Study Doubts Quantum Computer Speed 105
Alain Williams writes "The BBC reports that a new academic study has raised doubts about the performance of a commercial quantum computer in certain circumstances. In some tests devised by a team of researchers, the commercial quantum computer has performed no faster than a standard desktop machine. 'The study has been submitted to a journal, but has not yet completed the peer review process to verify the findings. And D-Wave told BBC News the tests set by the scientists were not the kinds of problems where quantum computers offered any advantage over classical types.'"
But it's QUANTUM! (Score:1)
Re:But it's QUANTUM! (Score:5, Funny)
But it's QUANTUM!
So's my dishwasher powder.
Re: But it's QUANTUM! (Score:1)
peer review does not verify findings.
Re: But it's QUANTUM! (Score:5, Interesting)
Reviewers often are working in similar spaces too. I had a paper that failed review the first couple times because the reviewer wanted more data on a particular area of the project. It didn't affect the main idea of the paper or in any way directly contribute to our argument. The reviewer needed some charts generated because he was working on a similar project and it would help for him to have some other paper that he could reference to get started/justify his paper. So not only low paid work sometimes low paid work for someone who you didn't even know was your boss :)
Re:But it's QUANTUM! (Score:5, Funny)
Just like digital improves the quality of everything.
Except music, if you're an audiophile who prefers vinyl.
I don't care one way or the other about the audio, but I'm a true hipster videophile, and I insist on watching everything on VHS. It's hard to describe, but VHS gives a warmer, softer, smoother picture without all those annoying dots distracting you from the filmmaker's true vision.
Messages Missed (Score:3, Insightful)
do not know if you measure it (Score:3, Funny)
it may be faster , slower or both
Re: (Score:1)
No, it's all of those at the same time.
Re: (Score:3)
Moot point: once you measure it, you've changed the results. Obviously, all measurements are invalid.
In other news... (Score:2, Insightful)
Re: (Score:2)
Prototype engine fails to win Formula One race.
D-Wave machine is hardly a prototype.
Re: (Score:3)
Re: In other news... (Score:2)
Re: (Score:2)
This is obviously a case of the observer influencing the outcome of the test.
Adjusts "Hiesenburg Compensators" hows that :)
Re:In other news... (Score:4, Informative)
It also hasn't been established that the D-Wave is at all a quantum computer. They've refused to say how it works in that regard, and there has been no proof that any quantum entanglements even take place inside the box.
Re:In other news... (Score:5, Insightful)
In more accurate news, Formula One car proves sucky at handling the monthly grocery shop...
Re: (Score:2)
Exactly what I heard. Especially with how badly the headline and first line contradict.
Study Doubts Quantum Computer Speed
raised doubts about the performance of a commercial quantum computer in certain circumstances
The headline makes it seem like someone is doubting the feasibility of quantum computing. Instead, someone is simply questioning it's practical applicability in "certain" circumstances. I already doubt the feasibility of my personal computer in "certain" circumstances. Like on the john... t
Re: (Score:3, Interesting)
you shitting me?
"prototype" engines won f1 all the fucking time when f1 was still cool as fuck and the cars were pretty much all prototypes.
d-wave on the other hand is a company taking millions for their product - which is a black box they say is good at doing something but refuse to tell why or how - and loads of government money is pouring into it.
try calling mb or whoever is actually building f1 engines still and try to buy their current race engine.
now try calling d-wave and try to buy whatever it is th
Re: (Score:2)
"prototype" engines won f1 all the fucking time when f1 was still cool as fuck and the cars were pretty much all prototypes.
They still are prototypes.
Re: (Score:3)
There is a difference between “prototype”, which implies a experiential design that might be refined and moved into production someday and “custom, bespoke”, where one takes true and tried engineering principles and push them to the extremes.
F1 today tends to fall into the latter category today. For example, the bodies use a lot of carbon fiber. The reason we don’t see a lot of carbon fiber into today’s production car is not because of engineering concerns (we know it wor
Re: (Score:2)
Re: (Score:2)
More like prototype racing engine beaten by Mom's old minivan.
