Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Tue, 12 Nov 2013 21:28:09 +0100
From: Katja Malvoni <kmalvoni@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: bcrypt-parallella on 64-core (was: Katja's weekly
 report #13)

Hi Alexander,

On Tue, Nov 12, 2013 at 9:18 PM, Solar Designer <solar@...nwall.com> wrote:

> On Tue, Nov 12, 2013 at 07:33:04PM +0100, Katja Malvoni wrote:
> > If I transfer keys only when changed, performance on E16 is 1207 c/s and
> on
> > E64 it's 4812 c/s.
> > And code seems to be reliable. I left "while :; do rm -f john.log
> john.pot;
> > seq 0 49999 | sudo -E LD_LIBRARY_PATH=$LD_LIBRARY_PATH ./john -stdin
> > -form=bcrypt-parallella ~/pw-bcrypt-2salt-50k 2>&1 > /dev/null | fgrep
> > 0:00; done" running for half an hour, it cracked all hashes on both E16
> and
> > E64 system.
>
> This is exciting news!  I recall you didn't expect to get all 64 cores
> working for bcrypt on this chip (because of the errata).
>

That is correct, since this approach of using local memory instead of
shared DRAM failed earlier I didn't even think of it when it turned out
that some cores have problems with fetching data from external memory.


> Is the code committed?
>

Yes it is -  git clone https://github.com/kmalvoni/JohnTheRipper -b
master-E64

Katja

Content of type "text/html" skipped

Powered by blists - more mailing lists

Your e-mail address:

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