Slashdot Mirror

← Back to Stories (view on slashdot.org)

VIA C3 Random Number Generator Reviewed

Posted by timothy on from the but-these-are-all-threes dept.

An anonymous reader writes "VIA has added a hardware random number generator to its Nehemiah C3 CPU. I found a recent review of its security. Interesting how it's done at the instruction level as opposed to the chipset level used by the i810 RNG (also reviewed there)."

11 of 289 comments (clear)

  1. Re:Cant the randomness be predicted? by jmv · · Score: 5, Interesting

    No. Generally the idea of devices is just to amplify thermal noise. Thermal noise is produced at the microscopic level with atoms bouncing into each other. There's no way to predict that... unless you tap directly into the generator (in which case it's simpler to just get the data on the computer before it's encrypted).

    --
    Opus: the Swiss army knife of audio codec
  2. Re:Artificial Intelligence vs Artificial Stupidity by jmv · · Score: 4, Interesting

    Actually, I've heard of some experiment where people were asked to fake a "coin tossing session" and write down the results. Generally, you could tell it's fake because when "generating random numbers" people tend not to repeat sequences.

    --
    Opus: the Swiss army knife of audio codec
  3. A Better System by lommer · · Score: 4, Interesting

    would be to use radioactive decay to generate random numbers. Very easy to implement using existeng technology, one of the few things that is completely random, and it's infinitely scalable to boot. A system I envision would simply moniter a radioactive sample for 1000 milli or micro seconds. Every sample time, it would record the number of fission events and if even, turn a bit on, if odd, turn the bit off. Then withing the space of a second you have a 1000 bit-long number that is COMPLETELY random.

    With this system perhaps it's possible to emulate the electric fields that generate the random number. Admittedly, with any complexity at all (as in a chip) this becomes impractical to do, but hey, why go for almost random when you can have truly random?

    1. Re:A Better System by Anonymous Coward · · Score: 1, Interesting

      intrinsic bias?

      use one sample, two detectors, one on the top and another on the bottom, AFAIK what triggers one can't trigger the other, esp if the sample emits beta instead of gamma rays.

  4. Re:Randomness by Anonymous Coward · · Score: 1, Interesting

    Very true. Human brains can also be tricked easily

    Example - If I ask you to "Pick a number between One and Four" about 90% of people will pick Two or Three.

    However, if I instead phrase the question "Pick a number from One To Four" about 90% of people will pick Three.

    Why? I verbalised the numbers One and Four, but also Two (To, Two, same thing) and the human brain trying to be random picks the one that wasn't mentioned.

    Don't believe me? Try it on your friends.

  5. Re:The Prototype: by Steven+Blanchley · · Score: 2, Interesting
    Interestingly enough, when asked to pick a number from 1 to 10, more people seem to pick 7 than anything else. My source is an unscientific poll, but I trust its general accuracy because I have noticed similar results myself. 7, 4, 5, 6, 3, 8, 2, 9, 10, 1. While the differences among 3, 4, 5, 6, and 8 could be off and the order of those probably cannot be trusted, 7 is definitely picked much more often than any of those five, which in turn are more likely picks than the four nearest either of the extremes.

    It isn't just one to ten either; I've noticed that when you ask people to pick a number from N to M, where N to M are relatively close together, a particular answer seems to come up more often than others; for example, 12 comes up often among numbers from 1 to 15. Can anyone suggest an explanation for this?

  6. Nehemiah core on EPIA boards by Stormie · · Score: 2, Interesting

    Does anyone know when VIA intend to release an EPIA MiniITX motherboard with a Nehemiah-cored C3 CPU? Apparently the M10000 they released recently was supposed to be so equipped, but turned out to only have a 1GHz version of the older Ezra-T C3 core. Since the Nehemiah core has a lot of improvements, this random number generator amongst them, I'd rather hang out for it than buy an M10000 now.. but how long must I hang?

  7. Where does noise come from? by mindpixel · · Score: 4, Interesting

    I remmeber going to the university science library when I was 14 to try to find out how to write a program to generate random numbers...found a big yellow book about pseudo-random number generators and thought, no, I want a real random number generator...of course I opened the book and discovered that it is impossible inside a deterministic system...you have to stick an antenna into an external universe...then I thought where the fuck did the universe get noise? Why isn't the universe one big symmetric crystal?

    Now I sit here looking at a 2 billion year-old hypernova and no one here can answer this question (There are at least 5 cosmoligists within spitting distance of me right now)...

    1. Re:Where does noise come from? by kinnell · · Score: 2, Interesting
      where the fuck did the universe get noise?

      I think there are several conclusions you could come to:

      1. The book (or mathematics) is wrong
      2. The universe is not a deterministic system
      3. Random numbers do not exist

      I would lean towards 2 - aren't quantum processes such as radioactive decay non-deterministic? This seems to be a matter of opinion, even among great physicists. If, however you hold that the universe is deterministic, I suppose you would have to come to the conclusion that nothing is entirely random.

      As a side not, noise does not imply non-determinism: gaussian white noise, for example, is entirely deterministic (depending on the above ;-) ), and is generated in a system where a large number of "random" events summed together contribute to the whole. For example chaotic movements of electrons in a hot conductor giving rise to electrical noise. In such a case, the noise is in practice unpredictable, so is treated as non-deterministic.

      --
      If I seem short sighted, it is because I stand on the shoulders of midgets
  8. Re:Finally by Anonymous Coward · · Score: 1, Interesting

    Not true, If the numbers were 32 bit floating point then the probability of getting 10 0's in a row is 2^320, which is about 10^96, there have been of the order of 10^17 seconds since the universe began, so the probability of anyone, anywhere, anytime getting that sequence is vanishingly small.

  9. Mmm.. thermal noise by Scorchio · · Score: 2, Interesting

    Similar to what Douglas Adams suggested as a random number generator, 25 years or so ago, I guess. This implementation is a little more convenient - although slightly less tasty - than a fresh really hot cup of tea.