[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20181125154942.GA6353@openwall.com>
Date: Sun, 25 Nov 2018 16:49:42 +0100
From: Solar Designer <solar@...nwall.com>
To: john-users@...ts.openwall.com
Subject: Re: Chunk of work specification
On Sun, Nov 25, 2018 at 01:23:48AM +0100, magnum wrote:
> --node=n/100000 which would split the job in 100,000 chunks and *n*
> could be eg. 123 (for 123rd chunk) or (for a 20x faster machine) eg.
> 400-419 (for 20 chunks starting at 400). Just think "chunks" instead of
> "nodes". You could use even larger numbers like --node=n/1000000.
Right, but there's not necessarily a need to consider machines' relative
speeds. It may be OK for the faster machine to be completing and be
assigned new tasks 20x more often. Alternatively, chunks may be
assigned per CPU core or per GPU rather than per machine - then the
relative performance differences are less, but the master script needs
to manage more "nodes".
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.
