Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Fri, 25 Jul 2014 18:32:37 +0000
From: Pyrex <pyrex@...networks.net>
To: "john-users@...ts.openwall.com" <john-users@...ts.openwall.com>
Subject: RE: Help Compiling John on ARM w/ MPI & CUDA

So I started looking into the "OMP_NUM_THREADS" variable a little more because I wanted to know why it is that I had to define it in the first place. I looked into how it gets defined in general. So this eventually got me to a point where I really started to look at my processor details on my dev boards and I discovered something very interesting. Since it's an ARM processor the only active core is the core 0 and the others are turned on ONLY when needed. This is part of the power saving methods for the ARM processor in general. Since I could care less about an extra watt or two I started researching exactly how to disable this function and force all cores to be on at all times.

To cut a long story short I created an RC script to run the following commands at boot:

echo 0 > /sys/devices/system/cpu/cpuquiet/tegra_cpuquiet/enable
echo 1 > /sys/devices/system/cpu/cpu0/online
echo 1 > /sys/devices/system/cpu/cpu1/online
echo 1 > /sys/devices/system/cpu/cpu2/online
echo 1 > /sys/devices/system/cpu/cpu3/online
echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor

This worked perfectly, not only did it enable all the cores but it allowed me to utilize OpenMP without the need for the "OMP_NUM_THREADS" variable declaration which it was probably not working prior to the above as the others cores would NEVER wake up even when the primary was maxed out.

I also enabled OpenMP with MPI by editing the john.conf and changing the following:

MPIOMPmutex = Y

to

MPIOMPmutex = N

I am still running tests to determine if OpenMP is in fact being utilized correctly as well as monitoring the cpu time with "htop".

I have noticed a small increase in numbers with WPA2 + CUDA + MPI + OMP...

If anyone has any feedback or suggestions for me I would appreciate it.

Thank you

-Pyrex

________________________________
From: Pyrex
Sent: Thursday, July 24, 2014 12:36 PM
To: john-users@...ts.openwall.com
Subject: RE: [john-users] Help Compiling John on ARM w/ MPI & CUDA

That worked perfectly! How would I fix this permanently and make use of it with cuda / wpapsk + MPI as magnum suggested?

-Pyrex
________________________________
From: Dhiru Kholia<mailto:dhiru.kholia@...il.com>
Sent: ?7/?22/?2014 4:34 AM
To: john-users@...ts.openwall.com<mailto:john-users@...ts.openwall.com>
Subject: RE: [john-users] Help Compiling John on ARM w/ MPI & CUDA

On Mon, 21 Jul 2014, Pyrex wrote:

> I am using GCC 4.8.2 but it doesn't seem to like the idea of OpenMP.
>
> I configured and I am getting an "OpenMP = Yes" but when I run the
> program I am getting the "Warning: OpenMP is disabled; a non-OpenMP
> build may be faster" message again.

Try running something like,

$ OMP_NUM_THREADS=4 ../run/john --test --format=net-md5

Does this work?

Dhiru
No malware was found: Scanned by TiSnetworks E-Mail Security Systems.


Content of type "text/html" skipped

No malware was found: Scanned by TiSnetworks E-Mail Security Systems.

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.