Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures

Ishan G. Thakkar; Sudeep Pasricha

doi:10.1109/VLSID.2016.13

Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures

Ishan G. Thakkar, Sudeep Pasricha

Electrical and Computer Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

9 Scopus citations

Abstract

This paper presents a novel, energy-efficient DRAM refresh technique called massed refresh that simultaneously leverages bank-level and sub array-level concurrency to reduce the overhead of distributed refresh operations in the Hybrid Memory Cube (HMC). In massed refresh, a bundle of DRAM rows in a refresh operation is composed of two subgroups mapped to two different banks, with the rows of each subgroup mapped to different sub arrays within the corresponding bank. Both subgroups of DRAM rows are refreshed concurrently during a refresh command, which greatly reduces the refresh cycle time and improves bandwidth and energy efficiency of the HMC. Our experimental analysis shows that the proposed massed refresh technique achieves up to 6.3% and 5.8% improvements in throughput and energy-delay product on average over JEDEC standardized distributed per-bank refresh and state-of-the-art scattered refresh techniques.

Original language	English
Title of host publication	Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems
Pages	104-109
Number of pages	6
ISBN (Electronic)	9781467387002
DOIs	https://doi.org/10.1109/VLSID.2016.13
State	Published - Mar 16 2016
Event	29th International Conference on VLSI Design, VLSID 2016 - Kolkata, India Duration: Jan 4 2016 → Jan 8 2016

Publication series

Name	Proceedings of the IEEE International Conference on VLSI Design
Volume	2016-March
ISSN (Print)	1063-9667

Conference

Conference	29th International Conference on VLSI Design, VLSID 2016
Country/Territory	India
City	Kolkata
Period	1/4/16 → 1/8/16

Bibliographical note

Funding Information:
This research is supported by grants from SRC, NSF (CCF-1252500, CCF-1302693), and AFOSR (FA9550-13-1-0110)

Publisher Copyright:
© 2016 IEEE.

Keywords

Bank-level parallelism
DRAM
Distributed refresh
Energy-ef-ficiency
Hybrid Memory Cube

ASJC Scopus subject areas

Hardware and Architecture
Electrical and Electronic Engineering

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/VLSID.2016.13

Cite this

Thakkar, I. G., & Pasricha, S. (2016). Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures. In Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems (pp. 104-109). Article 7434935 (Proceedings of the IEEE International Conference on VLSI Design; Vol. 2016-March). https://doi.org/10.1109/VLSID.2016.13

Thakkar, Ishan G. ; Pasricha, Sudeep. / Massed Refresh : An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures. Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems. 2016. pp. 104-109 (Proceedings of the IEEE International Conference on VLSI Design).

@inproceedings{736df5650335493eb3de8017fe57400f,

title = "Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures",

abstract = "This paper presents a novel, energy-efficient DRAM refresh technique called massed refresh that simultaneously leverages bank-level and sub array-level concurrency to reduce the overhead of distributed refresh operations in the Hybrid Memory Cube (HMC). In massed refresh, a bundle of DRAM rows in a refresh operation is composed of two subgroups mapped to two different banks, with the rows of each subgroup mapped to different sub arrays within the corresponding bank. Both subgroups of DRAM rows are refreshed concurrently during a refresh command, which greatly reduces the refresh cycle time and improves bandwidth and energy efficiency of the HMC. Our experimental analysis shows that the proposed massed refresh technique achieves up to 6.3% and 5.8% improvements in throughput and energy-delay product on average over JEDEC standardized distributed per-bank refresh and state-of-the-art scattered refresh techniques.",

keywords = "Bank-level parallelism, DRAM, Distributed refresh, Energy-ef-ficiency, Hybrid Memory Cube",

author = "Thakkar, {Ishan G.} and Sudeep Pasricha",

note = "Funding Information: This research is supported by grants from SRC, NSF (CCF-1252500, CCF-1302693), and AFOSR (FA9550-13-1-0110) Publisher Copyright: {\textcopyright} 2016 IEEE.; 29th International Conference on VLSI Design, VLSID 2016 ; Conference date: 04-01-2016 Through 08-01-2016",

