What is the breakthrough? There is no mention of the problem solved, nor the accuracy of the quantum solution, nor the amount of classical computing resources needed for a non-quantum solution.
The engineering probably has some cool novelty: 39 qubit and 10 million layers is a very large circuit. But simulating a large circuit is very different to achieving a quantum computing breakthrough in CFD.
This press release is uncomfortably close to the border between misleading and fraudulent. It's fairly harmless hype I guess, but claiming a breakthrough involving the world's largest quantum circuit when no quantum circuit has been built is just horrible. I'm not sure I could write this kind of stuff with a good conscience. Rolls-Royce and Nvidia are great companies who are genuinely innovative, it must be awful for the employees who work on the real innovations to read this stuff.
Ok, so I'm not sure what here _really_ benefited from a quantum computer.
I recognize that most folk on HN are concerned about QC applied to exact solutions (which realistically just means breaking crypto), but I'm more interested in QC for analog problem solving.
A real issue you have with digital computers is that they are fundamentally discrete. This means emulating or simulating analog systems (aka the real world) devolves to re-running the same simulation innumerable times and then basically averaging them. The benefit of a QC is that you _should_, in an idealized world, require many fewer runs to get an accurate final representation of the result.
If you go to YouTube you can find many examples of analog computers - often artillery and similar - and I think QC have the potential to offer some real gains in scientific computing by making it possible to create programmable analog computers which haven't really existed in the past.
This press release is simply there to prop up the ISC '23 conference and participants and research projects there. Really don't know if its a good way of doing that.
If a QC could be made to run for 2^20 (ish) longer and to have 2^3 better quality than SOTA then they could do this on it. That would be great because it would run in about 10 seconds, which I assume is much quicker than the run time on GPU's (although the article doesn't make that clear).
Also we have to solve the (not so famous) it-takes-2-days-to-turn-the-f-thing-on problem of QC, and also the the-f-thing-costs-a-fortune issue also.
But apart from this being technologically out of reach for 70 years it's absolutely a great and important result.
I don't think NV sales sagging will happen any time soon. It's more of a case of moving a small amount of product at once and getting their name in the press than that it is a side bet.
Note that if QC does take off NVidia won't be taking any of that cake because they are simulating what it would be like to have such a computer and if they would have one they wouldn't need GPUs to pretend they do. From a technical perspective this if for NVidia not much different from any other HPC sale.
It's weird that people are coming to my company and saying "we want to do a project on QC to do this interesting thing" and then when we say "sorry that's imaginary for 70 years" they go off and do it with someone else anyway. Literally they make a demo that they know won't work for 70 years.
Ok, Ok I could be wrong... it could just be 30 years. But even if it's 5 this is way way way out of the lightcone for these businesses.
This has been happening for three years at least. I feel like I did when people were asking about blockchain.
I look at it from another perspective. Nvidia well have experience with QC programming techniques when the QC hardware actually becomes available. Just as Ada Lovelace is recognized as being the first computer programmer, before the actual computer machines were available.
I am a complete noob w.r.t. Quantum computing, but how exactly does Quantum computing help with Computational Fluid Dynamics? Isn't CFD an issue of efficient parallel computation and scaling of computing resources?
There are certain classes of problems where (theoretical) QCs show an exponential speed-up over classical computers. These generally have to do with certain statistical properties of (theoretical) qbits (mainly related to the stronger-than-classical correlations between measurements on entangled particles, as far as I understand).
Note that I'm only saying theoretical QCs since no actually realized QC has been complex enough to show these speed-ups in a way that convinces everyone (though Google's device has shown a speed-up that is hard to argue with). However, it's important to note that any proof that physically-realized QCs wouldn't actually show these speed-ups would be a monumental discovery in quantum mechanics, as it would represent proof that quantum mechanics is wrong.
Yeah, but is this one of them? The press release is missing what the (theoretical) big-oh would be or any other discussion of what the expected performance gain would be, even in the best case.
Fourier methods are pretty common in PDEs and the QFT is the core ingredient of many quantum computing algos, such as the quantum phase estimation algorithm used to find eigenvectors (such as modes of wave equations) or the HHL algorithm used to solve linear systems of equations (for example linearized PDE). Let's check the literature though,
5 years ago we were on the verge of a working, scalable quantum computer. After 5 years of "breakthroughs" we are now about 10 years away from such a device. And any company who associates themselves with this vapourware crap just looks fraudulent...
Quantum supremacy is a technical term, which they achieved. It is an important milestone but it is not a working useful quantum computer, or anything even remotely close to one.
That's sort of my point: Every day we're making huge progress yet nothing actually changes. We have achieved quantum supremacy, but don't have a practical device yet. We are sick of winning but have not won yet. Every stone is a milestone, but we're 1bn miles from the actual goal so...
I don't think it is. I think if someone says they have a computer, it needs to be a computer, not a very large slide rule. The same applies to "breakthroughs" and "milestones": a non-quantum computer is never a breakthrough in quantum computers the same way a slide rule isn't in computers.
I think we're talking in circles, I just don't get how anyone can take these things seriously...
But quantum supremacy is not a general purpose quantum computer or anything close to one. At best its a proof of concept that you can do something a classical computer cannot. Certainly a neccesary condition, but miles upon miles away from a practical quantum computer.
Maybe the point is that the press/computers really ran with the "quantum supremacy" thing and severely mislead people as to what it actually means. Which i would agree with.
I think you are correct. There was so much money floating about no one bothered checking the value of anything (or the truthfulness of claims) and the key to getting funded was promises and claims not results and prototypes.
There also seems to be a bigger and bigger gap between expectation and reality in society in general that people are just fine ignoring (I'd cite climate change policy or deficits as examples of endless promises and zero delivery).
The engineering probably has some cool novelty: 39 qubit and 10 million layers is a very large circuit. But simulating a large circuit is very different to achieving a quantum computing breakthrough in CFD.