Of course... (Score:2, Flamebait)
Maybe you could take a moment to enlighten us so that we can put to bed the debate as to whether your product actually is a quantum computer.
Re: (Score:1)
Well if you test a microscope's ability to crack nuts, you may find that it is out-performed by legacy hammers in certain cases, too.
Re: Of course... (Score:1)
Having dropped such an instrument on my nuts, I can confirm it is at least as efficient as a hammer.
Re: (Score:2)
OTOH, if you test a microscope at a wide variety of tasks related to magnification and the best it ever manages is just slightly better than the $10 readers you bought at the drug store, you might begin questioning if it even is a microscope.
Re:Of course... (Score:4, Insightful)
Yeah, kinda what I was thinking.
Either tell us some of the things this can do faster, or we're going to have to assume this is smoke and mirrors.
If it is a real thing, there must be specific types of problems which can be identified where it is faster, and where that can be demonstrated as being significantly faster.
But if there aren't any of those, one does need to question if their claims are true.
Re: (Score:2)
But it's a quantum computer; it's both faster and slower at the same time!
Re:Of course... (Score:5, Funny)
Maybe it's quantum marketing, and is therefore both true and false at the same time.
Which would make it indistinguishable from all other marketing.
Re: (Score:2)
Quantum superpositions of true and false have the possibility of collapsing to the "true" state when closely observed --- not generally the case with "all other marketing," which tends to follow a classical hidden variable approach (i.e. they were lying all the time, you just didn't know where to look).
Re: (Score:2)
Marketing superpositions of true and false have the possibility of collapsing to the "true" state when closely observed as well.
Mostly that's by accident, or through the subtle inclusion of a single objective fact.
Re:Of course... (Score:5, Interesting)
DWave don't have to "enlighten us": the statistical tests that distinguish quantum and classical annealing are in the public domain and they've been open about which of those tests they think the machine should pass. The trouble is that it's hard to run those tests cleanly, which is what the study is about:
Here we show how to define and measure quantum speedup in various scenarios, and how to avoid pitfalls that might mask or fake quantum speedup.
Re: (Score:2)
Not news (Score:1)
the tests set by the scientists were not the kinds of problems where quantum computers offered any advantage over classical types.
So... why is this here?
Worthless BBC article (Score:4, Insightful)
Since I haven't read the actual paper, I'll give the researchers the benefit of the doubt. But the BBC reporting is terrible. What I got from the story is that a study has demonstrated that this Quantum computer isn't better at everything. Well, duh! Everyone who has even very casually followed Quantum computing knows that they are a new class of computing which can solve a limited set of problems very quickly. I'm really not much wiser after reading this story.
Re:Worthless BBC article (Score:4, Informative)
Since I haven't read the actual paper, I'll give the researchers the benefit of the doubt. But the BBC reporting is terrible. What I got from the story is that a study has demonstrated that this Quantum computer isn't better at everything. Well, duh! Everyone who has even very casually followed Quantum computing knows that they are a new class of computing which can solve a limited set of problems very quickly. I'm really not much wiser after reading this story.
What I got from it is that quantum computing researchers devised some tests for it and that it performed about as well as a desktop computer. I would *imagine* that quantum computing researchers at NASA and Google wouldn't just throw an unsuitable set of tests at it. I *imagine* that they know as much about the D-Wave computer as anyone outside D-Wave know about it and devised tests to, you know, *test* it.
I could be wrong, maybe Google and NASA quantum computing researchers know shit about quantum computing and threw totally unsuitable tests at it.
Re: Worthless BBC article (Score:2)
Re: (Score:1)
Maybe you know shit about reading comprehension. Google and NASA are DW2 customers; the researchers in question are at ETH Zurich.
Re: (Score:2)
Quote from the abstract:
The study asks a very specific question and acknowledge its limited scope.
