The University of Southampton
University of Southampton Institutional Repository

Experimental demonstration of RRAM-based computational cells for reconfigurable mixed-signal neuro-inspired circuits and systems

Experimental demonstration of RRAM-based computational cells for reconfigurable mixed-signal neuro-inspired circuits and systems
Experimental demonstration of RRAM-based computational cells for reconfigurable mixed-signal neuro-inspired circuits and systems
Modern electronics drive a shift from distributed, cloud and/or mainframe computing towards the ‘edge’. To realise this vision, we need access to hardware technologies that are both energy and scale efficient. During the last decade, the introduction of Resistive Random Access Memory (RRAM), also known as memristors, has fuelled interest in extending conventional circuits’ capabilities. Specifically, their capacity to act as scalable, non-volatile, finely tuneable, electrically programmable resistive elements render them promising candidates for future computer architectures. RRAM technologies have been considered by many as a promising candidate for implementing reconfigurable neuro-inspired circuits and systems capable of processing data in both digital and analogue formats. Presently, there is no extensive study of the behaviour of such circuits when realised physically with real RRAM devices. Hence, there are ample opportunities for developing novel electronic circuits for reconfigurable mixed-signal data processors in silico.
This thesis explores the design, implementation and testing of in-silico data processors capable of mapping data from one information domain to another and enabling a mixed-signal data processing. Through this research, I am introducing RRAM-based circuit designs operationally validated through simulations with state-of-art RRAM device model and then practically implemented proof-of-concept designs of these hybrid RRAM-CMOS circuits on hardware. The hardware implementation and testing of low-complexity primitive RRAM-based circuits that can process information in the analogue domain due to the introduction of programmable RRAM devices is the main contributions through this project. In this work, findings are presented regarding the implementation and testing in hardware of a RRAM-based primitive Multiply-Accumulate circuit, RRAM-enhanced Threshold Logic Gate design and as well as larger circuits on these primitive circuits that are easily integrated into RRAM-based In-Memory Computing (IMC) architectures. A primitive RRAM-based MAC circuit is designed and its behaviour exhibited in both simulation (Cadence Virtuoso Spectre) and in hardware. This circuit is used for a proof-of-concept Winner-Take-All systems that showcases 300fJ energy and 1.4ns delay per operation. Additionally, A RRAM-based Threshold Logic Gate is designed and showcased achieving 27.2µW and 0.14ns per operation. Finally, a RRAM-based Wake-Up Circuit is showcased to be feasible using the the aforementioned circuits inside an IMC system.
University of Southampton
Papandroulidakis, Georgios
518ddb08-ebeb-4026-829d-7a3db4fd3275
Papandroulidakis, Georgios
518ddb08-ebeb-4026-829d-7a3db4fd3275
Serb, Alexantrou
30f5ec26-f51d-42b3-85fd-0325a27a792c
Prodromakis, Themis
d58c9c10-9d25-4d22-b155-06c8437acfbf
Merrett, Geoffrey
89b3a696-41de-44c3-89aa-b0aa29f54020

Papandroulidakis, Georgios (2021) Experimental demonstration of RRAM-based computational cells for reconfigurable mixed-signal neuro-inspired circuits and systems. University of Southampton, Doctoral Thesis.

Record type: Thesis (Doctoral)

Abstract

Modern electronics drive a shift from distributed, cloud and/or mainframe computing towards the ‘edge’. To realise this vision, we need access to hardware technologies that are both energy and scale efficient. During the last decade, the introduction of Resistive Random Access Memory (RRAM), also known as memristors, has fuelled interest in extending conventional circuits’ capabilities. Specifically, their capacity to act as scalable, non-volatile, finely tuneable, electrically programmable resistive elements render them promising candidates for future computer architectures. RRAM technologies have been considered by many as a promising candidate for implementing reconfigurable neuro-inspired circuits and systems capable of processing data in both digital and analogue formats. Presently, there is no extensive study of the behaviour of such circuits when realised physically with real RRAM devices. Hence, there are ample opportunities for developing novel electronic circuits for reconfigurable mixed-signal data processors in silico.
This thesis explores the design, implementation and testing of in-silico data processors capable of mapping data from one information domain to another and enabling a mixed-signal data processing. Through this research, I am introducing RRAM-based circuit designs operationally validated through simulations with state-of-art RRAM device model and then practically implemented proof-of-concept designs of these hybrid RRAM-CMOS circuits on hardware. The hardware implementation and testing of low-complexity primitive RRAM-based circuits that can process information in the analogue domain due to the introduction of programmable RRAM devices is the main contributions through this project. In this work, findings are presented regarding the implementation and testing in hardware of a RRAM-based primitive Multiply-Accumulate circuit, RRAM-enhanced Threshold Logic Gate design and as well as larger circuits on these primitive circuits that are easily integrated into RRAM-based In-Memory Computing (IMC) architectures. A primitive RRAM-based MAC circuit is designed and its behaviour exhibited in both simulation (Cadence Virtuoso Spectre) and in hardware. This circuit is used for a proof-of-concept Winner-Take-All systems that showcases 300fJ energy and 1.4ns delay per operation. Additionally, A RRAM-based Threshold Logic Gate is designed and showcased achieving 27.2µW and 0.14ns per operation. Finally, a RRAM-based Wake-Up Circuit is showcased to be feasible using the the aforementioned circuits inside an IMC system.

Text
Thesis - Version of Record
Available under License University of Southampton Thesis Licence.
Download (5MB)
Text
PTD_Thesis_Papandroulidakis-SIGNED
Restricted to Repository staff only

More information

Published date: 2021

Identifiers

Local EPrints ID: 456862
URI: http://eprints.soton.ac.uk/id/eprint/456862
PURE UUID: 1dd93026-c3c7-4161-b4b2-11a85079dfd2
ORCID for Georgios Papandroulidakis: ORCID iD orcid.org/0000-0002-9203-2557
ORCID for Themis Prodromakis: ORCID iD orcid.org/0000-0002-6267-6909
ORCID for Geoffrey Merrett: ORCID iD orcid.org/0000-0003-4980-3894

Catalogue record

Date deposited: 13 May 2022 16:37
Last modified: 14 May 2022 01:44

Export record

Contributors

Author: Georgios Papandroulidakis ORCID iD
Thesis advisor: Alexantrou Serb
Thesis advisor: Themis Prodromakis ORCID iD
Thesis advisor: Geoffrey Merrett ORCID iD

Download statistics

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics

Atom RSS 1.0 RSS 2.0

Contact ePrints Soton: eprints@soton.ac.uk

ePrints Soton supports OAI 2.0 with a base URL of http://eprints.soton.ac.uk/cgi/oai2

This repository has been built using EPrints software, developed at the University of Southampton, but available to everyone to use.

We use cookies to ensure that we give you the best experience on our website. If you continue without changing your settings, we will assume that you are happy to receive cookies on the University of Southampton website.

×