Change search
ReferencesLink to record
Permanent link

Direct link
An efficient runtime power allocation scheme for many-core systems inspired from auction theory
Show others and affiliations
2015 (English)In: Integration, ISSN 0167-9260, Vol. 50, 147-157 p.Article in journal (Refereed) Published
Abstract [en]

Design of future many-core chips is experiencing a paradigm shift to the so-called power-budgeting design, due to the widening gap between instantaneous power consumption and the allowed maximum power, referred as the power budget. Critical to these many-core chips is the runtime power allocation mechanism which can help optimizing the overall system performance under a limited power budget constraint. In this paper, the power allocation problem (i.e., maximizing the system performance under a power budget) is modeled by a combinatorial auction. The problem can be transformed to a knapsack problem and the optimal solution reaches a Walrasian equilibrium. To solve the problem efficiently in a decentralized way, a Hierarchal MultiAgent based Power allocation (HiMAP) method is proposed with an optimal bound. In HiMAP, tiles bid for the opportunity to become active based on the chip's total power budget. Upon finishing an auction process, certain tiles will be power gated and/or frequency scaled according to the power allocation decision. Experimental results have confirmed that HiMAP can reduce the execution time by as much as 45% compared to four competing methods. The runtime overhead and cost of HiMAP are also small, which makes it scale well with many-core systems.

Place, publisher, year, edition, pages
2015. Vol. 50, 147-157 p.
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
URN: urn:nbn:se:kth:diva-171305DOI: 10.1016/j.vlsi.2014.11.001ISI: 000357054300015ScopusID: 2-s2.0-84930757737OAI: diva2:843182

QC 20150727

Available from: 2015-07-27 Created: 2015-07-27 Last updated: 2015-07-27Bibliographically approved

Open Access in DiVA

No full text

Other links

Publisher's full textScopus

Search in DiVA

By author/editor
Daneshtalab, Masoud
By organisation
Electronics and Embedded Systems
In the same journal
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Altmetric score

Total: 35 hits
ReferencesLink to record
Permanent link

Direct link