Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Memory Power Management for Java Processors Using Heap Partitioning and Power Gating

Gomez, Ricardo ; Gruian, Flavius LU orcid and Liu, Liang LU orcid (2016) 14th International Workshop on Java Technologies for Real-Time and Embedded Systems - JTRES 2016
Abstract
Power consumption is an important design parameter for battery operated devices, and an especially sensitive issue in embedded Java systems. This paper presents a solution to reduce leakage power consumption of the heap memory characteristic of Java processors. By partitioning the heap into several memory banks and taking advantage of a compacting garbage collector, our method dynamically powers off inactive regions.
The technique has been designed and evaluated for JOP, a Java Optimized Processor, and further implemented and verified in a 65nm CMOS technology using STM low-power high Vt (LPHVT) standard cell libraries. Experiments show that our method accurately follows the memory utilization profile in powering on and off banks,... (More)
Power consumption is an important design parameter for battery operated devices, and an especially sensitive issue in embedded Java systems. This paper presents a solution to reduce leakage power consumption of the heap memory characteristic of Java processors. By partitioning the heap into several memory banks and taking advantage of a compacting garbage collector, our method dynamically powers off inactive regions.
The technique has been designed and evaluated for JOP, a Java Optimized Processor, and further implemented and verified in a 65nm CMOS technology using STM low-power high Vt (LPHVT) standard cell libraries. Experiments show that our method accurately follows the memory utilization profile in powering on and off banks, achieving at least 50% leakage power reduction. The performance, area and power penalty introduced by the additional hardware are negligible. (Less)
Please use this url to cite or link to this publication:
author
; and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
host publication
The 14th International Workshop on Java Technologies for Real-Time and Embedded Systems
pages
8 pages
conference name
14th International Workshop on Java Technologies for Real-Time and Embedded Systems - JTRES 2016
conference location
Lugano, Switzerland
conference dates
2016-08-29 - 2016-09-02
external identifiers
  • scopus:85133481874
DOI
10.1145/2990509.2990514
language
English
LU publication?
yes
id
3440fbb9-bd4a-4840-b7a7-4a0b1912dd2a
date added to LUP
2016-08-22 13:43:47
date last changed
2024-01-03 14:17:30
@inproceedings{3440fbb9-bd4a-4840-b7a7-4a0b1912dd2a,
  abstract     = {{Power consumption is an important design parameter for battery operated devices, and an especially sensitive issue in embedded Java systems. This paper presents a solution to reduce leakage power consumption of the heap memory characteristic of Java processors. By partitioning the heap into several memory banks and taking advantage of a compacting garbage collector, our method dynamically powers off inactive regions. <br/>The technique has been designed and evaluated for JOP, a Java Optimized Processor, and further implemented and verified in a 65nm CMOS technology using STM low-power high Vt (LPHVT) standard cell libraries. Experiments show that our method accurately follows the memory utilization profile in powering on and off banks, achieving at least 50% leakage power reduction. The performance, area and power penalty introduced by the additional hardware are negligible.}},
  author       = {{Gomez, Ricardo and Gruian, Flavius and Liu, Liang}},
  booktitle    = {{The 14th International Workshop on Java Technologies for Real-Time and Embedded Systems}},
  language     = {{eng}},
  title        = {{Memory Power Management for Java Processors Using Heap Partitioning and Power Gating}},
  url          = {{http://dx.doi.org/10.1145/2990509.2990514}},
  doi          = {{10.1145/2990509.2990514}},
  year         = {{2016}},
}