Summary of "New chip opens door to ai computing at light speed"

 SUMMARY

Penn Engineers have developed a new chip that uses light waves instead of electricity to perform complex math needed for AI training, potentially revolutionizing computer processing speed and energy efficiency. By combining nanoscale material manipulation research with silicon-photonics technology, this chip enables computations at the speed of light, surpassing current chip limitations. The chip's unique design allows for faster calculations without the need for additional materials and offers privacy advantages by not storing sensitive information in the computer's memory. This innovation, ready for commercial use, could enhance AI systems' performance and security, making them virtually unhackable.

KEY POINTS 

• Penn Engineers have developed a new silicon-photonic (SiPh) chip that uses light waves to perform complex mathematical operations for AI training, potentially increasing processing speed and reducing energy consumption.
• The SiPh chip's design integrates Nader Engheta's research on nanoscale material manipulation for mathematical computations with light and the SiPh platform's use of silicon, aiming to surpass the limitations of traditional computing chips.
• The chip enables vector-matrix multiplication essential for neural networks by varying the height of silicon in specific regions, allowing light to scatter in patterns for rapid calculations.
• This design is ready for commercial applications and could be integrated into GPUs to enhance AI system development by speeding up training and classification processes.
• The chip offers privacy advantages by enabling simultaneous computations without the need to store sensitive information in working memory, making future computers using this technology virtually unhackable.

QUOTES 

“We decided to join forces,” says Engheta, leveraging the fact that Aflatouni’s research group has pioneered nanoscale silicon devices.

Instead of using a silicon wafer of uniform height, explains Engheta, “you make the silicon thinner, say 150 nanometers,” but only in specific regions.

“They can adopt the Silicon Photonics platform as an add-on,” says Aflatouni, “and then you could speed up training and classification.”

“No one can hack into a non-existing memory to access your information,” says Aflatouni.

SOURCES 

This study was conducted at the University of Pennsylvania School of Engineering and Applied science and supported in part by a grant from the U.S. Air Force Office of Scientific Research’s (AFOSR) Multidisciplinary University Research Initiative (MURI) to Engheta (FA9550-21-1-0312) and a grant from the U.S. Office of Naval Research (ONR) to Aflatouni (N00014-19-1-2248).

Additional co-authors include Vahid Nikkhah, Ali Pirmoradi, Farshid Ashtiani and Brian Edwards of Penn Engineering.

https://blog.seas.upenn.edu/new-chip-opens-door-to-ai-computing-at-light-speed by Ian Scheffler

AUTHORS

Ian Scheffler

Olivia J. McMahon

Melissa Pappas

Holly Wojcik

Lang: English
• Blog 
• Blogs 
• Blogging 
• Blogger 
• Bloggers