PEPON: Performance-aware hierarchical power budgeting for NoC based multicores

Akbar Sharifi, Asit K. Mishra, Shekhar Srikantaiah, Mahmut Kandemir, Chitaranjan Das

Research output: Contribution to journalArticle

28 Citations (Scopus)

Abstract

Targeting NoC based multicores, we propose a two-level power budget distribution mechanism, called PEPON, where the first level distributes the overall power budget of the multicore system among various types of on-chip resources like the cores, caches, and NoC, and the second level determines the allocation of power to individual instances of each type of resource. Both these distributions are oriented towards maximizing workload performance without exceeding the specified power budget. Extensive experimental evaluations of the proposed power distribution scheme using a full system simulation and detailed power models emphasize the importance of power budget partitioning at both levels. Specifically, our results show that the proposed scheme can provide up to 29% performance improvement as compared to no power budgeting, and performs 13% better than a competing scheme, under the same chip-wide power cap.

Original languageEnglish (US)
Pages (from-to)65-74
Number of pages10
JournalParallel Architectures and Compilation Techniques - Conference Proceedings, PACT
DOIs
StatePublished - 2012

Fingerprint

Budget control
Chip
Resources
Power Distribution
System Simulation
Experimental Evaluation
Cache
Workload
Partitioning
Network on chip
Network-on-chip
Model

All Science Journal Classification (ASJC) codes

  • Software
  • Hardware and Architecture
  • Theoretical Computer Science

Cite this

@article{e0e35794a6724a049dce9ad7a1ecd1f4,
title = "PEPON: Performance-aware hierarchical power budgeting for NoC based multicores",
abstract = "Targeting NoC based multicores, we propose a two-level power budget distribution mechanism, called PEPON, where the first level distributes the overall power budget of the multicore system among various types of on-chip resources like the cores, caches, and NoC, and the second level determines the allocation of power to individual instances of each type of resource. Both these distributions are oriented towards maximizing workload performance without exceeding the specified power budget. Extensive experimental evaluations of the proposed power distribution scheme using a full system simulation and detailed power models emphasize the importance of power budget partitioning at both levels. Specifically, our results show that the proposed scheme can provide up to 29{\%} performance improvement as compared to no power budgeting, and performs 13{\%} better than a competing scheme, under the same chip-wide power cap.",
author = "Akbar Sharifi and Mishra, {Asit K.} and Shekhar Srikantaiah and Mahmut Kandemir and Chitaranjan Das",
year = "2012",
doi = "10.1145/2370816.2370828",
language = "English (US)",
pages = "65--74",
journal = "Parallel Architectures and Compilation Techniques - Conference Proceedings, PACT",
issn = "1089-795X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

PEPON : Performance-aware hierarchical power budgeting for NoC based multicores. / Sharifi, Akbar; Mishra, Asit K.; Srikantaiah, Shekhar; Kandemir, Mahmut; Das, Chitaranjan.

In: Parallel Architectures and Compilation Techniques - Conference Proceedings, PACT, 2012, p. 65-74.

Research output: Contribution to journalArticle

TY - JOUR

T1 - PEPON

T2 - Performance-aware hierarchical power budgeting for NoC based multicores

AU - Sharifi, Akbar

AU - Mishra, Asit K.

AU - Srikantaiah, Shekhar

AU - Kandemir, Mahmut

AU - Das, Chitaranjan

PY - 2012

Y1 - 2012

N2 - Targeting NoC based multicores, we propose a two-level power budget distribution mechanism, called PEPON, where the first level distributes the overall power budget of the multicore system among various types of on-chip resources like the cores, caches, and NoC, and the second level determines the allocation of power to individual instances of each type of resource. Both these distributions are oriented towards maximizing workload performance without exceeding the specified power budget. Extensive experimental evaluations of the proposed power distribution scheme using a full system simulation and detailed power models emphasize the importance of power budget partitioning at both levels. Specifically, our results show that the proposed scheme can provide up to 29% performance improvement as compared to no power budgeting, and performs 13% better than a competing scheme, under the same chip-wide power cap.

AB - Targeting NoC based multicores, we propose a two-level power budget distribution mechanism, called PEPON, where the first level distributes the overall power budget of the multicore system among various types of on-chip resources like the cores, caches, and NoC, and the second level determines the allocation of power to individual instances of each type of resource. Both these distributions are oriented towards maximizing workload performance without exceeding the specified power budget. Extensive experimental evaluations of the proposed power distribution scheme using a full system simulation and detailed power models emphasize the importance of power budget partitioning at both levels. Specifically, our results show that the proposed scheme can provide up to 29% performance improvement as compared to no power budgeting, and performs 13% better than a competing scheme, under the same chip-wide power cap.

UR - http://www.scopus.com/inward/record.url?scp=84867523198&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867523198&partnerID=8YFLogxK

U2 - 10.1145/2370816.2370828

DO - 10.1145/2370816.2370828

M3 - Article

AN - SCOPUS:84867523198

SP - 65

EP - 74

JO - Parallel Architectures and Compilation Techniques - Conference Proceedings, PACT

JF - Parallel Architectures and Compilation Techniques - Conference Proceedings, PACT

SN - 1089-795X

ER -