University of Pennsylvania’s engineers announced a groundbreaking silicon-photonic (SiPh) chip. This chip performs sophisticated mathematical operations for AI training using light waves instead of energy. This innovative technology can boost computer processing speed and reduce energy use.
The SiPh chip, created by famous Nader Engheta and Firooz Aflatouni, uses silicon, a cheap and abundant computer chip material, and Engheta’s skill in manipulating nanoscale materials for mathematical computations using light, the fastest form of communication.
Engheta’s team worked with Penn’s School of Engineering and Applied Sciences’s associate professor of electrical and systems engineering, Aflatouni, on the Nature Photonics study. They wanted to create a platform for vector-matrix multiplication, a crucial mathematical process for neural networks, the core of modern AI systems.
The consistent height of silicon wafers is broken by this innovative chip architecture. Engheta explains how to thin silicon to 150 nanometers in specified places. These changes allow exact light transmission within the semiconductor without extra materials. Strategic height differences manipulate light scattering, speeding up chip mathematical calculations.
This solution could solve the problems with present computer chips, which are based on 1960s computing ideas. Using light waves instead of electricity allows for more advanced computers.
This revolutionary technology’s practical uses are widely anticipated by scientists. A future of AI computing at light speed with unequalled processing power and energy efficiency is possible with the SiPh chip.
