Google gets an error-corrected quantum bit to be stable for an hour
Using almost the entire chip for a logical qubit provides long-term stability.
About how far does this leave us from a usable quantum processor? How far from all current cryptographic algorithms being junk?
The latest versions of TLS already have support post-quantum crypto, so no, it's not all of them. For the ones that are vulnerable, we're way, way far off from that. It may not even be possible to have enough qbits to break those at all.
Things like simulating medicines, folding proteins, and logistics are much closer, very useful, and more likely to be practical in the medium term.
At least a week, probably more
Algorithms will be easier and faster to fix than the process of getting this breakthrough to viability
Maybe they can use the same techniques for keeping their product management and feature roadmap for more than an hour.
Just in time for the fall of American democracy. What could possibly go wrong.
Seeing quantum computers work will be like seeing mathemagics at work, doing it all behind the scenes. Physically (for the small ones) it looks the same, but abstractly it can perform all kinds of deep mathematics.
108 qubits, but error correction duty for some of them?
What size RSA key can it factor "instantly"?
Currently none, I think it's allegedly 2000 qbits to break RSA
afaik, without a need for error correction a quantum computer with 256 bits could break an old 256 bit RSA key. RSA keys are made by taking 2 (x-1 bit) primes and multiplying them together. It is relatively simple algorithms to factor numbers that size on both classsical and quantum computers, However, the larger the number/bits, the more billions of billions of years it takes a classical computer to factor it. The limit for a quantum computer is how many "practical qubits" it has. OP's article did not answer this, and so far no quantum computer has been able to solve factoring a number any faster than your phone can in under a half second.
Which hour ? If they create real quantum computer they can start identifying person that creates reality for all of us, assuming reality is broadcasted by collective mind, I doubt they can do it right now and I am sure the moment they start that person will log out from internet. Good bye then.
google can walk up the passageway of elton john for all i care!
I know a good therapist, if need be!
There's a dime stuck in the road behind our local store, tails side up, for over 15 years. And that doesn't even need error correction.
Why does it sound like technology is going backwards more and more each day?
Someone please explain to me how anything implementing error correction is even useful if it only lasts about an hour?
As stable as that dime is, it's utterly useless for all practical purposes.
What Google is talking about it making a stable qbit - the basic unit of a quantum computer. It's extremely difficult to make a qbit stable - and as it underpins how a quantum computer would work instability introduces noise and errors into the calculations a quantum computer would make.
Stabilising a qbit in the way Google's researchers have done shows that in principle if you scale up a quantum computer it will get more stable and accurate. It's been a major aim in the development of quantum computing for some time.
Current quantum computers are small and error prone. The researchers have added another stepping stone on the way to useful quantum computers in the real world.
Do you have any idea the amount of error correction needed to get a regular desktop computer to do its thing? Between the peripheral bus and the CPU, inside your RAM if you have ECC, between the USB host controller and your printer, between your network card and your network switch/router, and so on and so forth. It's amazing that something as complex and using such fast signalling as a modern PC does can function at all. At the frequencies that are being used to transfer data around the system, the copper traces behave more like radio frequency waveguides than they do wires. They are just "suggestions" for the signals to follow. So there's tons of crosstalk/bleed over and external interference that must be taken into account.
Basically, if you want to send high speed signals more than a couple centimeters and have them arrive in a way that makes sense to the receiving entity, you're going to need error correction. Having "error correction" doesn't mean something is bad. We use it all the time. CRC, checksums, parity bits, and many other techniques exist to detect and correct for errors in data.
I'm well aware. I'm also aware that the various levels of error correction in a typical computer manage to retain the data integrity potentially for years or even decades.
Google bragging about an hour, regardless of it being a different type of computer, just sounds pathetic, especially given all the money being invested in the technology.
It can be useful if they build enough of these that they can run programs that regular computers can't run at this scale, in less than an hour.
Quantum computers aren't a replacement for regular computers because they're much slower and can't do normal calculations, but they can do the type of problem where you have to guess-and-check too many answers to be feasible with regular computers in many fewer steps.
Because quantum physics. A qubit isn't 0 or 1, it's both and everything in between. You get a result as a distribution, not as distinct values.
Qubits are represented as (for example) quantumly entangled electron spins. And due to the nature of quantum physics, they are not very stable, and you cannot measure a value without influencing it.
Granted, my knowledge of quantum computing is very hand-wavy.
I do get that, yes it's more complicated than I can fully wrap my brain around as well. But it also starts to beg the question, how many billions of dollars does it take to reinvent the abacus?
Again, I realize there's a bit of a stark difference between the technologies, but when does the pursuit of over-complicated technology stop being worth it?
Shit, look at how much energy these AI datacenters consume, enough to power a city or more. Look at how much money is getting pumped into these projects..
Ask the AI how to deal with the energy crisis, I'll only believe it's actually intelligent when it answers "Shut me and all the other AI datacenters off, and recycle our parts for actual useful purposes."
Blowing billions on quantum computing ain't helping feed, clothe and house the homeless...
it only lasts for an hour
"Only"? The "industry standard" is less than a millisecond.
Show the academic world how many computational tasks the physical structure of that coin has solved in the 15 years.
You're forgetting the difference between processing and memory. The posted article is about memory.
If the memory sucks compared to standards of half a century ago, then they just suck.
"Let's use deuterium (stable) instead of tritium (t½ = 12.3y) so our nukes don't expire in a few years."
"Will it work?"
"No but it will be stable."
"I can't believe you haven't been fired earlier"
