Real random numbers generator

[daniel_st]

Hi again, guys.

I've jsut finished a Real (repeat, Real) Random numbers generator, you can get it on my homepage http://homedimension.webs.com
The generator is successfully tested on various Celeron processors, but Pentium 4 has failed. Still, if anybody wishes to try it, please do, and let me know which other processors are up to it.

enjoy

[boudino]

Sorry, but as far as I know, only possible real random number generator is hardware based (e.g. radio noise detection etc.). With software generators, you can only get worse or better statistical characteristics similar to those of random numbers.

[daniel_st]

Well, not any more. I guarantee you, this is 100% real random numbers generator, test it if you can. I don't very much like to reveal How it's made, though. But go, test it, it's real.

PS: It only does not work on some processors..

[laserlight]

Well, not any more. (...) PS: It only does not work on some processors..

So, the reason why it does not work on some processors is not because it is hardware based, but because your code has bugs?

I guarantee you, this is 100% real random numbers generator, test it if you can.

What are the statistical tests of randomness that you have tried it on, and for those processors on which it works, how well has it performed?

I don't very much like to reveal How it's made, though.

What's the idea? If you cannot reveal the principle, then it is very difficult to trust your guarantee, even though good performance for statistical tests of randomness might give us confidence to trust it as a PRNG rather than a "real" RNG.

By the way, rapidshare makes for a horrible file sharing website for those of us from countries with limited IP addresses. I'll have to wait for a long time to get a chance to download it.

[daniel_st]

Thanks for the opinion, shows interest, makes me proud.

So, the reason why it does not work on some processors is not because it is hardware based, but because your code has bugs?

No, my code does not have bugs. If it had, it wouldn't work on ANY processor, wouldn't depend on it, actually.

What are the statistical tests of randomness that you have tried it on, and for those processors on which it works, how well has it performed?

I haven't. I just developed it yesterday. If you need real random numbers, use it. If you need proof, test it. I will, when I have the time.

What's the idea? If you cannot reveal the principle, then it is very difficult to trust your guarantee, even though good performance for statistical tests of randomness might give us confidence to trust it as a PRNG rather than a "real" RNG.

Good, except it's not pseudo. I do not guarantee good stats. I guarantee real randomness. The idea is that it exploits certain phenomena, that occurs in most processors. Testes successfully on 4 different CPUs (Celeron 1.7, Celeron x64 1.8, AthlonII x2 and Atom) and one failed (Pentium 4 2.6)

By the way, rapidshare makes for a horrible file sharing website for those of us from countries with limited IP addresses. I'll have to wait for a long time to get a chance to download it

By the way paid hosts make for a horrible idea for smart developers with thin wallets, don't they? I know RS is not top notch, I will provide alternatives, when I have the actual possibility to provide good alternatives

Thank you

Like this one?

http://comscire.com/Home/

Note that just because a random generator is "real" doesn't guarantee that its statistic properties are better then what you can from a pseudo-random generator. Also natural sources of randomess can be skewed.

One major disadvantage with "real" generators lies in the fact that the random sequence isn't repeatable. This means for example that if you use them you can never exactly repeat (verify) a simulation experiment.

On a more philosophical note I'm not convinced a random generator utilizing some physical process is more "real" than one that's purely mathematical. I haven't seen any proof of that. I think Einstein was right. God doesn't play dice.

Yes exactly. Except it does not use noise, but other source of randomness. And of course, there are some very, very good PRNG, but some times, some people jsut need true randomness, that's it.

[nuzzle]

I've jsut finished a Real (repeat, Real) Random numbers generator,

Like this one?

http://comscire.com/Home/

Note that just because a random generator is "real" doesn't guarantee that its statistic properties are better then what you can from a pseudo-random generator. Also natural sources of randomess can be skewed.

One major disadvantage with "real" generators lies in the fact that the random sequence isn't repeatable. This means for example that if you use them you can never exactly repeat (verify) a simulation experiment.

On a more philosophical note I'm not convinced a random generator utilizing some physical process is more "real" than one that's purely mathematical. I haven't seen any proof of that. I think Einstein was right. God doesn't play dice.

[laserlight]

The idea is that it exploits certain phenomena, that occurs in most processors.

Then it is hardware based

[daniel_st]

Then it is hardware based

Well, all TRNG are. That's by definiton. But with this you don't need to buy RNG hardware, that's my point! The processor is a piece of HW everybody positively has. So you just need the SW, namely- my SW. And you can happily get real random.

[nuzzle]

And you can happily get real random.

Sure, but regardless of what physical phenomenon you monitor, how do you know it's random?

In my view, as long as it's not proved that physical random generators are better than mathematical random generators, why use them?

And above all, who needs them? Who gets happy from replacing a perfectly fine pseudo-random generator with well-defined properties, with some undefined unknown physical black-box?

[daniel_st]

undefined unknown physical black-box

That's the thing. Casinos, first, who can't risk someone starting to guess the next card.
Then military. I don't know of others, but those 2 are the major users of TRNG

[MrViggy]

Well, all TRNG are. That's by definiton. But with this you don't need to buy RNG hardware, that's my point! The processor is a piece of HW everybody positively has. So you just need the SW, namely- my SW. And you can happily get real random.

Not exactly. You've stated it doesn't work on all processors, therefore not everybody has the piece of h/w required to run your generator.

Viggy

[daniel_st]

Guys, can you clearly state what exactly is your **** problem!?! I made a program, that does what no other program does, and a program that is Free, where specialized hardware costs you MONEY! I don't think any of you really uses RNG, even if you do- not in a specialized case where you need true randomness. So if you don't need it- don't use it!

But stop the flood already. The programs works, yes, has some compatibility issues, but god damnit, it's only v1.0! And it works on 70% of the test machines.

[laserlight]

In my view, as long as it's not proved that physical random generators are better than mathematical random generators, why use them?

If we are talking about statistical randomness, then tests can give some confidence as to how these RNGs compare to good PRNGs.

And above all, who needs them?

Users of cryptography, e.g., to generate key material. For example, the key material for a one time pad cannot be generated by a PRNG, otherwise it ceases to be a one time pad. In such cases, the fact that the random sequence isn't repeatable is an advantage, not a disadvantage.

[daniel_st]

Update on the compatibility issue: The program turned out to actually Work on all CPUs, only not with the max bits. On P4 64-bit integer generation fails, but 63 and below work perfectly.

[daniel_st]

The PRO version of the assembly is now finished and available for purchase. This is The Only software, capable of generating True random numbers, now with added functionality. The PRO version can generate numbers with Unlimited amount of bits - 4096 and more-bit integers are easily generated, while retaining true randomness. Various tests are undergoing, and the generator is peforming even better than expected for now.

Thanks to all the sceptics.