3 ms·

Literally none of this is correct. > - Google’s Willow quantum chip significantly outpaces current supercomputers, solving tasks in minutes that would otherwis

Literally none of this is correct. > - Google’s Willow quantum chip significantly outpaces current supercomputers, solving tasks in minutes that would otherwise take billions of years. billions of years you say? Just what kinds of "computing tasks" we talkin about here?

r33b33 2y ago
Okay, explain why and how it is not correct then? I genuinely want to learn and figure out what is the truth and what is the best route of action when it comes to securing a portfolio.
- simonh 2y ago
  Willow only implements one single error corrected logical Qubit, using over 100 physical qubits. So far the test they have done is only applicable to an extremely narrow problem domain, approaching the limit of how practically useless a problem domain can be. I may be exaggerating slightly. As Google's own team report "Google will only consider itself to have created a “true” fault-tolerant qubit, once it can do fault-tolerant two-qubit gates with an error of ~10-6". That's two logical qubits. The general consensus is that to have a practically useful quantum computer we'd need one with about a million physical Qubits. That's not just 1,000 of these 105 qubit chips because those wouldn't be entangled, but one chip with about a million entangled physical qubits and therefore 1,000 logical qubits.
smallerize 2y ago
The benchmark is called "random circuit sampling". https://blog.google/technology/research/google-willow-quantum-chip/ https://blog.google/technology/research/google-willow-quantu... I call it a benchmark because a random circuit is probably not useful, but quantum computers are useful for simulating quantum systems including materials science, particle physics etc. This result is separate from the error-corrected qubit result and did not use error correction. There is a third result in the paper, which is a repetition-coded qubit that was stable for almost an hour. https://www.nature.com/articles/s41586-024-08449-y https://www.nature.com/articles/s41586-024-08449-y

r33b33 2y ago

Okay, explain why and how it is not correct then? I genuinely want to learn and figure out what is the truth and what is the best route of action when it comes to securing a portfolio.

simonh 2y ago
Willow only implements one single error corrected logical Qubit, using over 100 physical qubits. So far the test they have done is only applicable to an extremely narrow problem domain, approaching the limit of how practically useless a problem domain can be. I may be exaggerating slightly. As Google's own team report "Google will only consider itself to have created a “true” fault-tolerant qubit, once it can do fault-tolerant two-qubit gates with an error of ~10-6". That's two logical qubits. The general consensus is that to have a practically useful quantum computer we'd need one with about a million physical Qubits. That's not just 1,000 of these 105 qubit chips because those wouldn't be entangled, but one chip with about a million entangled physical qubits and therefore 1,000 logical qubits.

smallerize 2y ago

The benchmark is called "random circuit sampling". https://blog.google/technology/research/google-willow-quantum-chip/ https://blog.google/technology/research/google-willow-quantu... I call it a benchmark because a random circuit is probably not useful, but quantum computers are useful for simulating quantum systems including materials science, particle physics etc. This result is separate from the error-corrected qubit result and did not use error correction. There is a third result in the paper, which is a repetition-coded qubit that was stable for almost an hour. https://www.nature.com/articles/s41586-024-08449-y https://www.nature.com/articles/s41586-024-08449-y