Other Directory Sites: SeekWonder | Directory Owners Forum
The following archive was created by hippie-mail 7.98617-22 on Sat Apr 10 1999 - 01:18:25
bram (bram@gawth.com)
Tue, 2 Feb 1999 11:37:17 -0800 (PST)
On Mon, 1 Feb 1999, Mike Rosing wrote:
> There are lots of cheap RNG's, none of which seem to have passed things
> like DIEHARD. Good shielding to prevent external contamination is also
> expensive. Either you missed a smiley or the connection between
> technical know-how and cost of a product escaped your education.
>
> As I'm in the process of writing up a do-it-yourself description of how
> to build an RNG that does pass DIEHARD
I think building an RNG which passes that standard, while noble, is
unnecessary - a CSPRNG can fix most of the problems of an RNG, as long as
you have an accurate conservative estimate of how much real entropy is
coming out of the thing. Likewise, shielding can be much less (or
eliminated entirely) if you simply use the potential reduction in entropy
caused by external forces as a factor when estimating entropy of the
output.
> If there was a way to turn all the hours I've spent creating a working
> RNG that actually does pass DIEHARD into cash, I'd realy like to know
> about it.
It's possible that you could convince a chip company that they really need
an embedded RNG and that you're the person to design it for them.
I think that long term the solution to the RNG problem will be for all
CPU's to come with a built-in one, like the new Pentium III's will. It
might not take long for this to happen - CPU's are enough of a commodity
that anything you can do which adds real functionality might be worthwile
just for marketing reasons.
-Bram
The following archive was created by hippie-mail 7.98617-22 on Sat Apr 10 1999 - 01:18:25