Forgot your password?
typodupeerror
Google Supercomputing Science Technology

Google To Build Quantum Information Processors 72

Posted by Soulskill
from the remember-when-google-was-a-search-company dept.
An anonymous reader writes The Google Quantum AI Team has announced that they're bringing in a team from the University of California at Santa Barbara to build quantum information processors within the company. "With an integrated hardware group the Quantum AI team will now be able to implement and test new designs for quantum optimization and inference processors based on recent theoretical insights as well as our learnings from the D-Wave quantum annealing architecture." Google will continue to work with D-Wave, but the UC Santa Barbara group brings its own areas of expertise with superconducting qubit arrays.
This discussion has been archived. No new comments can be posted.

Google To Build Quantum Information Processors

Comments Filter:
  • ...and now are developing their own version based on their 'learnings'.

    • by gweihir (88907)

      In other words, they are stupid. Not a surprise, but the cost is petty cash for them.

  • It will be nice to see if Google's interests and motivations will yield interesting, more practical results, in this area.

    • by gweihir (88907)

      Highly unlikely. This is a hard, and possibly unsolvable problem. One small team within Google is not going to do better than 25 years of so-far failed Quantum Computing research.

      • "If I have seen further it is by standing on the shoulders of Giants"

        -Isaac Newton

        • by gweihir (88907)

          BS is BS, no matter what shoulders you stand on. It does require some domain knowledge recognizing that though, and a lot of people fall for the latest hype every few years, again and again. Still hype and still BS. We will neither have practical Quantum Computing, nor working AI that deserves the name, nor practical, affordable flying cars in the next 30 years. All of these could well turn out to be impossible in this universe.

          • If everyone had that attitude, we would still be huddled around fires cooking animals that we killed with spears.

    • by tlhIngan (30335)

      It will be nice to see if Google's interests and motivations will yield interesting, more practical results, in this area.

      Better advertising, I guess.

      Target uses metrics and analytics to guess when a family has a baby on the way, but I'm guessing using this may allow Google to go beyond simple pattern recognition into more complex fields.

      OK, you looked at websites A, B, D and F, well, you're likely trying to get into college, but G, and I imply you need help on your essay. Here are ads for essay farms!

      After

  • by koan (80826)

    Now we are proper fucked.

  • The D-Wave unit really doesn't help them. Perhaps a dedicated QUBO solver isn't sufficient for their needs or the D-Wave doesn't look like it will scale (we already know that it can be outperformed by equipment much less expensive than itself but investing might still be worthwhile if the technology looks like it will scale over time).
    • by Anonymous Coward

      They are going ahead and doubling the qubits in their D-Wave to 1024, but as expected, it doesn't look like D-Wave matches the hype. D-Wave will supposedly have 2048 qubits ready by next year. These iterations could easily trounce classical computing if a) people can program the thing and b) the narrow class of problems it solves is actually important.

      • by sarkeizen (106737)
        I'd quibble with "easily trounce" since we don't really know how this technology scale. Does doubling the D-Wave Qbits double performance? As it stands we know from other benchmarks that reasonably large single machine can operate equivalently to D-Wave. That hardware costs a few orders of magnitude less then D-Wave so unless D-Wave looks like it SCALES - i.e. performance goes up, costs go down. Then it's not going to be useful to anyone except in side effects like power efficiency. Also remember this
        • by gweihir (88907)

          It is exceedingly unlikely for D-Wave to scale. Normal interconnect is the limiting factor in current digital computers. Keeping entanglement scales far, far worse than simple electric connections. While somebody may eventually build a true Quantum Computer, (the D-Wave isn't one), it is quite possible that it will never scale to any useful size.

          • by sarkeizen (106737)
            Agreed. IIRC it's unclear if very much entanglement is going on in the D-Wave Two. Which is possibly why they are able to add Qbits at the rate they have been and also why their performance is muddy at best.
            • by gweihir (88907)

              The D-Wave is not fully entangled (that is why it is a quantum annealer, not a quantum computer), which makes its scaling pretty meaningless, as it does not actually lead to a scaling in the quantum computing performance. That means it will never be faster than conventional computers.

              • by sarkeizen (106737)
                The way you phrased your post makes it sound like you're saying "more entanglement would make it a quantum computer". Which isn't correct the difference between a Quantum Computer (or gate-model QC device if you prefer) and what D-Wave has built is that it's based around solving a specific Hamiltonian which happens to map to NP-Hard problem of some interest. You could, in theory solve decoherence in D-Wave's device through something not unlike quantum error correction. However even with that it still wo
                • by gweihir (88907)

                  I actually say that full entanglement would make it a QC. "More entanglement" may just make it more efficiently for what it can do. I do not even see that happening, and currently it is not match for a conventional computer of equal cost.

                  • by sarkeizen (106737)
                    What do you mean by "full entanglement"?
                    • by gweihir (88907)

                      Really? Look up how QC works some time...

                    • by sarkeizen (106737)
                      I know a thing or two about quantum information theory. However that's the first time I've heard "full entanglement" used to describe some entangled state. More frequently you talk about negativity or cluster states. That of course comes from knowing something about the topic and you...well...don't. :)
  • by msobkow (48369)

    Let me know when we have "quantum" processors that actually out-perform "traditional" CPUs on the tasks that quantum processing is supposed to be good at (e.g. graph optimization.) Until then, it's all marketting smoke and mirrors and not worth shit.

    • by resfilter (960880)

      even a failed attempt is worth quite a bit, and worth watching; as far as technological research goes, we usually tend to fuck it up a few dozen times before we get it just right.

    • by HiThere (15173)

      Well there's really two different problems. One is the hardware, the other is the algorithms. Currently, if we had quantum processors there's only a few things it could do better than conventional hardware. Prime number factorization is the leading light in that, This doesn't describe much of what computers currently do, so currently a real quantum processor wouldn't be really useful except for things like "Quantu Key Exchange", code breaking, etc. But there *ought* to be a large number of things it co

  • That summary reads like something out of Star Trek. Superconducting qubit arrays! Imagine a positronic network of those. ;-)

  • Oh, Hell Yea! However, it will take time for the average computer user unable to handle current system with watercooling needs... My guess is that anything current does need super cooling as in liquid Nitro cold... So the likely scenario of a broadly deployable technology is probably a very long time away... for those wondering why? Read this: http://www.academia.edu/240382... [academia.edu]

"No problem is so formidable that you can't walk away from it." -- C. Schulz

Working...