Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 11 Apr 2015 20:02:16 +0200
From: magnum <>
Subject: Re: [GSoC] John the Ripper support for PHC finalists

On 2015-04-11 16:39, Solar Designer wrote:
> On Thu, Apr 09, 2015 at 06:34:26PM +0200, magnum wrote:
>> md5crypt (opencl_cryptmd5_fmt_plug.c) would be an excellent base for a
>> single-kernel format. Is there any chance Pomelo would benefit from a
>> split/loop kernel?
> Definitely.  PHC finalists are expected to often be tuned to defensive
> running times of 100 ms or more (including multiple seconds), and this
> may be many seconds of latency on GPUs with their high concurrency and
> lower clock rates.
> We'll use lower cost settings when determining the CPU/GPU parity
> thresholds, which is one of the goals of this projects, but other than
> that the formats should work for high cost settings as well.
> So perhaps opencl_cryptmd5_fmt_plug.c isn't a good long-term choice to
> use as a base for these new formats, although it'd probably work for our
> initial experiments and it may be simpler.
> magnum, what base format would you suggest for a new split kernel format?

RAR5 might be a clean base. Many of the other ones support vectorizing
(decided at run-time) so are somewhat more complex (although that might
be interesting for Pomelo too?). I suspect some of Claudio's formats
might be a good start too, I haven't looked much at them as he actively
maintains them himself.


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.