year = "2016",

month = mar,

day = "16",

doi = "10.1109/VLSID.2016.13",

language = "English",

series = "Proceedings of the IEEE International Conference on VLSI Design",

pages = "104--109",

booktitle = "Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems",

}

Thakkar, IG & Pasricha, S 2016, Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures. in Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems., 7434935, Proceedings of the IEEE International Conference on VLSI Design, vol. 2016-March, pp. 104-109, 29th International Conference on VLSI Design, VLSID 2016, Kolkata, India, 1/4/16. https://doi.org/10.1109/VLSID.2016.13

Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures. / Thakkar, Ishan G.; Pasricha, Sudeep.
Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems. 2016. p. 104-109 7434935 (Proceedings of the IEEE International Conference on VLSI Design; Vol. 2016-March).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Massed Refresh

T2 - 29th International Conference on VLSI Design, VLSID 2016

AU - Thakkar, Ishan G.

AU - Pasricha, Sudeep

PY - 2016/3/16

Y1 - 2016/3/16

N2 - This paper presents a novel, energy-efficient DRAM refresh technique called massed refresh that simultaneously leverages bank-level and sub array-level concurrency to reduce the overhead of distributed refresh operations in the Hybrid Memory Cube (HMC). In massed refresh, a bundle of DRAM rows in a refresh operation is composed of two subgroups mapped to two different banks, with the rows of each subgroup mapped to different sub arrays within the corresponding bank. Both subgroups of DRAM rows are refreshed concurrently during a refresh command, which greatly reduces the refresh cycle time and improves bandwidth and energy efficiency of the HMC. Our experimental analysis shows that the proposed massed refresh technique achieves up to 6.3% and 5.8% improvements in throughput and energy-delay product on average over JEDEC standardized distributed per-bank refresh and state-of-the-art scattered refresh techniques.

AB - This paper presents a novel, energy-efficient DRAM refresh technique called massed refresh that simultaneously leverages bank-level and sub array-level concurrency to reduce the overhead of distributed refresh operations in the Hybrid Memory Cube (HMC). In massed refresh, a bundle of DRAM rows in a refresh operation is composed of two subgroups mapped to two different banks, with the rows of each subgroup mapped to different sub arrays within the corresponding bank. Both subgroups of DRAM rows are refreshed concurrently during a refresh command, which greatly reduces the refresh cycle time and improves bandwidth and energy efficiency of the HMC. Our experimental analysis shows that the proposed massed refresh technique achieves up to 6.3% and 5.8% improvements in throughput and energy-delay product on average over JEDEC standardized distributed per-bank refresh and state-of-the-art scattered refresh techniques.

KW - Bank-level parallelism

KW - DRAM

KW - Distributed refresh

KW - Energy-ef-ficiency

KW - Hybrid Memory Cube

UR - http://www.scopus.com/inward/record.url?scp=84964652359&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84964652359&partnerID=8YFLogxK

U2 - 10.1109/VLSID.2016.13

DO - 10.1109/VLSID.2016.13

M3 - Conference contribution

AN - SCOPUS:84964652359

T3 - Proceedings of the IEEE International Conference on VLSI Design

SP - 104

EP - 109

BT - Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems

Y2 - 4 January 2016 through 8 January 2016

ER -

Thakkar IG, Pasricha S. Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures. In Proceedings - 29th International Conference on VLSI Design, VLSID 2016 - Held concurrently with 15th International Conference on Embedded Systems. 2016. p. 104-109. 7434935. (Proceedings of the IEEE International Conference on VLSI Design). doi: 10.1109/VLSID.2016.13

Massed Refresh: An Energy-Efficient Technique to Reduce Refresh Overhead in Hybrid Memory Cube Architectures

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this