[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <5ea4da95517fc998cdf17539b964becf@smtp.hushmail.com>
Date: Sun, 26 May 2013 23:33:59 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: new dev box wishes (was: Contest)
On 26 May, 2013, at 23:09 , Solar Designer <solar@...nwall.com> wrote:
> On Sun, May 26, 2013 at 04:31:50PM -0300, Claudio Andr? wrote:
>> On 26-05-2013 15:59, Solar Designer wrote:
>>> We'll need to discuss where to put it. Replacing bull's 7970 with 7990
>>> is one easy option, but not necessarily the best one. Yes, I think we
>>> should make this 7990 gift card available for the entire JtR dev team
>>> online, in whatever machine.
>>
>> A new 'mamuth' machine having a 7990, a GTX 690 and any good APU would
>> be really a nice toy. Expensive, but powerfull.
>
> Do we really want a GTX 690, knowing that it's relatively slow for our
> purpose as compared to GTX 590 or two 580's? Is this just because some
> of our users will inevitably have these cards, so we have to optimize
> for them too? Or are we hoping that we'd be able to optimize enough
> that we'd outperform the 5xx cards?
The problem with 5xx is you can't buy them anymore. You could get used ones of course. IMO it doesn't matter much what we have, just anything so we can test and develop multi-GPU with CUDA.
> I think we may also want to install a Xeon Phi (despite of the price and
> lower performance per dollar than we achieve with GPUs for most formats),
> if we have people who would proceed to make use of it for JtR development
> soon enough (do we?)
That would be interesting. I haven't figured out how they will normally be used: Do we just get more cores to use with normal (?) OpenMP builds, or can we use OpenCL? Perhaps both alternatives are supported?
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.
