Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 3 Apr 2012 23:52:40 +0400
From: Solar Designer <>
Subject: Re: MSCash2 CUDA

Lukas -

On Tue, Apr 03, 2012 at 11:34:37PM +0400, Solar Designer wrote:
> And THREADS 512 is even faster:
> Benchmarking: mscash2-cuda [GPU]... DONE
> Raw:    29619 c/s real, 29619 c/s virtual

However, it failed to crack a test password.  I put this in a file:


It corresponds to qerwt per tests[].  Then I ran:

./john pw -fo=mscash2-cuda -i=alpha

with john.conf edited to use MinLen and MaxLen 5.  This completed
without getting the password cracked:

guesses: 0  time: 0:00:06:53 DONE (Tue Apr  3 23:44:49 2012)  c/s: 28730 trying: qcyqy - qvxqv

Then I tried:

$ echo qerwt | ./john pw -fo=mscash2-cuda -stdin
Loaded 1 password hash (mscash2-cuda [GPU])
guesses: 0  time: 0:00:00:06  c/s: 0.16  trying: qerwt

All of this was with THREADS 512.  I also noticed that when I increase
THREADS further to 768, I get ridiculously high performance per
benchmarks (over 1M c/s), yet the self-tests somehow "pass".  Perhaps we
need to improve the self-testing...

Finally, just to confirm the test vector:

$ echo qerwt | ./john pw -fo=mscash2 -stdin
Loaded 1 password hash (M$ Cache Hash 2 (DCC2) [SSE2i 8x])
qerwt            (?)
guesses: 1  time: 0:00:00:00  c/s: 2.85  trying: qerwt


Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.