Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
Supercomputing China Hardware

China's New Quantum Computer Has 1 Million Times the Power of Google's (interestingengineering.com) 143

Physicists in China claim they've constructed two quantum computers with performance speeds that outrival competitors in the U.S., debuting a superconducting machine, in addition to an even speedier one that uses light photons to obtain unprecedented results, according to a recent study published in the peer-reviewed journals Physical Review Letters and Science Bulletin. Interesting Engineering reports: The supercomputer, called Jiuzhang 2, can calculate in a single millisecond a task that the fastest conventional computer in the world would take a mind-numbing 30 trillion years to do. The breakthrough was revealed during an interview with the research team, which was broadcast on China's state-owned CCTV on Tuesday, which could make the news suspect. But with two peer-reviewed papers, it's important to take this seriously. Pan Jianwei, lead researcher of the studies, said that Zuchongzhi 2, which is a 66-qubit programmable superconducting quantum computer is an incredible 10 million times faster than Google's 55-qubit Sycamore, making China's new machine the fastest in the world, and the first to beat Google's in two years.

The Zuchongzhi 2 is an improved version of a previous machine, completed three months ago. The Jiuzhang 2, a different quantum computer that runs on light, has fewer applications but can run at blinding speeds of 100 sextillion times faster than the biggest conventional computers of today. In case you missed it, that's a one with 23 zeroes behind it. But while the features of these new machines hint at a computing revolution, they won't hit the marketplace anytime soon. As things stand, the two machines can only operate in pristine environments, and only for hyper-specific tasks. And even with special care, they still make lots of errors. "In the next step we hope to achieve quantum error correction with four to five years of hard work," said Professor Pan of the University of Science and Technology of China, in Hefei, which is in the southeastern province of Anhui.

This discussion has been archived. No new comments can be posted.

China's New Quantum Computer Has 1 Million Times the Power of Google's

Comments Filter:
  • by boner ( 27505 ) on Thursday October 28, 2021 @06:21AM (#61934779)

    Hyperspecific tasks hint at how they prepare the quantum computer for the calculation. Pristine environment hints at all the effort they are making to keep the quantum computer stable enough to complete that task.

    How long will it take before we have enough error correction stability that we can load a task on demand into a quantum computer and get a result before that error correction is overwhelmed? It already takes time today to load a stating state into a computer for most calculations, how is a quantum computer different?

    • It depends on what they want to do. If they want their fusion reactor to run in a stable fashion, they don't need to be able to load initial states on-the-fly, they only need to be able to upload cumulative data on-the-fly. A lot of it, very fast. If their quantum computer is up to that task, then they have a good chance of beating America and Europe to controlled fusion no matter what other limitations exist.

      (It's one of the very few problems where you need this kind of performance and where the setup that is being controlled takes so long to change that the QC can easily be updated along with it.)

      • Mod up. Technology without an application is a solution looking for a problem. Assuming the performance claimed is legitimate and this technology can in fact assist in developing stable fusion technology, then China may achieve what it's posturing itself as - the dominant global power of the 21st century
      • How do you come to the silly idea that you need a quantum computer to have "controlled fusion"?

        The hard problem of fusion is: fusion.
        Not calculating or gathering some data.

        • The hard problem in fusion is to avoid instabiities. For which you need horrific compute power. Fusion itself is easy, that's been handled for ages. Getting fusion to not cut out due to things like pinching is a major problem.

          My source? The fusion researchers. I do this, umm, thing called READ. It comes highly recommended.

          • And a Quantum computer (close to a fusion reactor, what a laugh) won't change that..

            I doubt you need more processing power then a 1980 machine to control a magnetic field.

            • I doubt you need more processing power then a 1980 machine to control a magnetic field.

              You really are a nut, aren't you?
              The magnetic fields in question are produced by some of
              the most complex systems ever imagined.

              • The magnetic fields in question are produced by some of
                the most complex systems ever imagined

                Nope.
                They are simply static - as in: never changing - magnetic fields in a super conductor.
                There is neither a computer required nor anything else to "manage them". You switch them on. And thats it.

            • by jd ( 1658 )

              Except that modern supercomputers are known not to be powerful enough, which either means the 1980s were more astonishing than I recall, or you really don't understand how much data needs to be processed in how short a time, with the correct responses given to millions - if not billions - of different control devices.

              Ok, how close are we? 101 seconds of sustained fusion has been achieved. 10 minutes will be achieved with the next generation of reactor.
              https://nation.com.pk/29-May-2... [nation.com.pk]
              https://www.euro-fusion [euro-fusion.org]

              • recall, or you really don't understand how much data needs to be processed in how short a time, with the correct responses given to millions - if not billions - of different control devices.
                A fusion reactor has not millions/billions of control devices for the magnetic field.

                It has exactly *one* control device: the power on/off switch. The resulting magnetic field is static. It is always the same.

  • I have a piece of paper that says "Xnqygiip3a". It's results are instant, though only for hyper-specific tasks. And even with special care, it still make lots of errors.

