Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAAepdCZwNBbvQEi_jvt7Oh3BT9y=A-MLW2jurLs5LVwnNNuHdQ@mail.gmail.com>
Date: Tue, 17 Sep 2013 01:58:03 +0200
From: Rafael Waldo Delgado Doblas <lord.rafa@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: dominosec optimization

Hello Alexander,

2013/9/15 Solar Designer <solar@...nwall.com>

> This is the easy optimization task I had mentioned:
>
> The DOMINOSEC_fmt_plug.c source file in bleeding-jumbo is in need of
> some optimizations.  Also, it should have proper *_binary_hash() and
> *_get_hash() functions added, and the effect from this needs to be
> tested on loading and cracking of a test file with millions of hashes of
> this type (you'll need to generate some fake ones, of correct format).
>
> For the kind of optimizations you can make, see:
>
> git show 57a0e272bbba7187b3240c9c605281100b97515f lotus5_fmt_plug.c
>
> Basically, I made the optimizations to lotus5_fmt_plug.c, but not yet to
> DOMINOSEC_fmt_plug.c, even though these two are similar.
>
> In particular, in lotus5_fmt_plug.c I introduced 2x interleaving, as
> needed to make use of the higher instruction issue rate that modern CPUs
> are capable of (multiple instructions per cycle per core, beyond the
> amount of instruction-level parallelism that the non-interleaved code had).
>
> Of course, both of these source files are rather dirty, just like most
> of jumbo is...  It'd be nice if you make some cleanups to
> DOMINOSEC_fmt_plug.c while you work on it.  A rewrite of this file from
> scratch would be even better - but only if you can make it much cleaner.
>

Ok sounds like something that I can do before GSoC finish.

Best regards,
Rafael.

Content of type "text/html" skipped

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.