IBM Q System One | Hacker News Reader

IBM Q System One

research.ibm.com

73 pointslainon50 comments

Maybe this will do for quantum computing what Watson has done for machine learning.

I cannot tell if this is intended as a very dry joke or not.

Neither can Watson.

It cannot be anything but...

Their Q-Experience FAQs are a much more substantive way to approach what exactly this thing will deliver: https://quantumexperience.ng.bluemix.net/qx/tutorial?section...

Beginner's guide to working with System Q: https://quantumexperience.ng.bluemix.net/qx/tutorial?section...

The number of words here talking about the eye catching design is proof that this is a decorative item that does nothing.

I kinda want to know what's going through the minds of the engineers pictured. If we could just remove all the marketing BS on that page and just have a 1 pager typed up by them I think we'd have a much better understanding of what this product actually is.

Design != Aesthetics.

The design is pretty important for a quantum computer.

> eye-catching design

> To design IBM Q System One, IBM assembled a world-class team of industrial designers, architects, and manufacturers

> ...industrial and interior design studios...

> ...manufacturer of high-end museum display cases...

They are absolutely talking about the aesthetic housing design, not the actual quantum engineering designs. About the closest they get in the Design section is talking about it being airtight and accessible for maintenance.

As far as the housing design: The way it is cylindrical, suspended from above and half lit in a dark room instantly made me think of the Poop Machine at MOMA.

The description talks a lot about commercial uses. Are there problems that the quantum computers of today are actually better at solving than traditional computers? I realize that for many things they're better in a theoretical big-O sense, but does that translate to better wall-clock speed? Presumably you could get a lot of AWS computing for the same cost as one of these.

If you need to factor the number 21, this is a great way to do it.

> Are there problems that the quantum computers of today are actually better at solving than traditional computers?

When the first hard disk (IBM RAMAC) was developed, people drew the same sorts of conclusions. The difference there was that they weren't convinced that the disk would be faster than a secretarial pool. The net was that IBM invented hashing and made it definitively the case that the disk WAS faster.

Hopefully, sixty years later, they are able to address your similar concern in a similar way.

People keep bringing this up. There was at least a theory about what the early computers would be good at that prior methods wouldn't. Then, they hoped they could improve them with research to do more. Even the ENIAC in the 1940's got used as a large calculator for defense applications:

https://www.thoughtco.com/history-of-the-eniac-computer-1991...

Similarly, you could say what value ARPAnet would provide, disks, email, telephone, and so on. Whereas, using this argument for quantum computers (here) or blockchains (another thread) makes no sense given they aren't doing anything better than existing techniques. They're actually a step back on most metrics. Q is both a research tool and a commercial waste of money until it proves to be better than a classical computer at some practical job. An example where quantum tech was theoretically better from the start was Quantum Key Exchange with modifications on the line detected. Even that goal was easier to solve without quantum methods, though.

> People keep bringing this up.

I think the idea that a new technology takes work to provide demonstrable value shouldn't be so surprising. That had to happen with RAMAC and may or may not happen here (or with Blockchain, etc.)

> When the first hard disk (IBM RAMAC) was developed, people drew the same sorts of conclusions. The difference there was that they weren't convinced that the disk would be faster than a secretarial pool. The net was that IBM invented hashing and made it definitively the case that the disk WAS faster.

Source? The impression that I got at the Computer History Museum was that they had specific performance requirements from a customer that they engineered to.

From here, iirc: https://mitpress.mit.edu/books/ibms-early-computers

Don’t have it with me, or I would check for the specific quote.

This has come up previously on HN, and I wrote a few more details back then: https://news.ycombinator.com/item?id=16274812

(Although still not an exact cite. :-( )

This is a nice story, but the problem is that in this case it's definitely not faster.

The point of the story is that people were saying that in 1953 too... Maybe this new machine is nothing, but we'll see.

Does it have enough error-corrected qubits to actually do something useful, or is it a high-tech piece of luxury furniture?

The latter.

More seriously though, there won't be a true error corrected quantum computer for a long time. People are hoping that there can be useful applications with imperfect physical qubits.

At the very least the machines are being used to experimentally verify quantum correction algorithms. See, e.g., https://arxiv.org/abs/1708.02297 and https://arxiv.org/abs/1709.01866 and https://arxiv.org/pdf/1709.00990.pdf

I love that product design. It's not since the Cray 2 and the CM-2 that I see such a striking piece of big iron.

Bummer it doesn't have a built in cushioned bench

it kind of looks to me like someone ran amok in the 8020 catalog. It looks pretty cool, in that prototypish kind of way.

Not to burst your bubble, but that is just the vacuum can on the dilution refrigerator and is a standard piece of hardware you will see in many physics labs. There wasn't really any product design beyond slapping the IBM logo on there.

Of course there was. Deciding what not to add is just as much of a design decision as what to add. I spend no time in physics labs, and like the parent i too find the aesthetic quite striking.

That's a lot of glass for the cleaning crew to wipe down each night. I wonder if it takes special training, or if someone just sticks a "Do not touch" Post-It note to the thing.

Or "Do Not Erase!"

Or "Do not think about it. Tantamount to observation which will interfere with the calculations."

Now I imagined replacing the glass panes with transparent LCDs or LED displays that would give an idea of system status at a glance.

These machines are, like the supercomputers I mentioned, meant to be displayed as a status symbol.

> replacing the glass panes with transparent LCDs or LED displays

I guess the CM-1/2/200 machines are the quintessential example of this. The Cray-2's cooling system isn't an active display, but it gets style points too (IMHO).

(But...I've worked in a few places that used glass windows (or other glass surfaces) as whiteboards... That's really what I was thinking of. :-) )

The Cray-2 had some indicator lights on the top of each lobe. I'm not sure how helpful they were to gain an understanding of what the computer was doing. I can't tell how much work my home cluster is doing by looking at the ethernet switch LEDs.

The LCD would nicely replicate the window-as-whiteboard experience though, at the sacrifice of some readability, because the object behind the glass is so interesting.

Reminds me of this announcement from AWS nine years ago: https://aws.amazon.com/blogs/aws/introducing-qc2-the-quantum...

Did they just put the refrigerator in a glass box and call it "System One"??

Why is this entirely focused on what amounts to case-modding for an alleged supercomputer? Quantum computing advances are presumably not an industrial design challenge...

>Quantum computing advances are presumably not an industrial design challenge...

Perhaps not, but getting laypeople to care enough about quantum computing to fund its development could very well be.

my bias would be towards achieving that goal with a demo of quantum computing :)

> In the second half of 2019, IBM will open the IBM Q Quantum Computation Center, located in Poughkeepsie, New York to expand IBM’s commercial quantum computing program, which already includes systems at the Thomas J. Watson Research Center in Yorktown, New York.

Does this mean, in true IBM fashion, the older, remote one will be terminated?

So is this the first quantum computer available for a company to buy?

How does one interface with it? VT100 terminal?

No, there have been quantum computers sold commercially for many years now. I think D-Wave was the first.

Surely you mean a 3270 terminal...

If they relaunch beam spring keyboards, I'll get one.

You need a terminal with a quantum dot display.

It looks like they are engineering one of those glass enclosed elevators, common in shopping malls, not a computer. Why the focus on outward appearance without explaining its capabilities?

Can someone enlighten me as to why these are always hung from the ceiling?

thermal and mechanical isolation

Good lord is that website slow to load/respond. Too many hero videos.

Maybe it was made by IBM's consulting division.

Reply on news.ycombinator.com