Researchers from China’s Tsinghua University propose all-analog photoelectronic chip for high-speed vision tasks
Chinese article by 项睿
English Editor 张未名
10-31 16:41

By Greg Gao

(JW Insights) Oct 31 -- Researchers in China's Tsinghua University introduced a novel computational framework known as the all-analog photoelectronic chip, which boasts computing power exceeding commercial high-performance chips by over 3,000 times. This research was recently published in the journal Nature.

This research in computing architecture that transcends Moore’s Law was carried out by Tsinghua University’s Department of Automation, led by Academician Dai Qionghai and Assistant Professor Wu Jiamin, in collaboration with the Department of Electronic Engineering, headed by Associate Professor Fang Lu and Associate Researcher Qiao Fei. The Chinese press call it a "groundbreaking milestone."

The team proposes an all-analog chip combining electronic and light computing (ACCEL). It has a systemic energy efficiency of 74.8 peta-operations per second per watt and a computing speed of 4.6 peta-operations per second, corresponding to more than three and one order of magnitude higher than state-of-the-art computing processors, respectively.

In a comparative analogy, if one were to liken the time it takes for information flow computations within the chip to travel to transportation, the appearance of this chip equates to reducing an eight-hour journey on the Beijing-Guangzhou high-speed railway to just eight seconds. The power needed to operate existing chips for one hour can now power this chip for over five centuries.

One of the lead authors of the paper, Dai Qionghai, remarked, “Developing a brand-new computational architecture for the era of artificial intelligence is a summit to conquer. Yet, bringing this new architecture into practical life to address significant national and societal needs is an even more crucial endeavor and our responsibility.”

The special review article published in Nature suggests that “perhaps the appearance of this chip will accelerate the integration of a new generation of computing architecture into everyday life much sooner than anticipated.”

linkedin twitter facebook line
Copy succeeded