IAMEM: Interaction-aware memory energy management

Mingsong Bi, Srinivasan Chandrasekharan, Chris Gniady

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

4 Scopus citations

Abstract

Energy efficiency has become one of the most important factors in the development of computer systems. As applications become more data centric and put more pressure on the memory subsystem, managing energy consumption of main memory is becoming critical. Therefore, it is critical to take advantage of all memory idle times by placing memory in low power modes even during the active process execution. However, current solutions only offer energy optimizations on a per-process basis and are unable to take advantage of memory idle times when the process is executing. To allow accurate and fine-grained memory management during the process execution, we propose Interaction-Aware Memory Energy Management (IAMEM). IAMEM relies on accurate correlation of user-initiated tasks with the demand placed on the memory subsystem to accurately predict power state transitions for maximal energy savings while minimizing the impact on performance. Through detailed trace-driven simulation, we show that IAMEM reduces the memory energy consumption by as much as 16% as compared to the state-of-the-art approaches, while maintaining the user-perceivable performance comparable to the system without any energy optimizations.

Original languageEnglish (US)
Title of host publicationProceedings of the 2013 USENIX Annual Technical Conference, USENIX ATC 2013
PublisherUSENIX Association
Pages267-278
Number of pages12
ISBN (Electronic)9781931971010
StatePublished - Jan 1 2019
Event2013 USENIX Annual Technical Conference, USENIX ATC 2013 - San Jose, United States
Duration: Jun 26 2013Jun 28 2013

Publication series

NameProceedings of the 2013 USENIX Annual Technical Conference, USENIX ATC 2013

Conference

Conference2013 USENIX Annual Technical Conference, USENIX ATC 2013
CountryUnited States
CitySan Jose
Period6/26/136/28/13

ASJC Scopus subject areas

  • Computer Science(all)

Fingerprint Dive into the research topics of 'IAMEM: Interaction-aware memory energy management'. Together they form a unique fingerprint.

  • Cite this

    Bi, M., Chandrasekharan, S., & Gniady, C. (2019). IAMEM: Interaction-aware memory energy management. In Proceedings of the 2013 USENIX Annual Technical Conference, USENIX ATC 2013 (pp. 267-278). (Proceedings of the 2013 USENIX Annual Technical Conference, USENIX ATC 2013). USENIX Association.