Teacher
Professional
- Messages
- 2,669
- Reaction score
- 819
- Points
- 113
From electricity to light: the transition that will change everything.
Researchers at the University of Pennsylvania have developed a new computer chip that runs on light, not electricity. This discovery promises to significantly speed up the training of artificial intelligence (AI) models, increasing the speed of data transfer and reducing the electricity required for their operation.
Today, humanity is creating exascale-class supercomputers capable of performing a quintillion operations per second. Despite the growth in computing power, modern technologies are still based on the principles of the 1960s.
Recently, scientists have been actively developing computing systems based on quantum mechanics, but their widespread adoption is still far away. At the same time, the growing need to process huge amounts of data for AI increases the requirements for computing power and leads to an increase in energy consumption.
A team of researchers led by Professor Nader Enghat from the School of Engineering and Applied Sciences has developed a silicon photonics (SiPh) chip that performs mathematical calculations using light. Light is the fastest way to transmit data, and the use of widely available silicon allows you to quickly scale the technology.
The main goal of the researchers was to create a chip capable of performing vector-matrix multiplication operations, which are widely used in neural networks that underlie the development of modern AI models.
Instead of completely rethinking the manufacturing process, the researchers changed the height of the chip in certain places to control the propagation of light inside it. This allowed the light to travel strictly in a straight line without scattering inside the chip.
The developed chip, which cannot be hacked, is already ready for implementation. The researchers collaborated with a commercial factory to produce their SiPh chips, adapting their design to fit existing market sizes.
Firoz Aflatouni, an associate professor in the university's Department of Electrical and Systems Engineering, noted that these chips could replace the graphics processing units (GPUs) used by companies to train and classify AI models. SiPh chips can serve as a complement to the existing infrastructure used by companies in the field of AI.
In addition, SiPh chips allow you to perform calculations faster and with less power consumption, as well as solve data privacy issues. Due to the possibility of parallel execution of multiple calculations, information does not need to be stored in RAM during processing, which makes unauthorized access to data impossible.
The results of the study were published today in the journal Nature Photonics.
Researchers at the University of Pennsylvania have developed a new computer chip that runs on light, not electricity. This discovery promises to significantly speed up the training of artificial intelligence (AI) models, increasing the speed of data transfer and reducing the electricity required for their operation.
Today, humanity is creating exascale-class supercomputers capable of performing a quintillion operations per second. Despite the growth in computing power, modern technologies are still based on the principles of the 1960s.
Recently, scientists have been actively developing computing systems based on quantum mechanics, but their widespread adoption is still far away. At the same time, the growing need to process huge amounts of data for AI increases the requirements for computing power and leads to an increase in energy consumption.
A team of researchers led by Professor Nader Enghat from the School of Engineering and Applied Sciences has developed a silicon photonics (SiPh) chip that performs mathematical calculations using light. Light is the fastest way to transmit data, and the use of widely available silicon allows you to quickly scale the technology.
The main goal of the researchers was to create a chip capable of performing vector-matrix multiplication operations, which are widely used in neural networks that underlie the development of modern AI models.
Instead of completely rethinking the manufacturing process, the researchers changed the height of the chip in certain places to control the propagation of light inside it. This allowed the light to travel strictly in a straight line without scattering inside the chip.
The developed chip, which cannot be hacked, is already ready for implementation. The researchers collaborated with a commercial factory to produce their SiPh chips, adapting their design to fit existing market sizes.
Firoz Aflatouni, an associate professor in the university's Department of Electrical and Systems Engineering, noted that these chips could replace the graphics processing units (GPUs) used by companies to train and classify AI models. SiPh chips can serve as a complement to the existing infrastructure used by companies in the field of AI.
In addition, SiPh chips allow you to perform calculations faster and with less power consumption, as well as solve data privacy issues. Due to the possibility of parallel execution of multiple calculations, information does not need to be stored in RAM during processing, which makes unauthorized access to data impossible.
The results of the study were published today in the journal Nature Photonics.
