Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Thu, 02 Feb 2012 12:03:22 +0100
From: Samuele Giovanni Tonon <samu@...uxasylum.net>
To: john-dev@...ts.openwall.com
Subject: Re: strange behaviour of c/s on raw-md5 and raw-md5-opencl

On 02/02/12 01:39, magnum wrote:
> On 02/02/2012 01:16 AM, Solar Designer wrote:
>> On Wed, Feb 01, 2012 at 03:43:09PM +0100, Samuele Giovanni Tonon wrote:
>>> same password file, same incremental mode on CPU it last for 7:13 at
>>> at 192911M c/s , on GPU it last for 5:12 at 48408 M c/s ? should'nt
>>> the c/s be higher ?
>>
>> This sounds like the integer overflow issue that I fixed in the CVS tree
>> a while ago.  It's triggered when you use a format that buffers a lot of
>> passwords (thousands) and at the same time you load a lot of hashes for
>> a certain salt (or total if saltless).
> 
> Samuele, the "bleeding-jumbo" branch on github includes these changes.

thanks,
i've tested it on bleeding jumbo the new raw-md5-opencl and
raw-sha1-opencl, now i get

./john -fo:raw-sha1-opencl -i:alpha7 /tmp/pass
[..]
Loaded 10000 password hashes with no different salts (Raw SHA-1 OpenCL
[raw-sha1-opencl])
guesses: 0  time: 0:00:05:03 DONE (Thu Feb  2 11:52:42 2012)  c/s:
275679M  trying: qqhjueq - qqjxjxq

./john -fo:raw-md5-opencl -i:alpha7 /tmp/pass
[..]
Loaded 10000 password hashes with no different salts (Raw MD5
[raw-md5-opencl])
guesses: 0  time: 0:00:05:42 DONE (Thu Feb  2 11:59:52 2012)  c/s:
244242M  trying: qqhjueq - qqjxjxq

finally getting some good numbers

Cheers
Samuele

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.