Most significant HPC stories of the year

HPCwire’s half-year retrospective gives the world’s most powerful QC third place as most significant HPC stories of the year so far. I think it should be #1 but I have a slight bias.

11 thoughts on “Most significant HPC stories of the year

  1. surname: might I suggest that you collect all your thoughts before posting and post one comment per post instead of 10+ per post? Also if you could try to keep the comments on topic that would be great also. As I mentioned to a different poster, if you’d like to contribute to the advancement of a field of science/technology, you have to first understand what people know and the evidence supporting that knowledge. If you really care about QC you should learn quantum mechanics (every university in the world has courses) and read the intro literature on QC. If you want to understand superconducting approaches to QC you can start by reading the NSA RSFQ review linked to in the sidebar http://www.nitrd.gov/pubs/nsa/sta.pdf .

  2. So you building quantum processor with ~100GHz… Perhaps it is your secret… then with 1024 qubits, maybe, maybe…

  3. I know how calculate, how many grover iterations need to grover algorithm, i can exactly calculate, what probabily to get wrong or good answer, but i can’t understand, how grover’s algorithm can speed up searching in unstructure database. And i feel very excidet, what i can’t understand somthing, and maybe i will understood leter, but for now grover’s algorithm to me is magical black box – true oracle… So i just want say to over’s, who interest in quantum computers, next. Supose we somthing searching with grover’s algorithm. If we something searching, it means, that we know, what we searching. So if we know, what we searching, then why we must search, what we know??? Example we searching state |101>. We know this state (it is |101>), so why we must search this state (if we know this state)???

  4. Surname: You search for |101> not because you know |101>, but because you need to know if that is a valid state within a list. For example, I know a dude with the name Mr. Blarg, and want to see if he’s in a New York phone book. I dropped the book, however, so all the names are non-alphabetic. To know whether or not he is in the phone book, a standard computer would need ~18 million steps (because he actually lives in Boise building landing strips for gay martians). A QC using Grover’s Algorithm would know he isn’t in NY in a thousandth of the time (once the speeds increase). But once the “phone book” is large enough, the QC search should be able to outperform the Classical Computer search even with a slower clock speed. Besides, everyone knows that classical computing is reaching the physical limits for speed (not to mention economical, nobody’ll be able to afford the new fab plants). If QC isn’t looked at, as well as other novel methods of computing, how’ll we ever know the question whose answer is 42?
    I suspect you are using a Babelfish to translate, which is why there is so much confusion (I almost never know what you are saying, know what I’m sayin’).

  5. But it still don’t clear. Okey, there is book, and we don’t know is the state |101> in this book or isn’t. But we STILL MARK THIS state |101> and computer will give answer THIS state |101>, becouse First oracle MARK |101>. So grover algorithm still give to us that state, which we mark!!!
    I don’t know maybe there is some dificult code, not just |101> state…
    You say quantum computer speed up quadratic if database is large, but you dont understand, that intel processors has quadratic more tranzistorrs, than Orion (quantum transitors) with 1024 qubits!!!

  6. The QC running Grover may say that there in NO |101> in the database. Intel currently has way more transistors, but we are reaching the physical limit for transistor/power density with classical computers. If QC is at the very beginning of the field (the quantum C64 if you will, which is still one of my favorite computers), it should be able to increase the qubits dramatically in the next few years.

    Besides, QC isn’t really meant to displace classical computers, they will just do certain things a hell of a lot better (like atomic modeling). Gaming, probably not.

  7. But how QC can say, that there no |101>. We search |101>, qubits go trough Hadamard gates. Next Oracle mark the state |101> with negative amplitude. All overs states is unmarked, then On output will be |101>, becouse it was marked. ?
    I don’t say, that quantum computers don’t has perspective, but only, that 1024 qubits is to small number to compete with Core2 processors in unstructure database searching. Maybe with 10000-100000 qubits things will change. Or maybe it just perspective to the future.
    But compare on such things like atomic modeling (where is exponentional speed up), then quantum computer of course even with 512 qubits will be more faster than millions supercomputers…

  8. @surname

    if you read Dr. Rose’s blog you might find out that the adiabatic system is fundamentally different then the standard gate model taught in university courses. Go read Dr. Farhi’s original papers so you see what the difference is mate and maybe Seth Llyod’s as well as he explains the basis for d-waves machine although the gadget is likely a shade different then the orignal theoretical conception of a superconducting Adiabatic QC SL came up with.

    if i get it correctly, they are letting the hardware take an encoded problem in the form of ferromagnetic and antiferromagnetic couplings in flux qubits (current in superposition?) and you chill it down to a ground state where the approximate answer to a query is the minimization of the encoded optimization problem functional or at least sufficiently close enough.

    Clearly the gate/circuit model algoritms do not apply to D-waves hardware and should be discarded as a source of questions. here is a nice dictionary type intro to what QC is

    http://plato.stanford.edu/entries/qt-quantcomp/

    @Dr. Rose: question of my own. was there any mention of the next level of qubits in the 2nd gen device? i am admittidly a skeptic of the show i attended in vancouver so i am piqued to see if you will increase the qubits to say — 64 or 128 or 3386 or whatever and show through a toy problem that you are getting quantum effects and a small speedup. For instance, can orion say, use the AQC algorithm to solve a sufficiently sized k-sat problem and then run the equivalent on the same hardware (turning off the quantum ^^) limiting it to the case of the most efficient classical k-sat solvers that are well documented in the black sea of literature? even a 10-20% speedup (although likely trivial in D-waves eyes) is enough to grab significant market share in the target industries you aim for. if so then you can have my whole bank account as investment because my nserc is not cutting it these days, lol.

  9. Geordie wrote:
    ” HPCwire’s half-year retrospective gives the world’s most powerful QC third place as most significant HPC stories of the year so far. I think it should be #1 but I have a slight bias “.

    I think, there is written the order is cronological, not order of importance.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s