[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <00a901cd3524$7524e760$5f6eb620$@net>
Date: Fri, 18 May 2012 13:31:35 -0500
From: "jfoug" <jfoug@....net>
To: <john-users@...ts.openwall.com>
Subject: RE: Performance Considerations of stdin
>From: Brad Tilley [mailto:brad@...ystems.com]
>
>What, if any, performance considerations are there when passing word
>candidates to JTR via stdin? I plan on doing some tests to see first-
>hand, but before doing so I wanted to ask here in case others have had
>some experience with this.
>
>Basically, I'm wondering if a word generation program that passed words
>into JTR via stdin would perform better than when JTR is doing both word
>mangling and hash generation, hash testing, etc all by itself.
>
>ehco -n word | john --sdtin
>
>Has anyone had experience performance testing something such as this?
Depends upon the OS. If it is winblows, then the performance is horrible.
I get about 30% or so performance for fast hashes, vs a dictionary read into
memory, with rules run on the dictionary. Windows has a TINY pipe_buffer.
I have found no way to speed this up (and I have tried).
However, for many other OS's, may likely be a performance boost by using
-stdin mode, IF doing some complex work, since you can do much more complex
things a lot more efficiently in a native language (like C, or even perl),
much faster than john's rules.
I have done this for things like a super elite builder, which can do all
sorts of permutations, such as full permutation of letter case, case, full
permutation of all elite possibilities (such as mississippi -> m1ssissippi,
miss1ssippi, m1ss1ssippi, ... m!ssiss1pp|, etc). Doing this in john, with
rules, or other extern stuff 'could' be done, but that is beyond my
capabilities, and I think you would either have to write a lot of rules that
either reject a lot of input words, looking for specific patterns, or the
rules would generate a lot of dupes.
I also did this, for a much more heavily complex time/date generator, that
did an all in one type run, and could handle much more than the external
date_time code in john.conf.
I would love to hear any speed comparisons between
john -rules:xxx -memory=9999999 -w=input.dat candidates.in
and
word_gen input.dat | john --stdin candidates.in
where the --rules:xxx (or externs, etc) and word_gen were 'comparable', when
run on other OS's.
Jim.
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.
