Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 18 Mar 2012 08:39:24 +0400
From: Solar Designer <>
Subject: Re: SSH thread-safety

On Sun, Mar 18, 2012 at 04:45:24AM +0400, Solar Designer wrote:
> Here's what I get on FX-8120 at stock clocks (3.1 GHz base, 4.0 GHz turbo):

Apparently, the documented base frequency is not relevant when the CPU
is under load and not overheated. gives 3.1 GHz
base, 3.4 GHz full load turbo, and 4.0 GHz half load turbo for this CPU.
In the BIOS overclocking settings, I can only specify two values for the
multiplier: base and turbo.  I've just tried setting base to 4.0 GHz and
disabling turbo (so should be 4.0 GHz as well) - got an unreliable
system (kernel panic shortly after starting to use XOP from all 8
threads).  Then I tried 3.8 GHz base and 4.2 GHz turbo - and this seems
to work.  I am getting almost exactly a 5% speedup for single-threaded
benchmarks and around a 10% speedup for OpenMP benchmarks (although it
varies between the different JtR formats).  Thus, it appears that
without overclocking I had 3.4 and 4.0, and now I have 3.8 and 4.2 GHz.

> Non-OpenMP build:
> user@...l:~/john/magnum-jumbo/src$ ../run/john -te -fo=ssh
> Benchmarking: ssh [32/64]... DONE
> Raw:    146448 c/s real, 146448 c/s virtual

Benchmarking: ssh [32/64]... DONE
Raw:    153300 c/s real, 153300 c/s virtual

> OpenMP:
> user@...l:~/john/magnum-jumbo/src$ ../run/john -te -fo=ssh
> Benchmarking: ssh [32/64]... (8xOMP) DONE
> Raw:    717852 c/s real, 89196 c/s virtual

Benchmarking: ssh [32/64]... (8xOMP) DONE
Raw:    793987 c/s real, 98967 c/s virtual

BTW, FX-8150 is documented to use 3.6 base / 3.9 full load turbo / 4.2
half load turbo, so maybe I should try 3.9 / 4.2.


Powered by blists - more mailing lists

Your e-mail address:

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