Tsinghua scholars collaborated to develop ultra-high-speed optoelectronic computing chips

In 1965, Gordon Moore, one of the founders of Intel, proposed the “Moore’s Law” that has affected the chip industry for more than half a century: predicting that every two years or so, the number of transistors that an integrated circuit can accommodate will double. In the field of semiconductors, the law of massage has flourished for decades, and “chips” have become an important engine for mankind to enter the intelligent era. However, as transistor size approaches the physical limit, Moore’s Law has slowed down or even failed in the last 10 years. How to build a new generation of computing architecture and establish a “new” order of chips in the era of artificial intelligence has become a frontier hot spot of great concern to the international community.

In response to this problem, Dai Qionghai, professor of the Department of Automation of Tsinghua University and academician of the Chinese Academy of Engineering, Wu Jiamin, assistant professor, Fang Lu, associate professor of the Department of Electronic Engineering, and Qiao Fei, associate researcher, jointly tackled key problems and proposed a new computing architecture that “breaks free” from Moore’s law: optoelectronic analog chips, with computing power more than 3,000 times that of current high-performance commercial chips.

The results were published in the journal Nature in the form of a long article.

If the running time of the vehicle is used as an analogy with the time calculated by the information flow in the chip, then the appearance of this chip is equivalent to shortening the 8-hour running time of the Beijing-Guangzhou high-speed railway to 8 seconds.

Photoelectric chip courtesy of Tsinghua University

The 2023 Nobel Prize in Physics was awarded to attosecond laser technology. As one of the fastest speeds in the universe known to man, light is indispensable in many fields of ultra-high-speed physics. However, it is not an easy task for scientists to do calculations with light. When the computational carrier changes from electricity to light, it needs to use the information carried in the light propagation to make the calculation.

Over the past few years, well-known teams at home and abroad have proposed a variety of designs, but in order to replace existing electronic devices to achieve system-level applications, there are still many international problems: first, how to integrate large-scale computing units on a chip and constrain the degree of error accumulation, second, how to achieve high-speed and efficient on-chip nonlinearity, and third, how to provide an efficient interface between optical computing and electronic signal computing in order to be compatible with the current information society with electronic signals as the main body. If we can’t solve these problems, it will be difficult for computing alone to truly replace the current electronic chips and show its skills in the information society.

In this regard, the research team of Tsinghua University creatively proposed a computing framework for deep optoelectronic fusion. Starting from the most essential physical principles, it combines optical computing based on electromagnetic wave space propagation and pure analog electronic computing based on Kirchhoff’s law, “breaking free” from the physical bottleneck of data conversion speed, accuracy and power consumption in the traditional chip architecture, and breaking through the three international problems of large-scale computing unit integration, efficient nonlinearity and high-speed optoelectronic interface on one chip.

Under the measured performance, the system-level computing power of the optoelectronic fusion chip is thousands of times higher than that of the existing high-performance chip architecture. However, such amazing computing power is only one of the many advantages of this chip.

In the intelligent vision task and traffic scene calculation demonstrated by the R&D team, the system-level energy efficiency (the number of operations that can be performed per unit of energy) of the optoelectronic fusion chip has reached 74.8 Peta-OPS/W, which is more than 4 million times that of the existing high-performance chips. Figuratively speaking, the amount of electricity that was originally enough for an existing chip to work for one hour can be used for more than 500 years.

One of the key factors that currently limits the limits of chip integration is the heat dissipation problem caused by excessive density. The optoelectronic fusion chip operating at ultra-low power consumption will help greatly improve the heating problem of the chip, bringing an all-round breakthrough for the future design of the chip.

Furthermore, the minimum linewidth of the optical part of the chip is only 100 nanometers, and the circuit part is only 180nm CMOS process, which has achieved several orders of magnitude performance improvement compared with the high-performance chip of 7 nanometer process. At the same time, the materials used are easy to obtain and cost only a few tenths of the latter.

Dai Qionghai introduced: “The development of a new computing architecture in the era of artificial intelligence is a peak, and it is more important and our responsibility to truly implement the new architecture into real life and solve the major needs of the national economy and people’s livelihood.” A special review of the research published in the journal Nature also pointed out that “perhaps the emergence of this chip will allow a new generation of computing architecture to enter daily life much earlier than expected.” (Source: Chen Bin, China Science News)

Related Paper Information:

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button