research-article

PHANES: ReRAM-based photonic accelerator for deep neural networks

Authors:

Jiang XuAuthors Info & Claims

DAC '22: Proceedings of the 59th ACM/IEEE Design Automation Conference

Pages 103 - 108

https://doi.org/10.1145/3489517.3530397

Published: 23 August 2022 Publication History

Abstract

Resistive random access memory (ReRAM) has demonstrated great promises of in-situ matrix-vector multiplications to accelerate deep neural networks. However, subject to the intrinsic properties of analog processing, most of the proposed ReRAM-based accelerators require excessive costly ADC/DAC to avoid distortion of electronic analog signals during inter-tile transmission. Moreover, due to bit-shifting before addition, prior works require longer cycles to serially calculate partial sum compared to multiplications, which dramatically restricts the throughput and is more likely to stall the pipeline between layers of deep neural networks.

In this paper, we present a novel ReRAM-based photonic accelerator (PHANES) architecture, which calculates multiplications in ReRAM and parallel weighted accumulations during optical transmission. Such photonic paradigm also serves as high-fidelity analog-analog links to further reduce ADC/DAC. To circumvent the memory wall problem, we further propose a progressive bit-depth technique. Evaluations show that PHANES improves the energy efficiency by 6.09x and throughput density by 14.7x compared to state-of-the-art designs. Our photonic architecture also has great potentials for scalability towards very-large-scale accelerators.

References

[1]

L. Song et al., "Pipelayer: A pipelined reram-based accelerator for deep learning," in HPCA. IEEE, 2017, pp. 541--552.

[2]

M. Imani et al., "Floatpim: In-memory acceleration of deep neural network training with high precision," in ISCA. IEEE, 2019.

Digital Library

[3]

W. Wan et al., "33.1 a 74 tmacs/w cmos-rram neurosynaptic core with dynamically reconfigurable dataflow and in-situ transposable weights for probabilistic graphical models," in ISSCC. IEEE, 2020, pp. 498--500.

[4]

A. Shafiee et al., "Isaac: A convolutional neural network accelerator with in-situ analog arithmetic in crossbars," in ISCA, 2016, pp. 14--26.

[5]

G. Yuan et al., "Forms: Fine-grained polarized reram-based in-situ computation for mixed-signal dnn accelerator," in ISCA, 2021.

[6]

A. Ankit et al., "Puma: A programmable ultra-efficient memristor-based accelerator for machine learning inference," in ASPLOS, 2019.

Digital Library

[7]

W. Li et al., "Timely: Pushing data movements and interfaces in pim accelerators towards local and in time domain," in ISCA. IEEE, 2020.

[8]

M. A. Lastras-Montano et al., "Resistive random-access memory based on ratioed memristors," Nature Electronics, pp. 466--472, 2018.

[9]

Y. Vlasov et al., "High-throughput silicon nanophotonic wavelength-insensitive switch for on-chip optical networks," pp. 242--246.

[10]

C. Kachris and I. Tomkos, "A survey on optical interconnects for data centers," IEEE Communications Surveys & Tutorials, 2012.

[11]

X. Ma et al., "Tiny but accurate: A pruned, quantized and optimized memristor crossbar framework for ultra efficient dnn implementation," in ASP-DAC, 2020, pp. 301--306.

[12]

Y. Wang, Q. Wang, S. Shi, X. He, Z. Tang, K. Zhao, and X. Chu, "Benchmarking the performance and energy efficiency of ai accelerators for ai training," in CCGRID. IEEE, 2020, pp. 744--751.

[13]

S. Devarajan et al., "A 16-bit, 125 ms/s, 385 mw, 78.7 db snr cmos pipeline adc," JSSCC, pp. 3305--3313, 2009.

[14]

W. Ke et al., "A 14-bit 100-ms/s cmos pipelined adc with 11.3 enob," Journal of Semiconductors, p. 085015, 2013.

[15]

W. Wang et al., "A 4.7-mw 12-bit 100-ms/s hybrid dac," in ICSICT. IEEE, 2018, pp. 1--3.

[16]

R. Ma et al., "A 10-bit 100-ms/s 5.23-mw sar adc in 0.18-μm cmos," Microelectronics Journal, pp. 63--72, 2018.

[17]

M. Baghbanmanesh et al., "A 10-bit radiation-hardened by design (rhbd) sar adc for space applications," in NGCAS. IEEE, 2017.

[18]

X. Chen et al., "Modeling and analysis of optical modulators based on free-carrier plasma dispersion effect," TCAD, pp. 977--990, 2019.

[19]

Z. Wang et al., "A holistic modeling and analysis of optical-electrical interfaces for inter/intra-chip interconnects," TVLSI, 2016.

[20]

B. E. Jonsson et al., "Area efficiency of adc architectures," in ECCTD. IEEE, 2011, pp. 560--563.

[21]

M. Chakir et al., "A low power 6-bit current-steering dac in 0.18-μm cmos process," in ISCV. IEEE, 2015, pp. 1--5.

[22]

Y. Hsu et al., "2.6 tbit/s on-chip optical interconnect supporting mode-division-multiplexing and PAM-4 signal," pp. 1052--1055.

