An article about Andy

Here is a great article from The New York Times about Andy Bechtolsheim and the very cool machine he is building. Just like Sun likes to say, the network is the computer! A good summary:

A number of Silicon Valley technologists are, however, betting on Mr. Bechtolsheim. “He’s a perfectionist,” said Eric Schmidt, Google’s chief executive, who worked with Mr. Bechtolsheim beginning in 1983 at Sun. “He works 18 hours a day and he’s very disciplined. Every computer he has built has been the fastest of its generation.”

23 thoughts on “An article about Andy

  1. The basic premise (of B. always building the fastest
    machines like Cray did in his days) seems untrue to me. e.g. he built 4 and 8 socket Opteron machines before his last startup got bought by Sun.

    While those are nice and fast boxes they are not really in the league of super computers.

    He also built some parallel streaming video server there which while also an interesting project was definitely not the fastest in CPU power (or even IO
    bandwidth) either.

    And Blue Gene and various SGI Altixes like Columbia easily hold the crown in these days.

    Now the interesting part is when will Orion beat a larger
    Blue Gene or even a 8socket Opteron at something?

    But I guess it’ll be already interesting if a given problem
    can be solved cheaper than on a large super computer.

  2. Why can’t orion to run shor algorithm? Becouse, orion dont use error corection, or becouse it is a adiabatic quantum computer?

  3. Nobody novadays disagree that universe is computer. And, yes, it is true. But it computer work only for itself. On we smalest particles level, computation is so haotics, that actualy, no can be build from it. It is the same, like from sand built mechanic computer. So, on the level where are the smallest particles, the computation is very powerful, but its don’t do anything solid with some non chaotics laws of particles. On the level, there particles is biger than particles on chaotic level, the computation have some laws of form. And on the level clasical phisics, computation have a very strong form of laws, so can be used to computate somthing not chaotic. But not chaotic computation is almost infinity times less poverfull than computation on chaotic level.

  4. But actualy, unknow phisics is classical phisics, but just in it pure form, without such ingredients like gravity, temperature and overs.

  5. So there no many unioverse or multiuniverse. There is just much smolest particles and more delicate lows o phisics. That if we do qubits much smaller than electron and they be very near each over, then no entanglent will be and no any superposition will be, and just it will be clasical computer. Becouse, qubits, can’t change information faster than light. So quantum computer is actualy classical computer, but in one pm say is as much small particles like atoms in universe and those particles, make lows of phisics, diference from low of classical phisics, and it is what we call entanglement and superposition.

  6. interesting is experiment with beam splitter. If single foton traveling from to beam splitter, then he split into two parts, and if we try to measure foton in one traveling direction and in second direction we get single foton, like he be before, with 50 % probability in one traveling direction and in anower traveling direction. But supose, that foton split and two parts go far away one from another. And at this time we measure foton in one direction. And if information still is in two parts then second part with speed of light must go back to the first part. if we measure foton in first part direction. And if two parts foton go say milions km one from anower than we must wait 10 seconds until foton from anower part goes back to the first part. And we measure foton 10 s leter than just if we measure foton in same traveling direction without beam splitter. Maybe sombody, already make such kind experiment?

  7. or maybe there is foton and some kind antifoton and we measure foton with 50 % probability in one of two direction, and antifoton never back if foton and antifoton never merge together. But if they join together then it will be like show in deutsch lecture2 experiment.

  8. How quantum register can be in so many states in same time? First of all, say all atoms go to superposition. Then smallest particles run with speed of light to each over and exchange information with all particles each. If some state probability increase, then like from sand clock, sand from each particles superposition goes to the potential gap, that made over particles going through each particle. So don’t nesasary that particles be infinity small, or that be infinity universes. Speed of light make his job, that so many qubits exange information, even in superposition.

  9. philosoph: In order to make a contribution to our understanding of nature you have to understand what is already known and the arguments supporting this knowledge. I would suggest taking the time to learn quantum mechanics (it’s taught at pretty much every university in the world).

  10. surname: the fundamental reason the Orion system wasn’t designed to run Shor’s algorithm isn’t noise, its AQC-ness or anything else technological. The real fundamental reason is that there is no good business case for building special-purpose code-breaking machines. You can see also my Red Herring post from a few months ago.

