News IBM Unlocks Quantum Utility With its 127-Qubit "Eagle" Quantum Processing Unit

[Admin]

IBM has announced that it has unlocked practical utility out of today's NISQ-era quantum computers - showing how a pairing of quantum and standard computing is required for reliable, consistent, truthful - and thus useful - results.

IBM Unlocks Quantum Utility With its 127-Qubit "Eagle" Quantum Processing Unit : Read more

 

[AndrewJacksonZA]

"'It immediately points out the need for new classical methods,' said Anand."
Am I the only person who paused a bit there for a second? #IYKYK

(Also, Anand's full name and introduction was used after only his surname, which is unusual in articles.)

 

[Deleted member 2731765]

Thanks for the news.

Yes traditionally, useful quantum computation can’t be done without fault tolerance, however, IBM’s paper does provide an important data point that demonstrates current quantum computers can provide value much sooner than expected by using error mitigation ( a total of 2,880 CNOT gates ? ).

Glad they focused on the "quantum Ising model" and tensor network methods, and the Pauli–Lindblad noise model for noise shaping in ZNE. And also shifting from Clifford to non-Clifford gates also helped provide a comparison between the quantum solution and the classical solution. But there are other techniques as well.

Actually, once QEC is attained, building fault-tolerant quantum machines running millions of qubits in a quantum-centric supercomputing environment, wouldn't be a far cry. It's about time we reach the ‘utility-scale' industry threshold.

Unfortunately, most claims about quantum advantage are usually based on either random circuit sampling or Gaussian boson sampling, but they are not considered to be as useful of an application, and there have been no useful applications demonstrating quantum advantage, since QCs are too error prone and too small.

Noise leads to errors, and uncorrected errors limit the number of qubits we can incorporate in circuit, which in turn limits the algorithm's complexity. Clearly error control is important.

We can also agree that realizing the full potential of quantum computers, like running Shor’s algorithm for factoring large numbers into primes, will surely require error correction.

IBM has actually done more error mitigation research than others. It's current roadmap shows a more detailed focus on error mitigation beginning in 2024 and leading to fault tolerance thence afterwards.

By the way, IBM isn’t claiming that any specific calculation tested on the Eagle processor exceeded the abilities of classical computers. Other classical methods may soon return MORE correct answers for the calculation IBM was testing.

 

[bit_user]

Maybe this is an oversimplification, but it sounds to me like they're essentially calibrating their quantum computer. The very next question I have is whether you'd have to do this per-machine, or is simply doing it per-architecture sufficient? And if the former, do you periodically have to re-calibrate?

The answers to these questions seem key to how practical the technique would be.

Also, given that I assume they're using a limited type of computation for this calibration, would you be able to do it in a more cost-effective and energy-efficient way using FPGAs?

 

[TechyIT223]

"'It immediately points out the need for new classical methods,' said Anand."
Am I the only person who paused a bit there for a second? #IYKYK

(Also, Anand's full name and introduction was used after only his surname, which is unusual in articles.)

Is sajat his real name or Anand ? Or maybe Anand is surname?

 

[bit_user]

Is sajat his real name or Anand ? Or maybe Anand is surname?

You guys are thinking of Anand Shimpi? I don't know what he's up to, these days, but he's not a quantum computing researcher. I'm sure those folks basically all have PhDs in particle physics.

 

[TechyIT223]

You guys are thinking of Anand Shimpi? I don't know what he's up to, these days, but he's not a quantum computing researcher. I'm sure those folks basically all have PhDs in particle physics.

Nope, I was not referring to that guy though. Usually most Indians basically swap/shift their username and surname in some cases I have seen.

So maybe ANAND is the first name and sajat his surname? But officially he is using his surname first. But I'm not sure... just a guess

 

[AndrewJacksonZA]

You guys are thinking of Anand Shimpi?

Lol, yeah, that's what popped into my head. And yes, I know. It was a nod to those of us who have been around for a while, especially since Tom's and Anandtech now share a corporate overlord.

 

[shoddyyMic1]

Shimpi is the founder of AnandTech I suppose?

 

[Deleted member 2731765]

Shimpi is the founder of AnandTech I suppose?

Yeah !

 

[TechyIT223]

Lol, yeah, that's what popped into my head. And yes, I know. It was a nod to those of us who have been around for a while, especially since Tom's and Anandtech now share a corporate overlord.

Yeah. TomHW and Anandtech are both related. I think one of them is a sister concern

 

[bit_user]

Yeah. TomHW and Anandtech are both related. I think one of them is a sister concern

They were competitors until a publisher ended up buying both (I forget if that was their current owner, Future, or a previous one). Even after that, their editors probably still viewed them as competitors, though the publisher might've had them try to carve out more distinct niches.

What I find amusing is that the writers can apparently contribute to both. Anton Shilov is the main example, though I've also seen content from Paul Alcorn on Anandtech (while he was also writing for Toms, but maybe before his editor position), and I'm pretty sure there are others.

 

[TechyIT223]

Thank for the info.