BEIJING, Feb. 17 (Xinhua) -- Engineers at the University of Pennsylvania have developed a new type of chip that uses light instead of electricity to perform complex calculations necessary to train artificial intelligence (AI). The chip has the potential to radically speed up the processing of computers while also reducing energy consumption. The study was published in the latest issue of Nature Photonics.
For the first time, the chip combines Benjamin Franklin Medalist Nader Enkhta's groundbreaking research on manipulating materials at the nanoscale with a silicon photonics (SiPH) platform. The former involves the use of light for mathematical calculations; The latter uses silicon, a cheap and abundant element used in the mass production of computer chips.
The interaction of light waves with matter represents a possible way to develop a computer that is not limited by the limitations of today's chips. The principle of the new chip is essentially the same as that of the chips at the beginning of the computing revolution in the 60s of the 20th century.
The researchers describe the development process of such a chip in **. Their goal was to develop a platform for performing vector matrix multiplication. Vector matrix multiplication is at the heart of the development and functionality of neural networks, which are the computer architectures that support AI tools today.
Enkhta explained that they can make the silicon wafer thinner, such as 150 nanometers, and use highly uneven silicon wafers, but this is limited to specific areas. Without the need to add any additional materials, these changes in height provide a way to control the propagation of light through the chip, as the change in height can cause the light to be scattered in a specific pattern, allowing the chip to perform mathematical calculations at the speed of light.
In addition to faster speeds and less energy consumption, the new chips also have privacy benefits. Because many computations can be done simultaneously, there is no need to store sensitive information in the computer's working memory, making future computers with such technologies virtually impossible to hack.
Editor-in-Chief Circle].
What does the efficiency of a computer depend on? Essentially, it depends on the speed at which its individual components are running and the density of their arrangement, and light is more desirable in both respects than electricity. Using light beams instead of electrons or currents for computing and storage will process a larger amount of information, higher computing speed, and extremely low power consumption, which is undoubtedly more suitable for the future computing era of AI and human-computer interaction. The research in this paper has great potential and advantages, but further development is needed to achieve commercial application.