Penn Engineers Introduce Groundbreaking Silicon-Photonics Chip for AI at Light Speed
In a significant leap forward for artificial intelligence (AI) computing, Penn Engineers have unveiled a cutting-edge Silicon-Photonics chip that harnesses light waves, rather than electricity, to execute complex mathematical operations crucial for training AI. This revolutionary chip not only has the potential to drastically enhance computer processing speed but also promises a considerable reduction in energy consumption.
The silicon-photonic (SiPh) chip represents a groundbreaking fusion of research by Benjamin Franklin Medal Laureate and H. Nedwill Ramsey Professor Nader Engheta and Associate Professor Firooz Aflatouni. Engheta’s expertise lies in manipulating materials at the nanoscale to conduct mathematical computations using light, the fastest mode of communication, while Aflatouni’s group has pioneered nanoscale silicon devices.
The chip’s design aims to overcome the limitations of current chips, which have roots in principles established during the early days of computing in the 1960s. Published in Nature Photonics, the paper details the collaboration’s success in developing a platform for vector-matrix multiplication, a fundamental mathematical operation in the realm of neural networks that power contemporary AI tools.
Engheta explains that instead of using a uniform silicon wafer, the team opted for variations in height, making the silicon thinner (around 150 nanometers) in specific regions. These height variations, without the introduction of additional materials, enable the control of light propagation through the chip. The resulting patterns of scattered light facilitate mathematical calculations at the speed of light.
Aflatouni notes that due to the chip’s design meeting the constraints of commercial foundries, it is ready for commercial applications and could potentially enhance graphics processing units (GPUs), which are in high demand for new AI system development.
“They can adopt the Silicon Photonics platform as an add-on,” suggests Aflatouni, “and then you could speed up training and classification.”
Beyond the remarkable gains in speed and energy efficiency, the chip also brings privacy advantages. The simultaneous occurrence of many computations eliminates the need to store sensitive information in a computer’s working memory, making a future computer powered by this technology virtually unhackable.
“No one can hack into a non-existing memory to access your information,” affirms Aflatouni.
This groundbreaking study, conducted at the University of Pennsylvania School of Engineering and Applied Science, was supported by grants from the U.S. Air Force Office of Scientific Research’s Multidisciplinary University Research Initiative (AFOSR MURI) and the U.S. Office of Naval Research (ONR).
Co-authors of the paper include Vahid Nikkhah, Ali Pirmoradi, Farshid Ashtiani, and Brian Edwards of Penn Engineering. The Silicon-Photonics chip promises a seismic shift in the AI computing landscape, opening new doors for unprecedented speed, efficiency, and security in artificial intelligence applications.
In a significant leap forward for artificial intelligence (AI) computing, Penn Engineers have unveiled a cutting-edge Silicon-Photonics chip that harnesses light waves, rather than electricity, to execute complex mathematical operations crucial for training AI. This revolutionary chip not only has the potential to drastically enhance computer processing speed but also promises a considerable reduction in…