How Can AI Researchers Save Energy? By Going Backward.
1 min read
Summary
Progress in conventional computing is slowing as the industry hits fundamental physical limitations regarding chip size, prompting researchers to consider other methods for improving computational efficiency.
Reversible computing, which involves computers that can run computations backward and forwards, is one such technology that could save energy and improve processor performance.
By never deleting data, computers could avoid wasting energy, while also being able to run computations in parallel could also offer performance advantages for AI algorithms.
Progress on reversible computing stalled in the 2000s, but recent advances and the growing need for more efficient computing due to the limits of conventional chips have prompted a revival in interest.
Vaire Computing, founded by University of Cambridge researcher Hannah Earley, is one company aiming to develop a commercial version of a reversible chip.
