Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 24 Jun 2015 05:13:08 +0300
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Agnieszka's weekly report #8

On Tue, Jun 23, 2015 at 02:34:26AM +0200, Agnieszka Bielec wrote:
> accomplishments:
> 
> I have exams now and I only ported Lyra2 CUDA  to JtR this week. with results:
> [a@...er run]$ ./john --test --format=lyra2-cuda
> Benchmarking: Lyra2-cuda, Generic Lyra2 [ ]... DONE
> Speed for cost 1 (t) of 8, cost 2 (m) of 8
> Raw:    303 c/s real, 304 c/s virtual
> 
> and 1k on my laptop but I have some problems with drivers now, My
> drivers stopped working, I installed drivers once again but when I
> restarted my laptop drivers stopped working again
> 
> It was suggestion to port CUDA code to JtR to comparison between
> OpenCl and CUDA but this version requires a lot of work to get better
> performance but I really don't see any sense to work on both versions
> at once

I am not suggesting that you optimize Lyra2 authors' CUDA code -
I merely suggested that we have it in JtR tree for easy comparison to
your OpenCL code.

It sounds like you got it in, but somehow it delivers poor speeds?
Can you please include a comparison against your OpenCL code on the same
GPUs?  Both super and your laptop.

Also, as magnum and I mentioned before, please do fill in the format
name and algorithm name in some reasonable way.  "Generic Lyra2 [ ]" is
obviously wrong.

> priorities:
> 
> Still will be working  on Lyra2 on GPU

OK.

Alexander

Powered by blists - more mailing lists

Your e-mail address:

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