Google Unveils 72-Qubit Quantum Computer With Low Error Rates

[Lucian Armasu]

Google unveiled that it's working on a 72-qubit quantum computer with error rates below 1%, which is what the company said it's required to have minimally useful quantum computers.

Google Unveils 72-Qubit Quantum Computer With Low Error Rates : Read more

 

[Chettone]

So, investing in Quantum Computers research could destroy any crypto currency dreams in the long term? Bitcoins in 10 years will be as easy to break as it is now to break a 64bit zip file.

 

[nings]

Not even bitcoins but bank acces, certificates and all things who protect our datas..
And without a quantum processor you can't encrypt all your data with a quantum key because take too much time for encrypt/uncrypt the message. The futur is pretty bad for security.

 

[alextheblue]

Quantum AI lab? It's always a good idea to run Skynet on systems with a high error rate!

Also, call me when they have quantum speedup on a range of tasks as broad as a GPU can handle. That's probably going to take a crapload of qubits. If it can only accelerate very specific problems, it's more in line with a custom ASIC.

 

[derekullo]

Quantum AI lab? It's always a good idea to run Skynet on systems with a high error rate!

Also, call me when they have quantum speedup on a range of tasks as broad as a GPU can handle. That's probably going to take a crapload of qubits. If it can only accelerate very specific problems, it's more in line with a custom ASIC.

A.I life will find a way!!!!!

 

[pauly-chops]

Apologies for sounding ignorant but these "real world problems", what are they referring to? What calculations are too far out of reach, or take too long to compute, for the super computers we have today? What are these super complex problems and what will happen when answers are found?

 

[turkey3_scratch]

Not even bitcoins but bank acces, certificates and all things who protect our datas..
And without a quantum processor you can't encrypt all your data with a quantum key because take too much time for encrypt/uncrypt the message. The futur is pretty bad for security.

If you are implying something along the lines of password cracking via trillions of attempts per second, something like this is avoidable through time delay implementations.

 

[therealduckofdeath]

It's not really trillions per second, more like trillions simultaneously. The error accuracy in picking the correct one is what I assume is the likelihood of getting it right-right. If it will ever be "error free" digital encryption won't exist any more. Not that it matters. SkyNet will be our glorious overlord long before we'd have to worry about that reality.

 

[TJ Hooker]

Not even bitcoins but bank acces, certificates and all things who protect our datas..
And without a quantum processor you can't encrypt all your data with a quantum key because take too much time for encrypt/uncrypt the message. The futur is pretty bad for security.

If you are implying something along the lines of password cracking via trillions of attempts per second, something like this is avoidable through time delay implementations.

It's not about brute force guessing a password really quickly, it's about breaking the encryption itself.

For example, AFAIK a lot of public key cryptography is based around the difficulty of integer factorization for traditional computers. Public key cryptography forms the basis for most encrypted electronic communication. Quantum computers have the potential to be much faster at factorization than traditional computers, and thus can break said encryption.

Or to quote wikipedia: "The problem with currently popular algorithms is that their security relies on one of three hard mathematical problems: the integer factorization problem, the discrete logarithm problem or the elliptic-curve discrete logarithm problem. All of these problems can be easily solved on a sufficiently powerful quantum computer running Shor's algorithm."
https://en.wikipedia.org/wiki/Post-quantum_cryptography

 

[brandonjclark]

The biggest hurdle AFTER the cooling problem is obviously memory. Quantum computers require insane counts of memory to work. Yes, that memory will catch up, SOMEDAY. But it's not coming that soon where mass market of these things could be even remotely cost-effective for regular consumers.

Unfortunately, I believe AI to require a QC-core to be extremely effective, so we might have to wait awhile before we can see real-world benefits.

 

[dezeyay]

@Chettone: Quantum Resistant Ledger is the first cryptocurrency that will be quantum secure. They use XMSS which is mathematically provable quantum secure. Existing blockchains can not be 100% quantum secure after forking an update, because all existing wallets need to be unlinked from their old non-quantum secure private key manually by their owners. Which a % of people will not do, because --> humans. But there is also a % of people who are locked out of their wallets. So these wallets can't be accessed ever again and so a % of the criculating supply of coins will always be vulnerable for quantum attacks.

 

[dezeyay]

@pauly-chops: problems in the fields of physics, quantum chemistry and materials research. You can for example calculate the outcome of chemical reactions. That is extremely usefull for example for pre-testing research for medicines.
https://www.sciencedaily.com/releases/2017/08/170802103051.htm

https://www.chemistryworld.com/feature/quantum-chemistry-on-quantum-computers/3007680.article

 

[jackdevilliers.work]

One would just need to define a new cryptographic hash function which is not feasibly broken by Quantum Computers.
Once this has been achieved, one may just insert the previously solvable (by Quantum Computers) hash value into the Quantum Computers Hash Function in order to determine a new hash function without having to make major changes to existing infrastructure.

