthal DK, Supinski BRD, Islam T, Mohror K, Rountree B et al. I/O Aware Power Shifting. In Proceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016. Institute of Electrical and Electronics Engineers Inc. 2016. p. 740-749. 7516070 https://doi.org/10.1109/IPDPS.2016.15
Savoie, Lee ; Lowenthal, David K ; Supinski, Bronis R De ; Islam, Tanzima ; Mohror, Kathryn ; Rountree, Barry ; Schulz, Martin. / I/O Aware Power Shifting. Proceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016. Institute of Electrical and Electronics Engineers Inc., 2016. pp. 740-749
@inproceedings{71e19c3f257d4a0faf356518be4ad3a5,
title = "I/O Aware Power Shifting",
abstract = "Power limits on future high-performance computing (HPC) systems will constrain applications. However, HPC applications do not consume constant power over their lifetimes. Thus, applications assigned a fixed power bound may be forced to slow down during high-power computation phases, but may not consume their full power allocation during low-power I/O phases. This paper explores algorithms that leverage application semantics - phase frequency, duration and power needs - to shift unused power from applications in I/O phases to applications in computation phases, thus improving system-wide performance. We design novel techniques that include explicit staggering of applications to improve power shifting. Compared to executing without power shifting, our algorithms can improve average performance by up to 8{\%} or improve performance of a single, high-priority application by up to 32{\%}.",
keywords = "HPC, Performance, Power",
author = "Lee Savoie and Lowenthal, {David K} and Supinski, {Bronis R De} and Tanzima Islam and Kathryn Mohror and Barry Rountree and Martin Schulz",
year = "2016",
month = "7",
day = "18",
doi = "10.1109/IPDPS.2016.15",
language = "English (US)",
pages = "740--749",
booktitle = "Proceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

TY - GEN

T1 - I/O Aware Power Shifting

AU - Savoie, Lee

AU - Lowenthal, David K

AU - Supinski, Bronis R De

AU - Islam, Tanzima

AU - Mohror, Kathryn

AU - Rountree, Barry

AU - Schulz, Martin

PY - 2016/7/18

Y1 - 2016/7/18

N2 - Power limits on future high-performance computing (HPC) systems will constrain applications. However, HPC applications do not consume constant power over their lifetimes. Thus, applications assigned a fixed power bound may be forced to slow down during high-power computation phases, but may not consume their full power allocation during low-power I/O phases. This paper explores algorithms that leverage application semantics - phase frequency, duration and power needs - to shift unused power from applications in I/O phases to applications in computation phases, thus improving system-wide performance. We design novel techniques that include explicit staggering of applications to improve power shifting. Compared to executing without power shifting, our algorithms can improve average performance by up to 8% or improve performance of a single, high-priority application by up to 32%.

AB - Power limits on future high-performance computing (HPC) systems will constrain applications. However, HPC applications do not consume constant power over their lifetimes. Thus, applications assigned a fixed power bound may be forced to slow down during high-power computation phases, but may not consume their full power allocation during low-power I/O phases. This paper explores algorithms that leverage application semantics - phase frequency, duration and power needs - to shift unused power from applications in I/O phases to applications in computation phases, thus improving system-wide performance. We design novel techniques that include explicit staggering of applications to improve power shifting. Compared to executing without power shifting, our algorithms can improve average performance by up to 8% or improve performance of a single, high-priority application by up to 32%.

KW - HPC

KW - Performance

KW - Power

UR - http://www.scopus.com/inward/record.url?scp=84983283657&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84983283657&partnerID=8YFLogxK

U2 - 10.1109/IPDPS.2016.15

DO - 10.1109/IPDPS.2016.15

M3 - Conference contribution

SP - 740

EP - 749

BT - Proceedings - 2016 IEEE 30th International Parallel and Distributed Processing Symposium, IPDPS 2016

PB - Institute of Electrical and Electronics Engineers Inc.

ER -