[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20210309180025.GA22890@openwall.com>
Date: Tue, 9 Mar 2021 19:00:25 +0100
From: Solar Designer <solar@...nwall.com>
To: john-users@...ts.openwall.com
Subject: Re: Multi-gpu setup
On Mon, Mar 08, 2021 at 04:22:47PM +0100, Micha?? Majchrowicz wrote:
> Quick question. Does john splits gpu tasks equally?
When you use "--fork" matching your GPU count, yes, it tries to split
equally.
> I just run a
> md5crypt dict attack on a multi gpu system and noticed one of gpus
> finished way faster than other one. If so can specify how big
> percentage will go to one gpu and how big to other ?
Not really, but as a hack you can list your faster GPU multiple times,
like this:
./john -fork=3 -dev=1,1,2 -format=something-opencl hash
This will run two concurrent instances of the OpenCL kernel on device 1,
but only one instance on device 2, still splitting the work between the
three instances equally - thus, giving device 1 twice more work.
As a cleaner workaround, you can run separate instances with "--node":
./john -se=1 -dev=1 -node=1-2/3 -format=something-opencl hash
and concurrently:
./john -se=2 -dev=2 -node=3/3 -format=something-opencl hash
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.
