[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20120712141347.GA18720@openwall.com>
Date: Thu, 12 Jul 2012 18:13:47 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Get hash number in cmp_all()
myrice -
On Thu, Jul 12, 2012 at 04:14:38PM +0400, Solar Designer wrote:
> [...] you should keep the grouping by salt. Specifically, when
> crypt_all() or cmp_all() is invoked for a specific salt, how do you
> identify just which hashes to compare the computed ones against?
> Scanning the entire hash array and skipping wrong-salt entries would be
> inefficient.
Actually, given that GPU formats tend to use high max_keys_per_crypt,
this naive approach may be OK for early experiments. It will only cause
substantial slowdown when the number of loaded hashes is significantly
larger than the number of hashes computed per crypt_all() call, since
you only need to do this scanning once per call.
For actual/"final" implementation, we'll need to do something better,
like the salt id thing I mentioned.
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.
