[Cryptech Tech] The PractRand randomness test suite

[Benedikt Stockebrand]

Hi Joachim and list,

thanks for the pointer; I'll take a look at what can be salvaged for
HWRNG testing.


As far as my own efforts on randomness testing are concerned: While
progress is painfully slow, I'm getting somewhat near finishing the
first part of the test framework I'm working on: An API for accessing
test data in a way that's convenient while still reasonably
performant---and threadsafe, too.  It took me a couple tries to beat
some unnecessary complexity out of it, but it really seems to be going
where I want it to go now.

Once I've got that sorted out, next will be some test suite handling
class that allows multiple tests to be run conveniently in parallel, to
avoid that rather annoying shortcoming of dieharder of running all
sequentially, and doing plenty of unnecessary I/O along the way.

After that it'll be another helper for the statistical stuff.  I'll have
to read up on statistical test methods first (at least by now I seem to
have found some reasonable textbook about this), and I'll also try to
find ways to avoid this (maybe use R?) if I can.

And once all this is sorted out it'll be all about coming up with tests
that actually match the needs of HWRNGs, rather than PRNGs.


Cheers,

    Benedikt

PS: @Rob:     Yes, C++ is every bit as painful as I remembered it---and
              then some:-(
    @Fredrik: Once this works out we hopefully have a reasonable tool to
              investigate those phenomena you found with the scope.

-- 
Benedikt Stockebrand,                   Stepladder IT Training+Consulting
Dipl.-Inform.                           http://www.stepladder-it.com/

          Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/