[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <55F7D56B.8070504@mailbox.org>
Date: Tue, 15 Sep 2015 10:23:07 +0200
From: Frank Dittrich <frank.dittrich@...lbox.org>
To: john-dev@...ts.openwall.com
Subject: Re: Judy array
On 09/15/2015 09:53 AM, magnum wrote:
> On 2015-09-15 09:29, Frank Dittrich wrote:
>> Even with fast salted hashes, it is a problem when you have may be 4
>> instances of john running, each with about 2 million hashes, and
>> cracking many passwords.
>
> You are using the default SALT_HASH_LOG of 12, right? You load a
> thousand times more salts than that number is really good for. Please
> try bumping it to 22 for the case of 2 million hashes. This will speed
> up things significantly, not just pot sync.
No, actually, I was using SALT_HASH_LOG 24. (I just verified that my
params.h change is older than my john binary.)
But without getting rid of that pot sync, you could really see when the
cracking stopped for quite a long time. And top showed that, the longer
the sync took, the more john processes stopped running and switched
their status.
(It took me some time to figure out why out of my 6 or 7 john processes
just one or two where running (status R). I ended up killing and
resuming the only running process, just to make the other processes run
again, until I figured out what caused this problem.
I was about to report that issue on github.)
> In fact we might want to bump that default a bit (at least in Jumbo).
May be a good idea.
Frank
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.
