Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 30 Apr 2012 07:31:56 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Password Generation on GPU

myrice, Samuele -

I'm sorry for postponing this discussion for a bit too long.  Let's get
to it now.  Please see below:

On Tue, Apr 17, 2012 at 03:19:18PM +0800, myrice wrote:
> On Fri, Apr 13, 2012 at 6:06 PM, Samuele Giovanni Tonon <
> samu@...uxasylum.net> wrote:
> >
> > yup, this is the other PITA with fast hashes:
> > with slow GPU the time spent on "hashing" is higher than time spent on
> > bus transfer; for this reason changing plaintext_length doesn't affect
> > it too much.
> >
> > Problem arise with newer GPU where "cpu time" is lower than "bandwith
> > time" .
> > If you all have some time try to change plaintext_length on ssha-opencl
> > you will see how it affects the speed; the other
> >
> > hashcat solved the problem, "sorting" password length from shorter to
> > longer; i suspect atom does that so he can create/destroy cl buffers
> > accordingly; however i don't know how this could be achieved in john.
> 
> I agree with you bandwidth may cause the problem.  Password generation on
> GPU will reduce the transfer time.
> 
> For term of "password generation", I just want to make it clear. Do you
> mean that in real password generation, the single, wordlist and incremental
> mode? I mentioned real because we split benchmark with real password
> generation. In benchmark, we just use bench_set_keys(set key from test_fmt
> one by one).
> 
> For now, I haven't explored so much of these modes. But I am familiar with
> ben_set_keys. So if I test it by modifying bench_set_keys, the result is
> also meaningful, right?

It is not clear what you're referring to by "test it" above.  Trying to
group candidate passwords (generated on CPU) by length, so that you
reduce the data transfer sizes?  If so, the benchmark should work fine,
as long as you have a wide enough variety of lengths in the tests[].

However, your message Subject is "Password Generation on GPU", so maybe
I misunderstood?  I'd like us to actually start discussing just that -
password generation on GPU - and I'll start to use this thread for that.

I'll post another message focusing on this actual topic.

Alexander

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