[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Wed, 12 Mar 2014 20:48:03 +0100
From: magnum <john.magnum@...hmail.com>
To: john-users@...ts.openwall.com
Subject: Re: pbkdf2-hmac-sha512 on GPU
On 2014-03-12 19:51, dy3afg9+halwpc@...rrillamail.com wrote:
> Is it at all possible to crack pbkdf2-hmac-sha512 on GPU, specifically, openCL?
>
> I’ve compiled the bleeding-jumbo from github which doesn’t appear to have pbkdf2-hmac-sha512-opencl as an option, yet pbkdf2-hmac-sha256-opencl is.
>
> On CPU, I’m getting around 20C/s with fork set to 8 (quad core 3.7ghz Xeon)
>
> If anyone could help speed up the process for this hash format it would be much appreciated.
> Joe.
It is basically trivial but there's nothing committed. Lukas might have
an experimental one. GPUs usually have a hard time with SHA-512 though,
more bugs than usual use to surface whenever 64-bit stuff are used.
Perhaps a CUDA kernel for this would be better as a first version - any
old CUDA driver is usually a couple orders of magnitude more stable than
whatever run of the mill AMD driver.
What build target did you use? Using the --test benchmark I get 228 c/s
on my 2.3 GHz laptop using OMP, or 240 c/s with fork=4. Maybe your
reported speed was for some IRL hash(es) with a higher iteration count?
magnum
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.
