Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 21 May 2013 16:21:53 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Incremental mode in 1.7.9.14

On Sun, May 19, 2013 at 08:16:29PM +0400, Solar Designer wrote:
> I started two longer runs of these (with otherwise the same settings)
> yesterday, printing status after every 50G candidates.  Here's what I am
> getting for up to 500G candidates so far.  JtR 1.7.9, the lines
> correspond to 50G, 100G, 150G, ... 500G:
> 
> guesses: 367569  time: 0:01:37:27  c/s: 3081G  trying: gh3rdnuj - gh3rdnd5
[...]
> guesses: 445900  time: 0:14:04:11  c/s: 2710G  trying: 142202lan2 - 142202luv5
> 
> The new code (charset.c patch that I posted yesterday), similar 10 lines:
> 
> 376211g 0:00:51:20 122.1g/s 16233Kp/s 16233Kc/s 5742GC/s icaldjcs..icalt75d
[...]
> 451358g 0:08:18:13 15.09g/s 16726Kp/s 16726Kc/s 4509GC/s TNJ0j7x..TNJ0ga8

Here's where these got to by 1e12 candidates.  Old:

guesses: 468112  time: 1:02:55:17  c/s: 2598G  trying: askbbzdgb - askbbzdfa
766630 password hashes cracked, 233339 left

New:

473686g 0:16:33:03 7.949g/s 16783Kp/s 16783Kc/s 4138GC/s bansetbes9..bansetta2k
773223 password hashes cracked, 226746 left

In other words, this is 76.66% vs. 77.32% of total passwords cracked
(including duplicates), although in practice the weakest passwords would
be cracked by other means first, and the percentages cracked by all
cracking modes combined would be higher.  On the other hand, in practice
the training set might not match the test set nearly so well.  (Here
both came from RockYou, although I took non-overlapping portions of
shuffled RockYou.)

Alexander

Powered by blists - more mailing lists

Your e-mail address:

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