Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Thu, 17 Jan 2013 15:07:41 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: CUDA multigpu first draft for pwsafe

Lukas -

On Wed, Jan 16, 2013 at 10:22:12PM +0100, Lukas Odzioba wrote:
> Thanks to one of john users  I got access to GTX 690.

Cool!  Can you benchmark other CUDA and OpenCL formats on it as well,
and compare those to the older NVIDIA cards that we have?

> For identical gpus code scales very well:
> before:
> Raw:    96376 c/s real, 97090 c/s virtual
> after:
> Raw:    192752 c/s real, 192752 c/s virtual

The single GPU run here is a bit slower than what we're getting on
bull's o/c GTX 570:

Benchmarking: Password Safe SHA-256 [CUDA]... DONE
Raw:    108881 c/s real, 108196 c/s virtual

Local worksize (LWS) 256, Global worksize (GWS) 57344
Benchmarking: Password Safe SHA-256 [OpenCL]... DONE
Raw:    128143 c/s real, 128862 c/s virtual

and a lot slower than what we're getting on HD 7970:

Local worksize (LWS) 64, Global worksize (GWS) 57344
Benchmarking: Password Safe SHA-256 [OpenCL]... DONE
Raw:    352524 c/s real, 17203K c/s virtual

I am not surprised.

Nice work on multi-GPU, although I guess we'll approach it differently
at a later time.

Thanks,

Alexander

Powered by blists - more mailing lists

Your e-mail address:

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