Pre-Charged Local Bit-Line Sharing SRAM Architecture for Near-Threshold Operation

Tae Woo Oh, Hanwool Jeong, Juhyun Park, Seong Ook Jung

Research output: Contribution to journalArticle

Abstract

In this paper, a pre-charged local bit-line sharing (PCLBS) static random access memory (SRAM) for near-threshold operation is proposed. In previous local bit-line sharing SRAMs, such as average-8T and full-swing local bit-line (FSLB) SRAMs, multiple bit-cells share a local bit-line pair with a small capacitance for high read stability. However, the average-8T SRAM has a considerably large delay because the full development of the local bit-line cannot be achieved. On the other hand, the FSLB SRAM reduces the delay but requires a timing constraint of control signals to achieve sufficient read sensing margin. The proposed PCLBS SRAM achieves high read speed by fully developing local bit-line pair without a timing constraint. Furthermore, the proposed PCLBS SRAM enhances the read stability and the write ability by, respectively, applying a pre-charged local bit-line scheme and transmission gates in write paths. Based on a 22-nm FinFET technology, the FSLB and proposed PCLBS SRAM have the minimum operating voltages of 0.44 and 0.4 V, respectively, while achieving the σ target read stability and write ability yields. Compared with the FSLB SRAM, the proposed PCLBS SRAM consumes 21% less energy at each minimum operating voltage and has 57% smaller read delay at the operating voltage of 0.4 V.

Original languageEnglish
Article number7932141
Pages (from-to)2737-2747
Number of pages11
JournalIEEE Transactions on Circuits and Systems I: Regular Papers
Volume64
Issue number10
DOIs
Publication statusPublished - 2017 Oct 1

Fingerprint

Memory architecture
Data storage equipment
Static random access storage
Electric potential

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Cite this

@article{568fe8f37b784d4297ddb8713c9ce795,
title = "Pre-Charged Local Bit-Line Sharing SRAM Architecture for Near-Threshold Operation",
abstract = "In this paper, a pre-charged local bit-line sharing (PCLBS) static random access memory (SRAM) for near-threshold operation is proposed. In previous local bit-line sharing SRAMs, such as average-8T and full-swing local bit-line (FSLB) SRAMs, multiple bit-cells share a local bit-line pair with a small capacitance for high read stability. However, the average-8T SRAM has a considerably large delay because the full development of the local bit-line cannot be achieved. On the other hand, the FSLB SRAM reduces the delay but requires a timing constraint of control signals to achieve sufficient read sensing margin. The proposed PCLBS SRAM achieves high read speed by fully developing local bit-line pair without a timing constraint. Furthermore, the proposed PCLBS SRAM enhances the read stability and the write ability by, respectively, applying a pre-charged local bit-line scheme and transmission gates in write paths. Based on a 22-nm FinFET technology, the FSLB and proposed PCLBS SRAM have the minimum operating voltages of 0.44 and 0.4 V, respectively, while achieving the σ target read stability and write ability yields. Compared with the FSLB SRAM, the proposed PCLBS SRAM consumes 21{\%} less energy at each minimum operating voltage and has 57{\%} smaller read delay at the operating voltage of 0.4 V.",
author = "Oh, {Tae Woo} and Hanwool Jeong and Juhyun Park and Jung, {Seong Ook}",
year = "2017",
month = "10",
day = "1",
doi = "10.1109/TCSI.2017.2702587",
language = "English",
volume = "64",
pages = "2737--2747",
journal = "IEEE Transactions on Circuits and Systems II: Express Briefs",
issn = "1549-8328",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "10",

}

Pre-Charged Local Bit-Line Sharing SRAM Architecture for Near-Threshold Operation. / Oh, Tae Woo; Jeong, Hanwool; Park, Juhyun; Jung, Seong Ook.

In: IEEE Transactions on Circuits and Systems I: Regular Papers, Vol. 64, No. 10, 7932141, 01.10.2017, p. 2737-2747.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Pre-Charged Local Bit-Line Sharing SRAM Architecture for Near-Threshold Operation

AU - Oh, Tae Woo

AU - Jeong, Hanwool

AU - Park, Juhyun

AU - Jung, Seong Ook

PY - 2017/10/1

Y1 - 2017/10/1

N2 - In this paper, a pre-charged local bit-line sharing (PCLBS) static random access memory (SRAM) for near-threshold operation is proposed. In previous local bit-line sharing SRAMs, such as average-8T and full-swing local bit-line (FSLB) SRAMs, multiple bit-cells share a local bit-line pair with a small capacitance for high read stability. However, the average-8T SRAM has a considerably large delay because the full development of the local bit-line cannot be achieved. On the other hand, the FSLB SRAM reduces the delay but requires a timing constraint of control signals to achieve sufficient read sensing margin. The proposed PCLBS SRAM achieves high read speed by fully developing local bit-line pair without a timing constraint. Furthermore, the proposed PCLBS SRAM enhances the read stability and the write ability by, respectively, applying a pre-charged local bit-line scheme and transmission gates in write paths. Based on a 22-nm FinFET technology, the FSLB and proposed PCLBS SRAM have the minimum operating voltages of 0.44 and 0.4 V, respectively, while achieving the σ target read stability and write ability yields. Compared with the FSLB SRAM, the proposed PCLBS SRAM consumes 21% less energy at each minimum operating voltage and has 57% smaller read delay at the operating voltage of 0.4 V.

AB - In this paper, a pre-charged local bit-line sharing (PCLBS) static random access memory (SRAM) for near-threshold operation is proposed. In previous local bit-line sharing SRAMs, such as average-8T and full-swing local bit-line (FSLB) SRAMs, multiple bit-cells share a local bit-line pair with a small capacitance for high read stability. However, the average-8T SRAM has a considerably large delay because the full development of the local bit-line cannot be achieved. On the other hand, the FSLB SRAM reduces the delay but requires a timing constraint of control signals to achieve sufficient read sensing margin. The proposed PCLBS SRAM achieves high read speed by fully developing local bit-line pair without a timing constraint. Furthermore, the proposed PCLBS SRAM enhances the read stability and the write ability by, respectively, applying a pre-charged local bit-line scheme and transmission gates in write paths. Based on a 22-nm FinFET technology, the FSLB and proposed PCLBS SRAM have the minimum operating voltages of 0.44 and 0.4 V, respectively, while achieving the σ target read stability and write ability yields. Compared with the FSLB SRAM, the proposed PCLBS SRAM consumes 21% less energy at each minimum operating voltage and has 57% smaller read delay at the operating voltage of 0.4 V.

UR - http://www.scopus.com/inward/record.url?scp=85019881434&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85019881434&partnerID=8YFLogxK

U2 - 10.1109/TCSI.2017.2702587

DO - 10.1109/TCSI.2017.2702587

M3 - Article

AN - SCOPUS:85019881434

VL - 64

SP - 2737

EP - 2747

JO - IEEE Transactions on Circuits and Systems II: Express Briefs

JF - IEEE Transactions on Circuits and Systems II: Express Briefs

SN - 1549-8328

IS - 10

M1 - 7932141

ER -