Improving tolerance to variations in memristor-based applications using parallel memristors

Jeyavijayan Rajendran, Ramesh Karri, Garrett S. Rose

Research output: Contribution to journalArticle

Abstract

Memristors are being explored for a wide variety of applications such as neuromorphic computing, memory and digital logic. However, they suffer from process variations like any other nanodevice, which in turn impacts their applicability. The effect of process variations, specifically variation in thickness, is highly non-linear on memristors; the effect is greater near the lower memristance region (near Moff) than in the higher memristance region (near Moff). Due to this non-linear effect, many applications do not use the lower memristance values. Consequently, the application's functionality and performance is affected. In this work, we propose a technique called parallel memristors. In this technique, instead of using a single memristor, the application uses several memristors connected in parallel. Each memristor in this parallel structure is programmed to a higher memristance value to tolerate variations. Since many memristors are connected in parallel, the effective memristance value can be near the Moff value, thereby achieving high-speed operation. We evaluate the parallel memristor technique in two different applications - memristor-based threshold logic and memristor-based memory. We also perform various optimizations to tradeoff between variation tolerance, power, delay, and area.

Original languageEnglish (US)
Pages (from-to)733-746
Number of pages14
JournalIEEE Transactions on Computers
Volume64
Issue number3
DOIs
StatePublished - Mar 1 2015

Fingerprint

Memristors
Parallel Applications
Tolerance
Process Variation
Logic
Nonlinear Effects
High Speed
Trade-offs
Threshold logic
Data storage equipment
Optimization
Computing
Evaluate

Keywords

  • Emerging technologies
  • logic design
  • memory design
  • reliability

ASJC Scopus subject areas

  • Hardware and Architecture
  • Software
  • Computational Theory and Mathematics
  • Theoretical Computer Science

Cite this

Improving tolerance to variations in memristor-based applications using parallel memristors. / Rajendran, Jeyavijayan; Karri, Ramesh; Rose, Garrett S.

In: IEEE Transactions on Computers, Vol. 64, No. 3, 01.03.2015, p. 733-746.

Research output: Contribution to journalArticle

@article{dfeb74f4082a472d8ddfb59a9ef2d7b2,
title = "Improving tolerance to variations in memristor-based applications using parallel memristors",
abstract = "Memristors are being explored for a wide variety of applications such as neuromorphic computing, memory and digital logic. However, they suffer from process variations like any other nanodevice, which in turn impacts their applicability. The effect of process variations, specifically variation in thickness, is highly non-linear on memristors; the effect is greater near the lower memristance region (near Moff) than in the higher memristance region (near Moff). Due to this non-linear effect, many applications do not use the lower memristance values. Consequently, the application's functionality and performance is affected. In this work, we propose a technique called parallel memristors. In this technique, instead of using a single memristor, the application uses several memristors connected in parallel. Each memristor in this parallel structure is programmed to a higher memristance value to tolerate variations. Since many memristors are connected in parallel, the effective memristance value can be near the Moff value, thereby achieving high-speed operation. We evaluate the parallel memristor technique in two different applications - memristor-based threshold logic and memristor-based memory. We also perform various optimizations to tradeoff between variation tolerance, power, delay, and area.",
keywords = "Emerging technologies, logic design, memory design, reliability",
author = "Jeyavijayan Rajendran and Ramesh Karri and Rose, {Garrett S.}",
year = "2015",
month = "3",
day = "1",
doi = "10.1109/TC.2014.2308189",
language = "English (US)",
volume = "64",
pages = "733--746",
journal = "IEEE Transactions on Computers",
issn = "0018-9340",
publisher = "IEEE Computer Society",
number = "3",

}

TY - JOUR

T1 - Improving tolerance to variations in memristor-based applications using parallel memristors

AU - Rajendran, Jeyavijayan

AU - Karri, Ramesh

AU - Rose, Garrett S.

PY - 2015/3/1

Y1 - 2015/3/1

N2 - Memristors are being explored for a wide variety of applications such as neuromorphic computing, memory and digital logic. However, they suffer from process variations like any other nanodevice, which in turn impacts their applicability. The effect of process variations, specifically variation in thickness, is highly non-linear on memristors; the effect is greater near the lower memristance region (near Moff) than in the higher memristance region (near Moff). Due to this non-linear effect, many applications do not use the lower memristance values. Consequently, the application's functionality and performance is affected. In this work, we propose a technique called parallel memristors. In this technique, instead of using a single memristor, the application uses several memristors connected in parallel. Each memristor in this parallel structure is programmed to a higher memristance value to tolerate variations. Since many memristors are connected in parallel, the effective memristance value can be near the Moff value, thereby achieving high-speed operation. We evaluate the parallel memristor technique in two different applications - memristor-based threshold logic and memristor-based memory. We also perform various optimizations to tradeoff between variation tolerance, power, delay, and area.

AB - Memristors are being explored for a wide variety of applications such as neuromorphic computing, memory and digital logic. However, they suffer from process variations like any other nanodevice, which in turn impacts their applicability. The effect of process variations, specifically variation in thickness, is highly non-linear on memristors; the effect is greater near the lower memristance region (near Moff) than in the higher memristance region (near Moff). Due to this non-linear effect, many applications do not use the lower memristance values. Consequently, the application's functionality and performance is affected. In this work, we propose a technique called parallel memristors. In this technique, instead of using a single memristor, the application uses several memristors connected in parallel. Each memristor in this parallel structure is programmed to a higher memristance value to tolerate variations. Since many memristors are connected in parallel, the effective memristance value can be near the Moff value, thereby achieving high-speed operation. We evaluate the parallel memristor technique in two different applications - memristor-based threshold logic and memristor-based memory. We also perform various optimizations to tradeoff between variation tolerance, power, delay, and area.

KW - Emerging technologies

KW - logic design

KW - memory design

KW - reliability

UR - http://www.scopus.com/inward/record.url?scp=84923241297&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84923241297&partnerID=8YFLogxK

U2 - 10.1109/TC.2014.2308189

DO - 10.1109/TC.2014.2308189

M3 - Article

VL - 64

SP - 733

EP - 746

JO - IEEE Transactions on Computers

JF - IEEE Transactions on Computers

SN - 0018-9340

IS - 3

ER -