A team of researchers from the University of California has built, for the first time, a working memory-crunching computer (memcomputer) prototype. This memcomputer is a computer that solves problems by crunching numbers and storing results simultaneously, rather than as completely separate processes, as is done with all modern computing machines. In that respect, it should work much more like the human brain.
Scheme of the memcomputing architecture used in this work to solve the subset sum problem. Credit: Science Advances
One of the theoretical physicists in charge of the project, Massimiliano Di Ventra, explains that he and his colleagues have managed to build such a computer that is capable of solving extraordinarily difficult computational problems.
“If we work with a different paradigm of computation, those problems that are notoriously difficult to solve with current computers can be solved more efficiently with memcomputers,” Di Ventra explains.
The team of researchers only has a proof of concept (POC) at their disposal, and certain issues must still be addressed in order to manufacture such memcomputers. Actually, such computers have been theoretically predicted 45 years ago. It was only in 2008 that researchers could manufacture the first memcomputer, though.
One such issue, Di Ventra explains, is to scale the current version up to a variety of memprocessors. The unique encoding system of memcomputers makes them vulnerable to unpredictable fluctuations which could cause errors. Di Ventra and his team is now investigating whether memcomputers could be designed to encode data differently.
The scientists detailed their findings online July 3 in the journal Science Advances.