lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20160616091416.GA5733@khazad-dum.debian.net>
Date:	Thu, 16 Jun 2016 06:14:16 -0300
From:	Henrique de Moraes Holschuh <hmh@....eng.br>
To:	Toralf Förster <toralf.foerster@....de>
Cc:	Linux Kernel <linux-kernel@...r.kernel.org>
Subject: Re: Why are the MB/s of avx and raid6: twice as high for a docked
 ThinkPad than for an undocked ?

On Wed, 15 Jun 2016, Toralf Förster wrote:
> This diff is reliable depending whether the T440s is docked (right) or not (left) :
> 
> Linux t44 4.5.7-hardened-r2 #1 SMP Wed Jun 15 23:39:10 CEST 2016 x86_64 Intel(R) Core(TM) i5-4300U CPU @ 1.90GHz GenuineIntel GNU/Linux
> 
> 215c215
> <    avx       : 10196.000 MB/sec
> ---
> >    avx       : 23504.000 MB/sec
> 225,234c225,234
> < raid6: sse2x1   gen()  2578 MB/s
> < raid6: sse2x1   xor()  2007 MB/s
> < raid6: sse2x2   gen()  3246 MB/s
> < raid6: sse2x2   xor()  2203 MB/s
> < raid6: sse2x4   gen()  3773 MB/s
> < raid6: sse2x4   xor()  2640 MB/s
> < raid6: avx2x1   gen()  5042 MB/s
> < raid6: avx2x2   gen()  5812 MB/s
> < raid6: avx2x4   gen()  6824 MB/s
> < raid6: using algorithm avx2x4 gen() 6824 MB/s
> ---
> > raid6: sse2x1   gen()  5816 MB/s
> > raid6: sse2x1   xor()  4515 MB/s
> > raid6: sse2x2   gen()  7261 MB/s
> > raid6: sse2x2   xor()  4968 MB/s
> > raid6: sse2x4   gen()  8511 MB/s
> > raid6: sse2x4   xor()  5916 MB/s
> > raid6: avx2x1   gen() 11378 MB/s
> > raid6: avx2x2   gen() 13117 MB/s
> > raid6: avx2x4   gen() 15398 MB/s
> > raid6: using algorithm avx2x4 gen() 15398 MB/s

Did you try setting the BIOS to "highest performance" mode even while on
battery?

BTW: update the BIOS of the T440s if you haven't done so already, AFAIK
there are some nasties in the outdated microcode and right now the only
way to get the fixed microcode for these processors is through a BIOS
update.  Latest BIOS for the T440s is from 2016-04-27.

> And why do the units differ (MB/sec versus MB/s) ?

For the typical "no good reason at all".

The proper SI units for these are actually "MiB/s", I think: AFAIK, the
kernel is measuring base-2 throughput, not base-10 throughput...

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