It is a known fact that today’s computers need a lot of energy to undertake
AI tasks
. To address the
energy consumption
problem and accelerate AI computation, a group of engineers from the School of Engineering and Applied Science at the University of Pennsylvania has developed a chip that uses light to perform the complex math essential to training AI.
Engineers manipulated materials at the nanoscale to perform mathematical computations using light — the fastest possible means of communication — with a silicon-photonic (SiPh) chip. Silicon is the cheap, abundant element used to mass-produce computer chips.
The interaction of light waves with matter is one of the possible avenues for developing computers that supersede the limitations of today’s chips.
Professor Nader Engheta and his group from the University of Pennsylvania, along with that of Firooz Aflatouni, Associate Professor in Electrical and Systems Engineering, collaborated to develop a platform for performing vector-matrix multiplication – a core mathematical operation in the development and function of neural networks.
A neural network is essentially a computer architecture that powers today’s AI tools. Consider it as a brain in artificial intelligence that teaches computers to process data in a similar way as a human brain.
How the chip works
As per the paper published by engineers, instead of using a silicon wafer of uniform height, “you make the silicon thinner, say 150 nanometers,” but only in specific regions, Engheta explained. He said that the variations in height — without the addition of any other materials — provide a means of controlling the propagation of light through the chip.
He said that these variations in height can be distributed to cause light to scatter in specific patterns, which in turn allows the chip to perform mathematical calculations at the speed of light.
Aflatouni noted that the design of the chip is already ready for commercial applications, and could potentially be adapted for use in graphics processing units (GPUs). GPU’s demand has skyrocketed as the interest in developing new AI systems has increased in recent years.
“They can adopt the Silicon Photonics platform as an add-on and then you could speed up training and classification,” Aflatouni said.
Privacy advantages of the chip
In addition to faster speed and less energy consumption, the chip is also said to offer privacy advantages over current chips. The engineers say that since there will be many computations that can happen simultaneously, there will be no need to store sensitive information in a computer’s working memory. This will render a future computer powered by such technology virtually unhackable.
“No one can hack into a non-existing memory to access your information,” says Aflatouni.
English (US) ·