[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CABh=JRHVXp6d7XhWcg2o60n1Uf+fo2XqWkzTqEeV8UtbobmPog@mail.gmail.com>
Date: Sat, 10 Nov 2012 12:17:27 +0200
From: Milen Rangelov <gat3way@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Split kernel for OpenCL WPA-PSK
Hello,
In fact the number on hashcat page is incorrect. Speed can be pushed to as
far as 143K c/s on 7970 @ stock clock with newer drivers. Of course that
means using the most aggressive settings. hashkill's performance can also
be tweaked to surpass 140K c/s if we are using higher GWS, however the
hardcoded maximum right now does not allow that (I should increase it for
the --double mode though). In both cases (hashkill/hashcat tweaked) desktop
becomes rather...unresponsive so that's the price you pay for the
performance.
On Sat, Nov 10, 2012 at 2:23 AM, Solar Designer <solar@...nwall.com> wrote:
> magnum -
>
> On Fri, Nov 09, 2012 at 08:00:50PM +0100, magnum wrote:
> > how about upgrading Bull to the driver versions recommended by HashCat?
> > It's Catalyst 12.8 and nvidia 304.32.
>
> I was/am going to install whatever is latest, but not right now as I
> have no time to deal with possible side-effects or perform a downgrade
> in the following couple of weeks. So let's revisit this in 2+ weeks
> from now.
>
> > Using device 0: Tahiti
> > Local worksize (LWS) 128, Global worksize (GWS) 262144
> > Benchmarking: WPA-PSK PBKDF2-HMAC-SHA-1 [OpenCL]... DONE
> > Raw: 129453 c/s real, 52428K c/s virtual
>
> Great!
>
> > This is a tad faster than HashCat unless Atom has tweaked it since the
> > figure I found (otherwise he will now... OTOH I'm not done yet =).
>
> The hashcat website says 131k c/s, and it did for a while (not a recent
> change). Milen mentioned getting 137k with hashkill in a tweet a while
> ago. All of these are for HD 7970 at stock clocks.
>
> > Jumbo-7 does 42723 c/s out of the box.
>
> IIRC, it did 55k when built with OpenMP. Still relatively slow, of course.
>
> > If I only could figure out how to compile clcc on Bull. I have no chance
> > to look at the referenced Ptx assembly. Or is there a way to keep temp
> > files?
>
> I hope someone else will answer this better, but we can try chattr +a on
> a directory (needs root privs) or LD_PRELOAD'ing a library with dummy
> unlink() (can be done as non-root) if nothing else helps.
>
> Alexander
>
Content of type "text/html" skipped
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.