    If there were a good reason to build factoring machines I think the main unresolved technical issues could be resolved. The central one is development of a superconducting AQC-friendly version of Shor’s algorithm. The constructs that I’m aware of for Shor-AQC are not practical.

  11. Hi Geordie!

    Correct me if I’m wrong but isn’t the fundamental reason for Orion’s design not being able to run Shor’s algorithm the very same reason for Orion’s design not being able to run commercially relevant quantum simulations? The fact that the system is not UQC without XZ couplings?

  12. Then, how goverment and banks can be sure, that Dwave, dont secretly work on shor algorithm or that dwave nobody sells orion copy to sombody else?
    Maybe you secretly want to crack RSA in bank say?

  13. All RSA, can’t be changed in something else per few years.

    And how about quantum simulation? Does quantum computer exponentionaly speedup, quantum simulation? Or quantum simulations algorithms still not exist, not created yet?

  14. What will be after then scale of tranzistor size reach 0.5nm? I think that it will be going slower and slower, and ofcourse nanotubes maybe be like tranzistors. And enginers will made processors time by time with more and more econimic (electro energy) architecture. And there is a question, maybe it is some more powerful and advanced computation than quantum computer? But if such kind computation even exist, it still will never be useful by humans (or aliens). You may ask why. Becouse if quantum computer give speedup on classical computation, and quantum computer is so sensitive to noise, then if exist more powerful computation, which can be used, than it will be hundred and milions time more sensetive to noise, than quantum computation. On such computation work our Universe… You just imagine, that our universe is made of infinite many balls, which is say smallest than electron is bilions and trilions times or maybe more. This balls is elementar computer… This balls just have some mass, but don’t has any kind of energy like gravity, or somethink else… So you may ask, how this balls can made say fotons, electrons and atoms, and if they don’t have any energy, then how they can compute somthing, and how they can make universe such like we it see? This is simple. Yes, balls, dont has any energy, but becouse, universe is infiniti in space, then you couldn’t know in what possition to this balls is good to be from point of energy. So if balls in universe is also infiniti, then they push each over, and it can be, in some period in time and space, found very strong power moving these balls to some directori. If there are many such moving many balls power they can hit each over, and made some simple law of new phisics, but if such powers is very much, then can be made say even electron or foton. And this balls move from one directory to over each and hit each over. Do you imagine how many computations those balls do? But they are so chaotic, that it computations can’t be used to compute even such simple equation like 2+2=4. So this balls has very powerfull computation power, but it power can’t be used to computate something stable. But this balls on more complex level made laws of quantum mechanic, and quantum mechanic make laws of classical phisics. But as i say, computation is more powerfull on lowest level, where is balls – the smallest particles of the Universe.

  15. No more smallest particles, than balls. Balls are truly elementary particles. Balls has ideal sphere form. All balls in universe is exactly the same mass. So supose, one ball push anower ball, thats anower ball push anower anower. But it is not just two balls, but infinity balls, and say in some qubic space is 10 such balls. And all of this balls push each another. So you imagine how complicate is that computation, then all balls push each anower from diferent trajectories? I think suck computation is much more powerfull than quantum computation with 10 qubits. 10 balls pushing each anower, from point of qomputation is much more powerfull, than quantum computer with 10 qubits. And also it is scalable, say 1000 balls pushing each over have (do) much more computation power, than ideal quantum computer with 1000 qubits.

  16. nextquant: it is true that the current design doesn’t have XZ or XX couplers, and these are required for universality, which is required for Shor & Qsim. Integration of XX or XZ couplers into a design isn’t all that difficult though.

  17. hmmm. maybe its a good idea to nullify the commentary and hire someone on site to ask a questions with a spell checker. im sure you could get a student for $7.50 an hour and youd be paying more then most companies or labs do. just a thought! ^^. hope to see more qubits soon.

  18. So Orion can run only Grover’s algorithm? Becouse, if shor’s algorithm and quantum simulation can’t be run at time, then only grover algorithm can be performed on Orion. Do you plan in future, for more qubits quantum computer made posible quantum simulation? Becouse in my opinion quantum simulation is somthing more than just grover searching algorithm. And quantum simulation, like i understand, gives exponentional speedup compare to quadratic for grover algorithm.

Leave a Reply

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

You are commenting using your 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