Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 06 Apr 2012 11:20:32 +0200
From: Samuele Giovanni Tonon <samu@...uxasylum.net>
To: john-dev@...ts.openwall.com
Subject: Re: JtR:Multi-GPU Setups

On 04/06/2012 09:55 AM, SAYANTAN DATTA wrote:
> 
> Hi,
> 
>     The AMD card needs to be running X, but the nvidia card only needs you
>     to run "nvidia-smi" and it will create the /dev nodes you need to use
>     the card. Don't waste resources on a second X server.
> 
> Did you run any openCL application? If so, which vendor's openCL did you
> use?   Does it detect both the cards?

i keep reading the list even if i'm not really active, so forgive me
if i come back from time to time and then i vanish, i have some works
to finish :-(

i really think a mixed environment (nvidia+amd+intel) is a big chaos.

this because of the /dev/* files created, the /etc/OpenCL/vendors/*.icd
that could create side effects unneeded.

i really believe at the moment we should test dual cards of the same
family, with nvidia it should not be a problem but with AMD it means
having a "dummy" x server with dummy devices connected to make it work .

Don't forget other important aspects:
- User must be logged in the X server
- and must be the same to use OpenCL: i tested this some months ago i'm
not able to test it again at the moment but feel free to try it.

This make me believe that you can't work on multiple card at the same
time, but you need to run a separate john session for the other card ;
at least for AMD and withouth taking in account things like SLI or
CrossFire (never used both of them, i saw some threads on hashcat forums)

Cheers
Samuele



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.