Google says it’s achieved it: Google says it's achieved quantum supremacy
Nope. But if it can make Notepad run faster I’m all for it.
Which is good, but do quantum computers grow at the same rate as classical computers? Will we go to 500-1000 qubit computers by 2030 or so that have lower error rates and more coherence?
No idea. I’m not sure what quantum computers will achieve. I keep hearing they will revolutionize fields like AI, chemistry, etc but I don’t know the details. Also I’ve heard some people imply we need millions of qubits to really make them work. Then again I’ve heard other people say 50 qubits will accomplish a lot. I have no idea.
Yes, I just read it in yesterday’s Science Times. Google claims to have a solved in a few minutes a problem that would take an ordinary computer 10,000 years to solve. The claim has been disputed.
Nonetheless, it seems like we’ve gone from 10 qubits to ~50 in just the last five years or so. So thats a good level of progress, we’ve gone from quantum computers that can barely do anything to quantum computers that can keep up with and surpass supercomputers in a few years.
Also the benchmark wasn’t against an ordinary computer doing something in 10,000 years, it was against the worlds fastest supercomputer doing something in 10,000 years.
At least thats how it seems to me.
Even if they work, quantum computers are good for very specific sets of problems. I heard someone the other day say that if they take off, AI is as good as solved.
Not even close.
I worked in finding manufacturing defects in microprocessors, which we do quite well. I don’t even want to think about the problem of finding manufacturing defects in quantum computers, and I’m too old to have to worry about it.
There were several big pushes for asynchronous logic, all of which died, and at least one because no one could figure out how to test it.
I recall that Intel made a prototype asynchronous Pentium back in the day, which was something like 3x as power efficient as the synchronous version. Never shipped, obviously. Any ASIC engineer would sell their soul for a 10% improvement, let alone 200%, so what went wrong? Turns out that if you can’t design, test, or manufacture your product, it doesn’t matter how good it is. As you say, it seems like there’s a good chance the same will be true of QCs except for highly specialized problems.
In fact, I would personally put odds that truly fast, general QCs (say, one that can factor the product of two million-digit primes) are not actually possible, but we’ll see about that.
The quantum computer is still over a thousand times faster at that speed.
If I understand right, we’re using “quantum supremacy” to mean a quantum computer can solve a problem that a classical computer cannot–at least in practice.
I’m happy to call a problem that takes 10,000 years to solve one that cannot be solved in practice. I don’t say that about one that take 2-3 days even if that’s long compared to the quantum computer.
I, for one or zero, welcome our Quantum Overlords.
Huh. Never pegged you for a Quantum Supremist.
Interesting related article. (Sadly, "Prototype Asynchronous Pentium doesn’t really work as a band name or as a porn name.)