[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <65067370.615c7d40.51dd35ab.74fb1@o2.pl>
Date: Wed, 10 Jul 2013 12:21:31 +0200
From: marcus.desto <marcus.desto@...pl>
To: john-dev@...ts.openwall.com
Subject: Re: PBKDF2 hash_out treatment
Hi Alexander,
Dnia 10 lipca 2013 12:04 magnum <john.magnum@...hmail.com> napisał(a):
> What are you ultimately trying to achieve?
I have written a python program to compute (a lot of) sha1 hashes by using pbkdf2-hmac-sha1.
This computation is running on CPU only.
To increase speed I would like to use my old GPUs (I have a lot of them).
So, I was looking for an implementation of pbkdf2-hmac-sha1 in opencl - I have found JtR.
> In general it's not a good idea to use JtR to produce complete, standard-formatted hashes because much of that is often optimized away. For example, since SHA-1 is big endian, chances are you will not see the little endian you are looking for (instead, we byte swap the target hash(es) - this saves us from gazillions of swapping operations). And the CPU formats would normally use SIMD on x86, so the output might be interleaved or coalesced.
So what do you suggest?
>
> Also, I'm not sure but I would assume the mscash2-opencl format does not support an iteration count of 1, again because of optimizations.
I am using pbkdf2_hmac_sha1_unsplit_kernel.cl not the mscash2-open opencl implementation.
>
> magnum
>
Regards,
Marcus
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.