  • Is it apathy, or is the west just simply not equipped to stay in the forefront of these types of technology developments? It seems lately that the west is outcompeted in many areas, especially around AI. Perhaps the west needs new Big Tech companies that thinks fresh ideas? Or perhaps more government regulation? Or what?
    I often think that the west is experiencing more than "monetary recession". It seems across the board - infrastructure wise, energy wise, etc. Is it synchronous with the west being weighed
    • by angel'o'sphere ( 80593 ) <angelo,schneider&oomentor,de> on Thursday October 28, 2021 @07:40AM (#61934921) Journal

      Is it apathy, or is the west just simply not equipped to stay in the forefront of these types of technology developments?

      Some parts in the west have no government with a foresight for the future that is putting money into research. Some parts in the west think if it is not private founded it is not good.

      Some parts in the west think every government program put into education/research (aka into an university) is a fraud to make PhDs rich.

      Actually one of those western governments destroyed the Japanese research programs and the famous "Miti"

      https://en.wikipedia.org/wiki/... [wikipedia.org]

      Because they found it anticompetitive if a government agency is funding research. Which it actually did not really do: 50% of the money came from the companies doing the research/development. In the end the assholes behind it even orchestrated a Japan wide banking crash and now wikipedia claims it was a speculation bubble, rofldi rofldi.

      Anyway: "the leaders of the west" do not believe in education, research and physics. Luckily the small countries in the west: do.

    • by sjames ( 1099 )

      We're far too busy spending money getting some guy to sing that he's a "GAINiac" and making sure people see it often enough to get a bit nauseated.

      That and making up rationalizations for why a medication that costs 30 cents to make should cost more than the median salary can pay for.

  • ... can only be used for very specific tasks in a highly controlled setting, and will generally give you the wrong answer (though you cannot yet tell when it is wrong and when right).

  • by Viol8 ( 599362 ) on Thursday October 28, 2021 @07:09AM (#61934855) Homepage

    I doubt the reviewers had first hand usage of the machine so they have to believe what the chinese scientists are telling them and frankly with something this politically as well as technologically important I'd take the data with a large bucket of salt given how well known the chinese are for exaggeration when it suits them.

    • by AmiMoJo ( 196126 ) on Thursday October 28, 2021 @09:00AM (#61935115) Homepage Journal

      This is why we keep falling further and further behind. It's 5G all over again.

      Didn't put in the R&D. When China figured it out, went into denial. Finally went whining to the government that waaah it's so unfair China got all the patents waaah.

      Instead of going into the denial stage this should be a wake-up call and a chance to learn from their paper and work on replicating it ASAP. Otherwise in a few years companies will be whining to the government again that China has all the good quantum computing patents and they have to pay royalties.

      5G. AI. Hypersonic weapons. How many times does this have to happen?

    • "So the number of continuous parameters describing the state of such a useful quantum computer at any given moment must be... about 10^300... Could we ever learn to control the more than 10^300 continuously variable parameters defining the quantum state of such a system? My answer is simple. No, never."

      - Physicist Mikhail Dyakonov

      The Case Against Quantum Computing [ieee.org]

  • From the article.

    "And even with special care, they still make lots of errors. "In the next step we hope to achieve quantum error correction with four to five years of hard work," said Professor Pan of the University of Science and Technology of China, in Hefei, which is in the southeastern province of Anhui."

    So how do the two compare for computing accurate results? With the "lots of errors" does the quantum computer even produce practical results? How good of a prediction is getting good error correction

    • If you want to see what it is like trying to get a reliable computation among noise errors, just look at the Slashdot 'Firehose > All' page. It is pretty useless just like that.
      • If you want to see what it is like trying to get a reliable computation among noise errors, just look at the Slashdot 'Firehose > All' page. It is pretty useless just like that.

        Well if it is a billion (I think the article used a larger number but...) times faster than normal computations and we run the problem a million times, maybe we could get a wisdom of the crowd like filter: https://en.wikipedia.org/wiki/... [wikipedia.org]

        Though I have my doubts.

  • by gweihir ( 88907 )

    As no "Quantum Computer" exists at this time (all that exist are clever fakes and analog computers miss-classified as "Quantum Computers"), the speed of all these is exactly the same: zero.

  • by Locutus ( 9039 ) on Thursday October 28, 2021 @09:40AM (#61935279)
    Does the world really want to see China in control of Deep Thought?

    Would they ask it the ultimate question about life, the universe and everything?

    What would they be asking it?

    LoB
    • by vbdasc ( 146051 )

      Does the world really want to see China in control of Deep Thought?

      I personally would be more worried of the possibility to see China in control of all our encrypted communications.

  • Winnie was lying in Grandma's bed looking at Little Red Riding Hood saying, "All the better to see you with, my dear."

  • "China's New Quantum Computer Has 1 Million Times the Power of Google's".

    Means nothing. They obviously stole the technology from Google.

    Oh wait...

  • A civil war field cannon can drive nails many times faster than a common nail gun. They won't be well driven, they may not be driven where you want them, and it takes a while to set up, but there it is, 10-20 pounds of nails driven in just a second.

  • "So the number of continuous parameters describing the state of such a useful quantum computer at any given moment must be... about 10^300... Could we ever learn to control the more than 10^300 continuously variable parameters defining the quantum state of such a system? My answer is simple. No, never."

    - Physicist Mikhail Dyakonov

    The Case Against Quantum Computing [ieee.org]

Intel CPUs are not defective, they just act that way. -- Henry Spencer

Working...