Re: (Score:2, Informative)
I'll agree that
but it's not 'worthless'...you have to hack through it b/c they aim for the dumbest, most non-tech reader they can imagine...then dumb it down more...but the info is there
you don't need to just "give the researchers the benefit of the doubt"...that's foolish
"Quantum" computing is hype & the research saying it exists is flawed.
This is big news for some people. For me, I split time between the biz world & academia so this matters to me for many reasons.
One e
Re: (Score:2)
Re: (Score:1)
Comment removed (Score:5, Informative)
Re: (Score:2)
Compare it instead to a traditional annealer.
Correct me if I misunderstood, but doesn't calculation effectively produce solutions at least on a similar level of precission as annealing?
Re:this story never seems to be correct. (Score:5, Informative)
D-Wave has yet to demonstrate, in the open literature, that their quantum annealing is faster than classical computing annealing methods using considerably cheaper hardware. Early "look how fast we are" comparisons involved comparing against really terrible algorithms on classical hardware --- independent researchers were able to beat D-Wave when not using intentionally crippled approaches.
Re: (Score:1)
Exactly. It's a classic story in R&D:
- Invent a promising new technology based on some groundbreaking principle.
- Announce commercial availability 'real soon now'.
- Initial implementations turn out to be equal or worse to proven implementations based on the 'classic' technology.
-> Now the company has to dial up the PR to 11 in order to stay afloat while the boys in the lab create version 2.0 that actually delivers..
Same thing happens with drug discovery all the time: A brand new drug, targeting some
Re: (Score:2)
They are. That's the problem, the computation using the dwave for annealing isn't going any faster than the one using the conventional approach on a desktop PC.
Dwave snake oil? (Score:2)
I had to chuckle at the thought that the D Wave is nothing more than a desktop hidden inside a large case full of magic quantum gizmos. Kinda like those free energy machines with hidden motors and batteries inside them. The article even reads like a free energy hoax:
"The comparison found no evidence D-Wave's $15m (£9.1m) computer was exploiting quantum mechanics to calculate faster than a regular machine."
"And D-Wave told BBC News the tests set by the scientists were not the kinds of problems wh
Re: (Score:2)
I had to chuckle at the thought that the D Wave is nothing more than a desktop hidden inside a large case full of magic quantum gizmos.
I hear the D-Wave folks also have been developing a new Retro Encabulator [youtube.com].
Re: (Score:2)
Actually it is the claim that it performs useful quantum computation that I take with a grain of salt.
"has not yet completed the peer review process" (Score:2)
Re: (Score:1)
There are a set of problems that a linear. They depend upon the results before them. No amount of parallel computing will change that problem. You can make *new* algs that will do that task in parallel sometimes. Parallel is where quantum computing really shined. As you could in theory have a beaker of goo and it 'does stuff'. The reality is much more like a standard computer laid out on a board and waiting for memory to show up.
This is not really news to anyone who put some thought into it. It has t
Because they were observing (Score:1)
The commercial quantum computer has performed no faster than a standard desktop machine because the benchmark test was being observed by a team of researchers. Had they not have looked at it, the results would have been different.
Re: Because they were observing (Score:2)
Well (Score:2)
And D-Wave told BBC News the tests set by the scientists were not the kinds of problems where quantum computers offered any advantage over classical types
From the abstract:
We illustrate our discussion with data from a randomized benchmark test on a D-Wave Two device with up to 503 qubits.
What is a randomized benchmark test? What is randomized? The algorithm itself? Then I guess that's not a good test.
And was D-Wave not involved in the study?
Re: (Score:2)
It's obviously referring to randomly chosen problem instances, which are then approximately-solved by both the D-wave device and a state-of-the-art classical algorithm. Don't be a fool.
And why should D-wave be involved in the study? Is Intel involved in every benchmark test of x86 hardware? That would be very suspicious indeed.
