Tuesday 22 May 2018

Lasers Could Make Computers 1 Million Times Faster




Lasers Could Make Computers 1 Million Times Faster





An artist's rendering shows polarized light interacting with the honeycomb lattice.
Credit: Stephen Alvey, Michigan Engineering
A billion operations per second isn't cool. Know what's cool? A million billion operations per second.
That's the promise of a new computing technique that uses laser-light pulses to make a prototype of the fundamental unit of computing, called a bit, that could switch between its on and off, or "1" and "0" states, 1 quadrillion times per second. That's about 1 million times faster than the bits in modern computers.
Conventional computers (everything from your calculator to the smartphone or laptop you're using to read this) think in terms of 1s and 0s. Everything they do, from solving math problems, to representing the world of a video game, amounts to a very elaborate collection of 1-or-0, yes-or-no operations. And a typical computer in 2018 can use silicon bits to perform more or less 1 billion of those operations per second. [Science Fact or Fiction? The Plausibility of 10 Sci-Fi Concepts]
 
In this experiment, the researchers pulsed infrared laser light on honeycomb-shaped lattices of tungsten and selenium, allowing the silicon chip to switch from "1" to "0" states just like a normal computer processor — only a million times faster, according to the study, which was published in Nature on May 2.
That's a trick of how electrons behave in that honeycomb lattice.
In most molecules, the electrons in orbit around them can jump into several different quantum states, or "pseudospins," when they get excited. A good way to imagine these states is as different, looping racetracks around the molecule itself. (Researchers call these tracks "valleys," and the manipulation of these spins "valleytronics.")
When unexcited, the electron might stay close to the molecule, turning in lazy circles. But excite that electron, perhaps with a flash of light, and it will need to go burn off some energy on one of the outer tracks.
 The tungsten-selenium lattice has just two tracks around it for excited electrons to enter. Flash the lattice with one orientation of infrared light, and the electron will jump onto the first track. Flash it with a different orientation of infrared light, and the electron will jump onto the other track. A computer could, in theory, treat those tracks as 1s and 0s. When there's an electron on track 1, that's a 1. When it's on track 0, that's a 0.
 
Crucially, those tracks (or valleys) are sort of close together, and the electrons don't need to run on them very long before losing energy. Pulse the lattice with infrared light type one, and an electron will jump onto track 1, but it will only circle it for "a few femtoseconds," according to the paper, before returning to its unexcited state in the orbitals closer to the nucleus. A femtosecond is one thousand million millionth of a second, not even long enough for a beam of light to cross a single red blood cell.
So, the electrons don't stay on the track long, but once they're on a track, additional pulses of light will knock them back and forth between the two tracks before they have a chance to fall back into an unexcited state. That back-and-forth jostling, 1-0-0-1-0-1-1-0-0-0-1 — over and over in incredibly quick flashes — is the stuff of computing. But in this sort of material, the researchers showed, it could happen much faster than in contemporary chips.
The researchers also raised the possibility that their lattice could be used for quantum computing at room temperature. That's a kind of holy grail for quantum computing, since most existing quantum computers require researchers to first cool their quantum bits down to near absolute zero, the coldest possible temperature. The researchers showed that it's theoretically possible to excite the electrons in this lattice to "superpositions" of the 1 and 0 tracks — or ambiguous states of being kind-of-sort-of fuzzily on both tracks at the same time — that are necessary for quantum-computing calculations.
"In the long run, we see a realistic chance of introducing quantum information devices that perform operations faster than a single oscillation of a lightwave," study lead author Rupert Huber, professor of physics at the University of Regensburg in Germany, said in a statement. However, the researchers didn't actually perform any quantum operations this way, so the idea of a room- temperature quantum computer is still entirely theoretical. And in fact, the classical (regular-type) operations the researchers did perform on their lattice were just meaningless, back-and-forth, 1-and-0 switching. The lattice still hasn't been used to calculate anything. Thus, researchers still have to show that it can be used in a practical computer.
Still, the experiment could open the door to ultrafast conventional computing — and perhaps even quantum computing — in situations that were impossible to achieve until now.
  SOURCE : Live Science.

No comments:

Post a Comment