Cited By

Liu JMa YNakamura YWang Y(2025)OpticalHDC: Ultra-fast Photonic Hyperdimensional Computing AcceleratorProceedings of the 30th Asia and South Pacific Design Automation Conference10.1145/3658617.3697709(748-753)Online publication date: 20-Jan-2025
https://dl.acm.org/doi/10.1145/3658617.3697709
Li XLiu JZhang YLiu YZhang JLi CChen SFu YTian FZhang WXu J(2024)PhotonNTT: Energy-Efficient Parallel Photonic Number Theoretic Transform Accelerator2024 Design, Automation & Test in Europe Conference & Exhibition (DATE)10.23919/DATE58400.2024.10546638(1-6)Online publication date: 25-Mar-2024
https://doi.org/10.23919/DATE58400.2024.10546638
Wang RWang PLyu CLuo GMa JZhou XZhang YPan J(2023)Photonic Binary Convolutional Neural Network Based on Microring Resonator ArrayIEEE Photonics Technology Letters10.1109/LPT.2023.327214835:12(664-667)Online publication date: 15-Jun-2023
https://doi.org/10.1109/LPT.2023.3272148

Recommendations

IMC: energy-efficient in-memory convolver for accelerating binarized deep neural network
NCS '17: Proceedings of the Neuromorphic Computing Symposium

Deep Convolutional Neural Networks (CNNs) are widely employed in modern AI systems due to their unprecedented accuracy in object recognition and detection. However, it has been proven that the main bottleneck to improve large scale deep CNN based ...
Signed Convolution in Photonics with Phase-Change Materials Using Mixed-Polarity Bitstreams
ASPDAC '24: Proceedings of the 29th Asia and South Pacific Design Automation Conference

As AI continues to grow in importance, in order to reduce its carbon footprint and utilization of computer resources, numerous alternatives are under investigation to improve its hardware building blocks. In particular, in convolutional neural networks (...
Accelerating On-Chip Training with Ferroelectric-Based Hybrid Precision Synapse
In this article, we propose a hardware accelerator design using ferroelectric transistor (FeFET)-based hybrid precision synapse (HPS) for deep neural network (DNN) on-chip training. The drain erase scheme for FeFET programming is incorporated for both ...

Comments

Information & Contributors

Information

Published In

cover image ACM Conferences

DAC '22: Proceedings of the 59th ACM/IEEE Design Automation Conference

July 2022

1462 pages

ISBN:9781450391429

DOI:10.1145/3489517

General Chair:
Rob Oshana
NXP

Copyright © 2022 ACM.

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected].

Sponsors

SIGDA: ACM Special Interest Group on Design Automation
IEEE CEDA

Publisher

Association for Computing Machinery

New York, NY, United States

Publication History

Published: 23 August 2022

Permissions

Request permissions for this article.

Request Permissions

Check for updates

Author Tags

Qualifiers

Research-article

Conference

DAC '22

Sponsor:

SIGDA

DAC '22: 59th ACM/IEEE Design Automation Conference

July 10 - 14, 2022

California, San Francisco

Acceptance Rates

Overall Acceptance Rate 1,770 of 5,499 submissions, 32%

Upcoming Conference

DAC '25

Sponsor:
sigda

62nd ACM/IEEE Design Automation Conference

June 22 - 26, 2025

San Francisco , CA , USA

Contributors

Other Metrics

View Article Metrics

Bibliometrics & Citations

Bibliometrics

Article Metrics

3
Total Citations
View Citations
689
Total Downloads

Downloads (Last 12 months)136
Downloads (Last 6 weeks)8

Reflects downloads up to 05 Mar 2025

Other Metrics

View Author Metrics

Citations

Cited By

Liu JMa YNakamura YWang Y(2025)OpticalHDC: Ultra-fast Photonic Hyperdimensional Computing AcceleratorProceedings of the 30th Asia and South Pacific Design Automation Conference10.1145/3658617.3697709(748-753)Online publication date: 20-Jan-2025
https://dl.acm.org/doi/10.1145/3658617.3697709
Li XLiu JZhang YLiu YZhang JLi CChen SFu YTian FZhang WXu J(2024)PhotonNTT: Energy-Efficient Parallel Photonic Number Theoretic Transform Accelerator2024 Design, Automation & Test in Europe Conference & Exhibition (DATE)10.23919/DATE58400.2024.10546638(1-6)Online publication date: 25-Mar-2024
https://doi.org/10.23919/DATE58400.2024.10546638
Wang RWang PLyu CLuo GMa JZhou XZhang YPan J(2023)Photonic Binary Convolutional Neural Network Based on Microring Resonator ArrayIEEE Photonics Technology Letters10.1109/LPT.2023.327214835:12(664-667)Online publication date: 15-Jun-2023
https://doi.org/10.1109/LPT.2023.3272148

View Options

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Full Access

Get this Publication

View options

PDF

View or Download as a PDF file.

eReader

View online with eReader.

Figures

Tables

Media

View Table of Conten