Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 28 Mar 2015 13:22:33 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: [GSoC] JtR SIMD support enhancements

On 2015-03-28 03:13, magnum wrote:
> On 2015-03-28 02:36, Lei Zhang wrote:
>> As there're a couple of subtasks in this project I apply for, I'd
>> like to know if you have any suggestions on what to do next.
> 
> While I'm not sure of the full answer to that, I suspect you will
> benefit from familiarizing yourself with sse-intrinsics.c (currently
> handling anything from SSE2 up to and including XOP/AVX) and macros like
> SIMD_COEF_32 and SIMD_COEF_64, and for some formats the GETPOS macro.

BTW an easier start, if you like, is to modify rawSHA256_ng_fmt_plug.c
and/or rawSHA512_ng_fmt_plug.c (possibly even rawSHA1_ng_fmt_plug.c) so
they take advantage of AVX2 and/or AVX-512 when available. These are (I
think) the only Jumbo formats that doesn't use the shared code - they
are more or less self-contained.

The work you'd invest in these formats can be re-used in
sse-intrinsics.c later on.

On a side note the main reasons these formats are kept at all (they are
currently slower than their non-ng alternatives) is they can potentially
be optimized easier, eg. reversing steps and such things.

magnum

Powered by blists - more mailing lists

Your e-mail address:

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