A significant discovery, which holds the potential to revolutionize the field of computing, has been made by a research team at the University of Limerick.
This discovery pertains to the design of molecules, offering fresh avenues for probing, controlling, and tailoring materials at the most basic molecular scale. The results have contributed to a global project striving to develop an entirely new hardware platform for artificial intelligence (AI). This platform shows marked improvements in computational speed and energy efficiency.
Professor Damien Thompson, the team leader at the Bernal Institute of the University of Limerick and SSPC's director, discusses the inspiration behind the design, which predominantly revolves around the human brain.
According to Professor Thompson, the natural wiggling and jiggling of atoms is used to process and store information. He states, "The molecules create a multitude of individual memory states as they pivot and bounce around their crystal lattice. We can trace out the path of the molecules inside the device and map each snapshot to a unique electrical state."
This innovative approach is believed to have potential transformative benefits for various computing applications, including those that are memory-intensive and energy-hungry such as online gaming and digital maps.
Previously, neuromorphic platforms, which draw inspiration from the functional structure of the human brain, have only been successful in executing low-accuracy operations. This falls short in comprehensive computing tasks that require higher levels of resolution. The team's newly restructured computing architecture, however, promises to meet this high resolution challenge.
The project lead at the Indian Institute of Science (IISc), Professor Sreetosh Goswami, explained that their most significant achievement is a fully functional neuromorphic accelerator integrated into a circuit board. It has precision high enough to handle AI and machine learning workloads, including neural networks, auto-encoders and generative adversarial networks.
The team looks forward to making further progress and enhancements. The ultimate goal is to replace traditional computers with high-performance 'everyware'. These will be based on energy-efficient and eco-friendly materials and provide ubiquitous information processing integrated into everyday items, from clothing to food packaging to construction materials.
Disclaimer: The above article was written with the assistance of AI. The original sources can be found on ScienceDaily.
