Slashdot Mirror
A Peek Inside D-Wave's Quantum Computing Hardware
JeremyHsu writes: A one-second delay can still seem like an eternity for a quantum computing machine capable of running calculations in mere millionths of a second. That delay represents just one of the challenges D-Wave Systems overcame in building its second-generation quantum computing machine known as D-Wave Two — a system that has been leased to customers such as Google, NASA and Lockheed Martin. D-Wave's rapid-scaling approach to quantum computing has plenty of critics, but the company's experience in building large-scale quantum computing hardware could provide valuable lessons for everyone, regardless of whether the D-Wave machines live up to quantum computing's potential by proving they can outperform classical computers. (D-Wave recently detailed the hardware design changes between its first- and second-generation quantum computing machines in the the June 2014 issue of the journal IEEE Transactions on Applied Superconductivity.)
"We were nervous about going down this path," says Jeremy Hilton, vice president of processor development at D-Wave Systems. "This architecture requires the qubits and the quantum devices to be intermingled with all these big classical objects. The threat you worry about is noise and impact of all this stuff hanging around the qubits. Traditional experiments in quantum computing have qubits in almost perfect isolation. But if you want quantum computing to be scalable, it will have to be immersed in a sea of computing complexity.
"We were nervous about going down this path," says Jeremy Hilton, vice president of processor development at D-Wave Systems. "This architecture requires the qubits and the quantum devices to be intermingled with all these big classical objects. The threat you worry about is noise and impact of all this stuff hanging around the qubits. Traditional experiments in quantum computing have qubits in almost perfect isolation. But if you want quantum computing to be scalable, it will have to be immersed in a sea of computing complexity.
Well yeah, but only upon observation.
There has also been a couple of articles in Slashdot where the customers didn't even know if it was a real quantum computer or not. This D-Wave's product has been a very weird thing from the beginning.
Here's a fairly complete anti D-Wave article.
http://www.scottaaronson.com/b...
And rather than just try for first post ... here are the articles in question:
http://tech.slashdot.org/story...
http://science.slashdot.org/st...
Build it, and they will come^Hplain.
You can't LOOK at it!
"A one-second delay can still seem like an eternity for a quantum computing machine capable of running calculations in mere millionths of a second."
Millionths!?!
Um, last time I looked, most microcontroller chips could make calculations in "mere millionths of a second".
Maybe this photo will help to make the concept more clear. The quantum entanglement pairs are seen on the table, with the operator holding one of them. The little colored spheres represent the current state. Calculations performed by the processor can be seen on the background display. Of course this is just a radically simplified representation of the general idea.
And supposedly it is no faster than a real computer. What gives?
It's hard to say because it's all "secret sauce" (so everybody just plunks their heels down on some position rather than admit "I don't know") but one thing that's interesting to me is that a handful of blokes out of Canada appear to have built a computer that's about as fast as a Xeon that Intel needed a few billion dollars, thousands of people, and forty years experience to create.
And that was their first commercial version. Maybe somebody will rip one apart and find out it says "Xeon 2650" on the inside, but until that happens I'm willing to give them the benefit of the doubt because they seem to have at least one fairly remarkable accomplishment under their belts.
If the Google guys buy the upgrade, I'd be willing to bet five bucks that it's real, just very early in the development cycle still.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
In a sci-fi world...
The fact that nobody has been able to measure what the frig is going on with D-Wave's quantum computer is probably a requirement for the thing to work.
The moment we are all allowed to look under the hood, it'll collapse into some useless state and all that hardware will be rendered no more valuable than your average cat.
However, if you read the article (which I did), they are doing real engineering. They are building very sophisticated superconducting quantum circuits. Their second generation machine has four times the qubits and cycles much more quickly. This is very difficult and advanced work, and they are making it happen.
So why is DWave getting so much flack on Slashdot? Somehow I doubt it's because there are vast number of quantum physics types just waiting to display their deep knowledge whenever the subject comes up. What I see are Slashdot Pundits: hoards of pseudotechnical wanna-be's who pile on with meaningless criticism. The motivation is not to have a useful debate but to pretend to be smart by talking trash. Maybe they impress each other, but from my vantage point it looks like a lot of eight year olds shouting curse words they don't understand and giggling over how cool they are.
Why is Snark Required?
"a handful of blokes out of Canada appear to have built a computer that's about as fast as a Xeon "
not really, as theyve leveraged 40++ years of digital/electronic technical development/process/production/knowhow/infrastructure which allows you to assemble the chips you need for a third world produced toy in a matter of weeks
The results so far appear to just be a stepping stone for this new type of technology - a milestone rather than a 'breakthrough'
the d-wave quantum "computer" isn't nearly identical to modern CPUs. it is, in fact, far inferior.
"They were pure niggers." – Noam Chomsky
don't bother trying to discuss this here, you won't get anywhere. read Nielsen and Chuang's "Quantum Computation and Quantum Information" if you can handle it. suffice it to say, non-locality is an accepted part of quantum theory. i know it seems really exotic and impressive to lay-people, but it's just a part of quantum theory. part of what you need to keep in mind is that, roughly speaking, the entanglement is determined at read-time. you can't simultaneously write to a nearby and far-away qubit at the same time. they need to be together at some point, close enough to be entangled by the write.
would you be impressed by a CPU that was a mile across? well, maybe. but would it be a fundamentally different thing than a CPU than was an inch across? not really, though there would be some engineering impracticalities. it's the same with QC; non-locality is just a feature of the system. it's there, we've shown it's there. it doesn't, in principle, matter if the two qubits are an inch apart, or 17 miles apart.
non-locality is interesting, but for QC you just need to accept that it's there.
"They were pure niggers." – Noam Chomsky
The D-Wave machine comes with a software library to solve a specific problem where the vendor claims the machine has much better performance than a conventional PC. They gave the performance numbers of a PC running non-optimized code as a baseline. The thing is experts on the field have created optimized code for solving this same problem which runs faster than the D-Wave machine. On a standard desktop PC. So the D-Wave machine, at least for the problem they claimed they were good at, is many orders of magnitude more expensive than a regular PC with worse performance.
Real quantum computers are supposed to have much better performance than classic computers at solving certain classes of problems but the D-Wave does not seem to have better performance at all.
the reason D-wave not performing as they (used to) claim is that they were hucksters and charlatans all along.
now they've admitted that they're doing adiabatic annealing (a huge step down from what they were coyly hinting at) and, guess what?, it's not really that great at that either. it's not the "talking to classical reality" part of their machine that sucks; it's just not doing very much inside either.
"They were pure niggers." – Noam Chomsky