Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 16 Apr 2015 22:10:03 +0300
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: [GSoC] John the Ripper support for PHC finalists

Agnieszka,

On Thu, Apr 16, 2015 at 08:55:27PM +0200, Agnieszka Bielec wrote:
> if gpu isn't faster than cpu, implementing opencl to john the ripper
> makes any sense?

Yes, it does!  Support for PHC finalists is primarily a research
project.  Results such as yours might affect the selection of PHC
winners, and/or eventual adoption of these hashing schemes.

Also, performance may vary across devices, including e.g. future GPUs
vs. future CPUs, and for different m_cost and t_cost (what if you try
even lower m_cost?)  We should be able to re-run such performance tests
in the future.

And it may be easier and even more cost-effective to install multiple
GPUs than multiple CPUs in a system.  Currently, it is cost-effective to
build up to 8 GPU machines, whereas a machine with 8 CPU chips may cost
substantially more.  So it may be 8 GPUs + 2 CPUs vs. 2 CPUs alone per
chassis.  Obviously, a machine with 10 chips will win over one with 2
chips even with same per-chip performance.

We similarly have bcrypt-opencl, even though it's not faster than its
host CPU counterpart per-chip.

> why is only 1 GPU on super now?

I am still seeing all GPUs:

[solar@...er ~]$ clinfo | grep '^  Name:'
  Name:                                          Tahiti
  Name:                                          Tahiti
  Name:                                          Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz
  Name:                                                 Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz
  Name:                                          Intel(R) Many Integrated Core Acceleration Card
  Name:                                          GeForce GTX TITAN

If "john" does not see the two Tahiti's for you, and it's not because of
any change in your tree, then I guess this indicates an AMD driver
issue, and we probably need to reboot.  Please let me know (off-list).

Thanks,

Alexander

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