Work on a U.K. supercomputer is turning heads in an era where processing power is expanding rapidly, and artificial intelligence is taking advantage of these huge speed and performance boosts.
Reuters reports researchers and engineers are working on a “Cambridge-1” computer in the renowned university town, built on an Nvidia DGX SuperPod system to accomplish 400 petaflops, an impressive data crunching rate by any standard.
According to reports, it’s part of a plan to build an AI Center of excellence at Cambridge where the Cambridge-1 computer represents the 29th fastest machine in the world, according to estimates from Top500.
What other computers rank high on this list? There’s the Trinity computer in Los Alamos, as well as SuperMUC-NG in Leibniz Germany, the ABCI in Japan, and the Piz Daint in Switzerland.
Around the world engineers are linking myriad components into elaborate mainframe supercomputers with the potential to move from traditional strictly linear data processing into the wide world of neural network models and artificial intelligence programs driven by inscrutable black box algorithms. Then there’s the quantum computer, a model that is still in its infancy, where scientists are trying to build reliability into a fundamentally dynamic framework, albeit in a puzzle that computer science people have approached for decades:
“Scientists’ approach of spreading the information of one qubit—a ‘logical qubit’—among many physical ones traces its roots to the early days of ordinary computers in the 1950s,” writes Adrian Cho at ScienceMag, in a broad survey of how quantum builds on conventional supercomputing methods like those displayed by the Cambridge-1 and its peers. “The bits of early computers consisted of vacuum tubes or mechanical relays, which were prone to flip unexpectedly. To overcome the problem, famed mathematician John von Neumann pioneered the field of error correction.
It’s an exciting frontier in technology, but one that has to be approached carefully. With experts like Bill Gates and Elon Musk trumpeting the necessity of ethical AI, the human in the loop is critically important in these types of innovations.