Re: (Score:1)
idiotic (Score:2)
My cell phone isn't very good at editing 1080p videos or hosting Exchange 2013 and my toaster isn't good at cooking eggs either. The headline should read "quantum computer not good at calculations they weren't designed to do." What a useless story.
Not news (Score:2)
It's long been known that quantum computing offers speed-ups only on certain problems. Thing is, many of those problems are of immense practical import (like factoring large numbers).
Re: (Score:2)
And thing is that this D-wave gizmo seems to be useless for these important problems.
I assure you that if they would show factorization of very big numbers in instant time, nobody would doubt them, but instead line up with checkbooks.
So far, as far as I'm aware, they have failed to show ANY kind of important problem which can be solved by their machine in time orders of magnitude faster than comparable size hardware (not to mention, comparable _cost_ hardware).
Re: (Score:3)
Is factoring a prime number such an important problem?
I do that in constant time, actually. Or log(N) if I have to print the result.
Re: (Score:1)
It's long been known that quantum computing offers speed-ups only on certain problems. Thing is, many of those problems are of immense practical import (like factoring large numbers).
Factoring large numbers would get a lot less important if it weren't computationally wasteful to do it. The main practical application of them is cryptography where the whole point is having a problem who's solution can be verified quickly but takes preposterous amounts of time to compute.
If it became trivial to factor large numbers no one would use crypto schemes based on factoring large numbers and the technology would have rendered itself obsolete. (unless of coarse it can also do something else that's m
What I hate with quantum computers (Score:2)
As soon as you know how fast it is, you have no Idea where it is and lose all the gained time searching for it.
D-Wave machine Quantum computer (Score:4, Informative)
Re: (Score:1)
> under the principles of an adiabatic quantum computer
That's not true -- most of the nerds involved have type II diabetes.
Re: (Score:1)
Re: (Score:2)
Is the thing actually Turing complete (in the practical sense)?
Re: (Score:2)
Yes, we should. But we don't have reporters anymore, we have polemicists. Apologies to the few scientific reporters that still have honor.
New Resource Testing (Score:2)
People need to realize that this is not the government waste/quantum computing expose it is made out to be in this article. Whenever the supercomputing community comes out with a new resource, we test it and find the best algorithms for that resource. We have a long history of different algorithms working better for different resources. Take for instance, the transition from the Cray vector processors, to commodity Intel processors, then back to vector processors with the Earth Simulator, then back to In
Peer Review (Score:2)
Why put trust into peer review when we're doing such a poor job of it now?
http://www.economist.com/news/leaders/21588069-scientific-research-has-changed-world-now-it-needs-change-itself-how-science-goes-wrong [economist.com]
Have I misunderstood something (Score:2)
As I understand it, it's fairly common knowledge that a quantum computer wouldn't be that much faster than a normal computer in most cases. As I've heard it, there's only a few applications where a quantum computer would be significantly faster, the main one being the math required to crack RSA and other asymmetric encryption algorithms.
It could probably mine bitcoins stupidly fast too.
My independent study. (Score:2)
They could have saved themselves the trouble and just read the Wikipedia article. This has been known since at least 2007, according to one of the quotes there.
It's junk. (Score:1)
Re: (Score:2)
Just to add a little to what you said. An old GPU/ASIC can perform certain tasks far faster than modern CPUs. If D-Wave ends up only able to solve that kind of problem faster than a traditional computer then it must also be able to outperform the GPU/ASIC setup as well.
no fair!! (Score:1)
I call bullshit!!
they obviously altered the performance of the computers by measuring their performance.
There are obviously two ways to look at this (Score:5, Informative)
The Google Quantum AI lab puts this news into perspective [google.com] and I put my positive spin on it here [wavewatching.net].
Having talked with one of the co-authors of the paper, he actually came away impressed at how far D-Wave has come in ten years. Although not yet far enough that I'd win my bet with him, that the D-Wave two could beat classical computing across the board.
So in short, yes, the BBC's reporting on quantum computing is atrocious. Not the first time either. [wavewatching.net]