Yet another attempt to benchmark the D-Wave quantum computer – this time, in its 512 qubit configuration – has come up with disappointing results, with the company responding that it was the wrong kind of test. The work, led by Matthias Troyer of ETH Zurich, pitted a D-Wave Two (the machine that Google bought to much fanfare …
"Speaking to New Scientist, D-Wave says what matters isn't speed, but whether it has genuinely built a computer that uses quantum effects."
A Rube Goldberg machine does wonderful things to, that doesn't mean I want to use a Rube Goldberg dishwasher though, or a Rube Goldberg toothbrush, etc.
I wouldn't mind a Rube Goldberg toothbrush
It would probably work better than the one I was using tomorrow.
Re: I wouldn't mind a Rube Goldberg toothbrush
Entangle the bristles with the one I didn't use yesterday - that'll sort it.
So, the cat is alive
it just doesn't run any faster than any of the cats outside the box.
So *what* class of problems does it give a 10x speedup to?
Otherwise it's the classic computer mfg problem.
performance << design spec
Issue upgrades/ workrounds
Obviously the usual suspects of digital slowness are not going to be much use, but I'll presume their tech has a bunch of speed sapping pathologies of its own.
IOW D-Wave better get their collective finger out.
It does work faster
But at the same time, it also doesn't.
And observing it makes it not work the way you expect it to.
This sort of...
reminds me of the dawn of computing, when the machines were arguably slower than trained human calculators. However, the machines needed no sleep or lunch/pub/loo breaks, rendering the discussion of "speed of calculation" pretty much moot, especially as they got faster as tech progressed.
Here we have something similar, with a new tech at the dawn of its practical application, up against established tech in a "performance match". I can't help but note that the new tech already requires the established tech to step up it's game and requires custom built hardware to beat the new tech.
Quantum-effect tech is still in its' toddler years, and from the article it's clear that the chip itself is more or less hampered in reaching its full potential, and needs some work. Yet it can already chug along with some of the best we can put together.
Maybe I'm an optimist, but that's still not bad.
> ... that's still not bad
Maybe, and maybe not. :) As I understand it, the more precisely they pin down the speed of the D-Wave, the less precision there can be about its physical location. A really, really good benchmark could smear the thing out beyond the walls of the lab., no?
The fact that the Planck constant is so small is definite proof that God loves us.
class of problem....
“We're now trying to identify a class of problems for which the current quantum hardware might outperform all known classical solvers,”
The paper is worth looking at - you can get the PDF.
The important question is, does this machine solve this NP problem in polynomial time?
With a much larger machine, it would be easy to distinguish between exponential and polynomial runtimes. Because it is currently small, this is harder. A further confusing factor is that "classical" machines can use simulated annealing to solve this problem in what seems to be better than exponential time most of the time.
So it is possible that the machine has no or limitted quantum entanglement, yet still performs well because it is annealing.
It would be great if this machine really works, but so far the evidence looks marginal, and dwave's agressive PR gives a bad taste
I skimmed(!) the paper, although it is really not my field, I am not sure it is asking the right question.
I went to the D-wave talk at SC'13 and spoke to the presenter.
We define speedup as T(classically)/T(quantum) for problems that can be solved classically , I sort of think we are missing the point unless we realises that T(quantum) is not an equivalent quantity.
For the readers, at SC'13 the D-wave talk focussed on using the D-wave machine as part of a feedback cycle to inform classical calculations.
I am just saying, with so few of these machines in circulation (although they are soliciting ideas , I've not heard back yet...), we might have to wait for the methods that show the real utility.
I agree though, the agressive PR is annoying.
A cryogenist friend of mine commented "It is a research project to get 20mK, not sure about the computing..."
Somebody needs to define a Turing test for quantum computing
Something obvious, like going through a trap-door math function backwards. Something that would take a normal supercomputer years, and repeatedly demonstrate doing it in, say, less than a day.
Re: Somebody needs to define a Turing test for quantum computing
Yes, just what I was thinking - if it cannot solve a classically hard problem of that sort, what use is it?
"the company responding that it was the wrong kind of test."
Quantum Computing, the homeopathic computing solution.
- Product round-up Ten excellent FREE PC apps to brighten your Windows
- Hi-torque tank engines: EXTREME car hacking with The Register
- Review What's MISSING on Amazon Fire Phone... and why it WON'T set the world alight
- Product round-up Trousers down for six of the best affordable Androids
- Why did it take antivirus giants YEARS to drill into super-scary Regin? Symantec responds...