[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130522103908.GA6672@openwall.com>
Date: Wed, 22 May 2013 14:39:08 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: X32 build, anyone?
On Wed, May 22, 2013 at 11:24:40AM +0530, Dhiru Kholia wrote:
> On Wed, May 22, 2013 at 6:40 AM, magnum <john.magnum@...hmail.com> wrote:
> > Has anyone tried an X32 Linux? And in particular, benchmarked John the Ripper on it?
>
> I tried x32 some time back but I didn't see any noticeable performance
> gains in JtR.
How/what did you test?
There may be slight performance gains when going from x86_64 to x32 for:
1. Lots of fast hashes (thousands or millions) loaded for cracking,
where the savings from smaller pointer size may reduce the working set
size and thus result in more cache hits. However, this effect should be
reduced in bleeding (as compared to 1.7.9-jumbo-7) due to us starting to
use bitmaps (instead of hash tables), which are same size and speed
regardless of pointer size.
2. Bitslice DES in OpenMP-enabled builds, where DES_bs_all structs
contain arrays of pointers, and together these structs don't fit in L1
data cache. (They could fit with fewer instances computed per thread,
but then the OpenMP overhead would be higher.) Without OpenMP, we have
just one DES_bs_all, which fits in L1 data cache, so there shouldn't be
much difference.
Alexander
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.