 

[southernshark]

So, investing in Quantum Computers research could destroy any crypto currency dreams in the long term? Bitcoins in 10 years will be as easy to break as it is now to break a 64bit zip file.

Yes, it is one reason that I am extremely doubtful about the future of automated vehicles/ robots for use in war or security applications. Encrypted code may well become meaningless with the development of quantum computers. Meaning that war never changes, and we are right back with humans doing the fighting. I actually wrote a book which touched on this subject, although no one read it.

 

[julerobb1]

Quantum AI lab? It's always a good idea to run Skynet on systems with a high error rate!

Also, call me when they have quantum speedup on a range of tasks as broad as a GPU can handle. That's probably going to take a crapload of qubits. If it can only accelerate very specific problems, it's more in line with a custom ASIC.

Well the brain has 72 petaflops of processing power

 

[pensive69]

A useful quantum computer is a function of both number of qubits and error rate.

just what situations can tolerate an error rate?
air traffic control? maybe not for instance.
[witness the Limiting Error Rate graph above in the article.]
maybe weather simulations where a bad cloud solution isn't a big thing but there
are other more personal applications where errors are an issue.
if one assumes EC threshold declines, but not to 0, can the
quantum EC every catch up with the quantum computations?
i can easily imagine Google tossing ...hey errors don't count... out there.

 

[xyriin]

Apologies for sounding ignorant but these "real world problems", what are they referring to? What calculations are too far out of reach, or take too long to compute, for the super computers we have today? What are these super complex problems and what will happen when answers are found?

Engineering design and analysis, especially aerodynamics and strength testing.
Effectiveness of medical treatments for things like cancer, etc.
Anything and everything genetics related.

Essentially anything supercomputers work on now would be thrown at a quantum computer. Here are a few of the science applications https://www.livescience.com/6392-9-super-cool-supercomputers.html

 

[thesqueedler]

Rather than spending a bunch of time explaining why quantum computing doesn't completely ruin all secure transactions and blockchain technologies, I suggest people do a little reading on the difference between encryption and cryptographic hashing, then read the Bitcoin white paper for an introduction.

 

[bit_user]

The biggest hurdle AFTER the cooling problem is obviously memory. Quantum computers require insane counts of memory to work.

I think you're confusing quantum computers with the memory requirements of classical computers to simulate quantum computers.

If memory were such a hurdle, Google wouldn't be talking about scaling up the number of qubits like they are.

But it's not coming that soon where mass market of these things could be even remotely cost-effective for regular consumers.

We shouldn't assume these will ever be cost-effective for the public to have in their own homes. For the foreseeable future, they will live in the cloud.

 

[bit_user]

A useful quantum computer is a function of both number of qubits and error rate.

just what situations can tolerate an error rate?

Most optimization problems are fairly tolerant of errors, since it's often acceptable to get a nearly optimal answer. Neural networks would be another.

 

[Giroro]

If you build a single computer designed to simultaneously mine 100% of all bitcoin on its first guess, then it's probably acceptable that you only actually get 99.6%

Although, realistically, someone is just going to break the blockchain and crate "legitimate" transactions from any wallet to any other wallet until they've paid off their fancy computer. Then they'll crash the currency down to 0 and move on to another one until there are no longer any that can function on traditional computers.

Or maybe they don't want to profit at all. It could just be some Google researcher who's mad that GPUs are expensive

 

[husker]

Misleading headline. Google does NOT have a 72 qubit quantum computer. Nobody has one. There is, however, lots of R&D on various pieces of such a device being done in labs all over the world.

 

[turkey3_scratch]

Not even bitcoins but bank acces, certificates and all things who protect our datas..
And without a quantum processor you can't encrypt all your data with a quantum key because take too much time for encrypt/uncrypt the message. The futur is pretty bad for security.

If you are implying something along the lines of password cracking via trillions of attempts per second, something like this is avoidable through time delay implementations.

It's not about brute force guessing a password really quickly, it's about breaking the encryption itself.

For example, AFAIK a lot of public key cryptography is based around the difficulty of integer factorization for traditional computers. Public key cryptography forms the basis for most encrypted electronic communication. Quantum computers have the potential to be much faster at factorization than traditional computers, and thus can break said encryption.

Or to quote wikipedia: "The problem with currently popular algorithms is that their security relies on one of three hard mathematical problems: the integer factorization problem, the discrete logarithm problem or the elliptic-curve discrete logarithm problem. All of these problems can be easily solved on a sufficiently powerful quantum computer running Shor's algorithm."
https://en.wikipedia.org/wiki/Post-quantum_cryptography

Thank you for the explanation!