john-users - Re: core by core speed

Openwall GNU/*/Linux 3.0 - a small security-enhanced Linux distro for servers

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Tue, 23 Mar 2010 00:10:18 +0100
From: "Magnum, P.I." <rawsmooth@...dband.net>
To: john-users@...ts.openwall.com
Subject: Re: core by core speed

> thread: 6 guesses: 73  time: 0:00:00:56  c/s: 4759588M  trying: VLBj -
> VL.!
> thread: 7 guesses: 82  time: 0:00:00:56  c/s: 4703820M  trying:
> 48semc97 - 48semmos
> thread: 5 guesses: 78  time: 0:00:00:56  c/s: 4603104M  trying:
> shidds2m - shiddsli
> thread: 4 guesses: 97  time: 0:00:00:56  c/s: 4613056M  trying: PEV2i3
> - PEV2jb
> thread: 3 guesses: 66  time: 0:00:00:56  c/s: 4646022M  trying: hctagk7
> - hcta3am
> thread: 2 guesses: 73  time: 0:00:00:56  c/s: 4712449M
> thread: 0 guesses: 71  time: 0:00:00:56  c/s: 4727283M  trying: jmig1mm
> - jmikdor
> thread: 1 guesses: 144  time: 0:00:00:56  c/s: 4679187M  trying: Jodbg8
> - Jod0pu
>   8 nodes: guesses: 684  time: 0:00:00:56  c/s: 987845K total
> @@...@...@...@...@...@...@...@...@...@...@...@@@
>

This is off-topic, but some of those "total" c/s figures are amazingly 
incorrect. I don't think it would happen on my gear even with such 
extreme figures per core, but I'm not 100% sure. I'll test it.

I wonder if MPI_UNSIGNED_LONG could be 32-bit on a system that otherwise 
use 64-bit for longs (as indicated by the per-core figures)? That sounds 
pretty stupid if you ask me, but what else could it be?

magnum