[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 24 Jun 2012 17:59:48 -0400
From: cc <cc@...heads.com>
To: john-users@...ts.openwall.com
Subject: Re: dynamic formats tutorial
On Sun, Jun 24, 2012 at 5:04 PM, Solar Designer <solar@...nwall.com> wrote:
> Yet I think this is valuable, hence the posting. In fact, as I said
> before, I'd appreciate it if folks announce their JtR-related blog posts
> and such in here.
I recently put together a hack to use JtR in a distributed fashion -
pre-dividing the keyspace and using a cgi program to distribute work
units to the workers. Each client gets a unit of work and the
currently uncracked hashes (this means that, for salted hashes, units
get faster as the salts get reduced).
It's currently a dumb brute force because I wanted to use it quickly,
and the overhead of pre-dividing Incremental was too high. I've been
able to speed it up since, but it's still quite large (months to
generate) - as long as you have a small number of worker cores (~10)
they get starved out by unit generation. Thus brute force - which
takes a few seconds to divide the keyspace into ranges. (The keyspace
division only needs to be done once for a given unit size, so if
anybody has "entry, length, fixed, count, numbers[0-8]" snapshots that
divide the entire keyspace into ~100-300k chunks that would be really
helpful). I was originally going to make in an External mode but I
needed numbers bigger than maxint, so it uses --stdin.
Anyway the usual caveats apply, your mileage may vary, etc. It works
well if you have more heterogeneous compute than patience to optimize
guesses.
https://github.com/ccdes/brutus
-cc
Powered by blists - more mailing lists
Powered by Openwall GNU/*/Linux -
Powered by OpenVZ
