Chinese Journal of Lasers, Volume. 50, Issue 5, 0500001(2023)

Future of Optical Computing: Analog or Digital?

Guoqing Ma1,2, Changhe Zhou3、*, Rongwei Zhu1,2, Fenglu Zheng1,2, Junjie Yu1,2、**, and Guohai Situ1,2、***
Author Affiliations
  • 1Laboratory of Information Optics and Optoelectronic Technology, Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800, China
  • 2Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences,Beijing 100049, China
  • 3The Institute of Photonics Technology, Jinan University, Guangzhou 510632, Guangdong, China
  • show less
    Figures & Tables(7)
    Methods to improve computing power and energy efficiency ratio in the “post-Moore era” (FinFET is field effect transistor, EUV is extreme ultraviolet lithography, GAA is gate-all-around FET, UCIe is universal chiplet interconnect express, and chiplet is an integrated circuit block)
    Overview of development model and main technologies of optical computing
    Representative achievements of early digital optical computing technologies
    Fundamental building blocks and releated technical solutions of major optical computing based on photonic integrated circuits (PICs). (a) Main technical schemes of analog optical computing for PICs; (b) basic building blocks required for optical computing based on PICs; (c) main technical schemes of digital optical computing for PICs
    Major advances in free-space analog optical computing. (a) Image filtering and convolution operations are achieved by modulating point spread function (PSF); (b) operations such as optical differentiation/integration and VM-MAC are achieved by modulating MTF (modulation transfer function) of the system on the spectral plane; (c) modulate parameters of each pixel in each diffractive or scattered layer to realize optical interconnection and optical transformation
    Optical matrix convolution computing system based on multi-imaging-casting architecture[138]
    • Table 1. Characteristics and typical techniques of analog and digital optical computing

      View table

      Table 1. Characteristics and typical techniques of analog and digital optical computing

      Optical computingAdvantageDisadvantageTypical technique
      Analog optical computing

      ● High information density,no quantization error

      ● The design freedom of optoelectronic materials and micro-nano structures is high

      ● Fast computing speed

      ● Match with deep learning algorithm

      ● Signals are greatly affected by cascaded system noise,and new fault tolerance methods need to be developed

      ● Dedicated information processing

      ● High computing accuracy requires high linearity and dynamic range of system

      ● Optical Fourier transform

      ● Smart metamaterials

      ● ONNs

      ● Optical convolutional filtering

      Digital optical computing

      ● No cascading noise,strong anti-interference ability

      ● The computing results can be easily transplanted into existing electronic digital devices for storage and processing

      ● Under the same system,higher computing accuracy can be obtained through coding

      ● The resolution of time and space is reduced,and the computing scale is limited

      ● The complexity of signal processing algorithm is high

      ● Each coded bit must be calculated accurately,which requires high signal-to-noise ratio(SNR)of the system

      ● Optical logic device

      ● Optical coding methods

      ● Directed logic

      ● TOC

    Tools

    Get Citation

    Copy Citation Text

    Guoqing Ma, Changhe Zhou, Rongwei Zhu, Fenglu Zheng, Junjie Yu, Guohai Situ. Future of Optical Computing: Analog or Digital?[J]. Chinese Journal of Lasers, 2023, 50(5): 0500001

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: reviews

    Received: Sep. 2, 2022

    Accepted: Oct. 11, 2022

    Published Online: Feb. 23, 2023

    The Author Email: Zhou Changhe (chazhou@mail.shcnc.ac.cn), Yu Junjie (Junjiey@siom.ac.cn), Situ Guohai (ghsitu@siom.ac.cn)

    DOI:10.3788/CJL221209

    Topics