When it comes to energy efficiency, today’s computers eat up some eight orders of magnitude more than the theoretical minimum. Clearly, there’s room for improvement.
Today, Quentin Herr and pals at Northrop Grumman Systems in Baltimore reveal a superconducting logic device that uses 300 times less power than conventional silicon chips. That’ll be a shock to chip-makers whose attempts at energy efficiency improvements pale in comparison.
The laws of physics help, of course. The power dissipated by a conventional transistor depends on its size and on the material from which it is made.
By comparison, the energy dissipated by the superconducting equivalent, the Josephson junction, is determined only by thermal noise, the system being entirely free of the troublesome resistance that eats up energy in conventional devices.
To exploit this, Herr and co developed and built a new type of device known as a reciprocal quantum logic chip. Their superconducting chip is made of niobium, operates at 4.2 kelvin and consists of 1600 Josephson junctions .
They say it can operate at clock speeds of 6 Ghz with a 6 mW power supply, that faster speeds are possible and that this performance can scale to chips with a million junctions. And it does it with a bit error rate of less than 10^-40. That’s practically nothing.
But the jaw dropper is the power consumption which is only three orders of magnitude above the theoretical limit.
Maximize business value with data-driven strategies
Every organization is now collecting data, but few are truly data driven. Here are five ways data can transform your business.
Modern security demands an empathy-first approach to insiders
While attention is often focused on threats from outside the organization, employees too can pose a risk to security—even inadvertently.
The book ban movement has a chilling new tactic: harassing teachers on social media
Educators who stand up to conservative activists are being harassed and called “groomers” online, turning them into potential targets for real-world violence.
OpenAI is ready to sell DALL-E to its first million customers
But the company has had to rush out fixes to the image-making model’s worst flaws to do so.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.