Feb 19 2024
Penn Engineers have created a novel chip that does the intricate math required to teach artificial intelligence (AI) using light waves rather than electricity. The device might significantly reduce the energy usage of computers while also increasing their processing speed.
To perform mathematical computations using light, the fastest possible form of communication, Benjamin Franklin Medal laureate and H. Nedwill Ramsey Professor Nader Engheta's groundbreaking research in material manipulation at the nanoscale is brought together for the first time in the design of a silicon-photonic (SiPh) chip.
The SiPh platform uses silicon, an inexpensive and plentiful element used to mass-produce computer chips.
One potential path toward creating computers that surpass the capabilities of current chips, which are fundamentally built on the same ideas as chips from the early days of the computing revolution in the 1960s, is the interaction of light waves with matter.
The development of the new chip is described by Engheta's group and Associate Professor Firooz Aflatouni of Electrical and Systems Engineering in the journal published in Nature Photonics.
“We decided to join forces,” says Engheta, leveraging the fact that Aflatouni’s research group has pioneered nanoscale silicon devices.
Their objective was to create a platform that could carry out vector-matrix multiplication, a fundamental mathematical operation used in the construction and operation of neural networks, the computer architecture that underpins modern AI systems.
Instead of using a silicon wafer of uniform height, Engheta explained, “You make the silicon thinner, say 150 nm.”
Without the use of any additional materials, those height variations offer a way to regulate how light travels through the chip. This is because the height variations can be distributed to cause light to scatter in particular patterns, enabling the chip to execute mathematical operations at the speed of light.
Aflatouni claims that this design is already ready for commercial applications and could be modified for use in graphics processing units (GPUs), the demand for which has increased dramatically with the widespread interest in creating new AI systems due to the limitations imposed by the commercial foundry that produced the chips.
They can adopt the Silicon Photonics platform as an add-on, and then you could speed up training and classification.
Firooz Aflatouni, Associate Professor, Department of Electrical and Systems Engineering, Penn Engineering
The privacy benefits of Engheta and Aflatouni's chip go beyond its increased speed and reduced energy consumption. Since multiple computations can occur concurrently, sensitive data would not need to be kept in a computer's working memory, making future computers equipped with this technology essentially unhackable.
Aflatouni said, “No one can hack into a non-existing memory to access your information.”
The research was conducted at the University of Pennsylvania School of Engineering and Applied Science and supported by a grant from the US Air Force Office of Scientific Research’s Multidisciplinary University Research Initiative and as a part of a grant from the US Office of Naval Research (ONR) to Afaltouni.
Vahid Nikkhah, Ali Pirmoradi, Farshid Ashtiani, and Brian Edwards of Penn Engineering also co-authored the research.
Journal Reference:
Nikkhah, V., et.al. (2024). Inverse-designed low-index-contrast structures on a silicon photonics platform for vector–matrix multiplication. Nature Photonics. doi.org/10.1038/s41566-024-01394-2