[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 14 Dec 2011 12:57:00 +0400
From: Solar Designer <solar@...nwall.com>
To: john-users@...ts.openwall.com
Subject: faster Lotus5 hash cracking (was: JtR CUDA ????)
On Wed, Dec 14, 2011 at 10:13:23AM +0800, Gu George wrote:
> I have tried to crack some Lotus Domino hashes by John-1.7.7-jumbo. I used
> following command: .\john.exe --incremental -format:lotus5
> .\noCrackedHash_1.txt,
BTW, there's a newer build for Windows here:
http://openwall.info/wiki/john/custom-builds#Compiled-for-Windows
As far as I am aware, it is not any faster at Lotus5 hashes, though.
> and the benchmark is about 94508K c/s.
I guess this is not the benchmark, but rather speed seen while cracking.
It depends on the number of hashes loaded for cracking a lot.
> I have
> sucessfully carcked about 100 hashes in about 2 weeks, but there are about
> 300 hashes which are not carcked.
Sounds reasonable.
> My PC has an Intel i7,8 core CPU@...z.
You mean quad-core, 8 logical CPUs.
> So, my question is: is there a more fast version of John-Ripper, such as
> CUDA support?
For this hash type, no.
> I have checked http://openwall.info/wiki/john/GPU, it seems
> current patches do not support Lotus5 format.
That's correct.
> Another question: Is it possible to configure John-Ripper to run more
> faster in my PC?
Yes: you need to run 8 instances of it in parallel, configured to try
different candidate passwords and using different --session names.
If doing this for 8 instances is difficult, then do it for at least 4 -
e.g., you may split by length:
http://openwall.info/wiki/john/parallelization#IncrementalAllN
You can also try building with MPI support, but that's more difficult.
I guess we should implement OpenMP parallelization for this hash type,
but we haven't done so yet.
Alexander
P.S. Your reuse of the old thread for a mostly unrelated topic was in
fact undesirable.
Powered by blists - more mailing lists
Powered by Openwall GNU/*/Linux -
Powered by OpenVZ
