Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Fri, 29 Apr 2011 09:10:17 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: sha256 format patches

On Wed, Apr 27, 2011 at 11:07:07PM +0200, ?ukasz Odzioba wrote:
> Finally i've got something faster than 9600m, it is 9800GT. Still
> slower than Alexander's 8800GTS 512 but for the next few weeks should
> be enought.

Sounds fine.

> For sha256cuda slow i've got ~11k c/s (8800gts - 16k c/s).
> I found bug in my phpasscuda code, i was getting wrong results, and
> many weird things because of to restrictive max_registers limit. After
> two days of tracking this bug code went messy, so I need to clean it
> up before uploading on wiki.

That's curious.  Thank you for sharing this info.  I guess, this might
also be what caused the false positive in my testing of the SHA-256
"slow" patch.  (I think the correct output was left somewhere in memory
from a test.)

> As zero revision sha256cuda patch performance is poor, but it should
> be easy to double it.
> Currently it is about 30k c/s of phpass $P$9.

If it's for phpass, then why do you call it sha256cuda?  Is that just an
error in your posting above?

The "9" means 2048 iterations, so your 30k c/s means 60 million
iterations of the MD5 compression function per second.  This is
reasonable for a low-end/old GPU, although as you correctly say there's
a lot of room for improvement.

Please upload your phpass code to the wiki soon!

> oclHashcat+ can do 135k c/s on faster 8800GTS 512, so there is much to be done.

Thanks,

Alexander

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