Engineers at the University of Pennsylvania have created a groundbreaking chip that uses light waves instead of electricity to conduct complex mathematical operations crucial for training artificial intelligence. This chip has the potential to greatly enhance the speed of computer processing while simultaneously reducing energy consumption.
The silicon-photonic (SiPh) chip merges the innovative research of Benjamin Franklin Medal Laureate and H. Nedwill Ramsey Professor Nader Engheta on nanoscale material manipulation for mathematical computation using light, the fastest form of communication, with the SiPh platform that utilizes silicon, a cost-effective and abundant element employed in the mass production of computer chips.
The utilization of light waves interacting with matter signifies a promising route towards developing computers that surpass the limitations of current chips, which are fundamentally built on principles originating from the early days of computing in the 1960s.
In a publication in Nature Photonics, Engheta's team collaborated with Firooz Aflatouni, Associate Professor in Electrical and Systems Engineering, to introduce this new chip. Engheta highlights the strategic partnership, leveraging Aflatouni's expertise in nanoscale silicon devices to create a platform specializing in vector-matrix multiplication, a critical mathematical operation essential for neural networks, the foundational architecture supporting modern AI applications.
Through a unique design approach involving the manipulation of silicon thickness in specific regions, light propagation control is achieved without additional materials. This design feature permits the chip to execute mathematical computations at the speed of light, revolutionizing processing capabilities.
Aflatouni emphasizes that the chip design is primed for commercial applications, potentially extending to integration with graphics processing units (GPUs) to accelerate training and classification tasks amidst the growing demand for AI technologies.
Apart from speed and energy efficiency benefits, the Engheta and Aflatouni chip offers enhanced privacy features. By enabling simultaneous computations, sensitive data does not need to be stored in the computer's memory, heightening security by making it exceedingly difficult to access personal information. Aflatouni notes the superior security aspect: "No one can hack into a non-existing memory to access your information."
Supported in part 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), this study was conducted at the University of Pennsylvania School of Engineering and Applied Science, showcasing the cutting-edge advancements in computer chip technology.
