Re: Fwd: Effect of Processor and Memory on KDE4 execution speed

On Tue, Feb 21, 2012 at 6:07 PM, Devin Teske <devin.teske_at_fisglobal.com>wrote:

> On Tue, 2012-02-21 at 14:42 -0800, Robison, Dave wrote:
> >
> >
> > -------- Original Message --------
> >                           Subject:
> > Effect of Processor and Memory on
> > KDE4 execution speed
> >                              Date:
> > Tue, 14 Feb 2012 09:39:36 -0500
> >                              From:
> > Mehmet Erol Sanliturk
> > <m.e.sanliturk_at_gmail.com>
> >                                To:
> > FreeBSD Current
> > <current_at_freebsd.org>
> >
> >
> > Dear All ,
> >
> > Today I have encountered a case which I think informing you about it may
> be
> > useful .
> >
> > In my previous messages , I have mentioned very slowness of KDE4 .
> >
> >
> > Onto another computer I have installed DruidBSD 9.0 b56 amd64 , and KDE4
> .
> > In that installation KDE4 worked surprisingly fast .
> >
>
> Great! Glad to hear it's fast!
>
> Though, a couple corrections about the statement above.
>
> First correction:
>
> You meand "FreeBSD Druid", not "DruidBSD".
>
> The "FreeBSD Druid" is a FreeBSD installer.
> The "DruidBSD" is a micro distribution that can't/doesn't install
> anything.
>
> Second correction:
>
> The "FreeBSD Druid" installs a vanilla, completely unmodified
> RELENG_9_0_RELEASE copy of FreeBSD (albeit using sysinstall(8) instead
> of bsdinstall(8)).
>
>
> > To understand whether difference is among FreeBSD or DruidBSD , I have
> > installed
> > FreeBSD 9.0 Release amd64 and KDE4 on the same computer instead of
> DruidBSD
> > .
> >
> > The KDE4 has worked flawlesly i.e. , means very fast .
> >
>
> Cool. Glad to hear it's fast there too (makes sense; see below)
>
> With respect to the binaries that are being laid-down by the FreeBSD
> Druid, there is no difference in comparison to the official installer.
>
> However, the FreeBSD Druid does do some post-install configuration of
> the system that the official installer does not (though none of these
> customizations should have any effect on KDE in any way shape-or-form):
>
> 1. It enables sshd root login in /etc/ssh/sshd_config
> 2. It enables netwait in /etc/rc.conf (by adding netwait_enable="YES")
> 3. It installs perl-5.14 (as a package so if you don't want it, you can
> pkg_delete it to get rid of it)
> 4. It sets the root password
> 5. It installs /usr/local/sbin/host-setup
> 6. It configures root login-shell to be /usr/local/sbin/host-setup for
> easy setup on first-login (once exited, must be run manually henceforth)
> 7. It installs rsync (as a package, so if you don't want it you can
> pkg_delete it)
> 8. It installs bonnie (as a package; don't want -- pkg_delete)
> 9. Enables SU+J on /tmp /var and /usr
>
> None of these customizations should have any effect on system
> performance whatsoever.
>
> So I recommend not treating this as a "FreeBSD Druid" versus "FreeBSD"
> problem.
>
> NOTE: and again, "DruidBSD" is not an installable OS, only a micro Live
> Distribution used for doing rescues or simply booting into a mfs
> environment.
>
>
>
> > To make equivalent the installations on both computers , I have installed
> > FreeBSD 9.0 Release amd64 and KDE4 on the slow computer exactly as in
> fast
> > computer .
> >
>
> I'm confused.
>
> In the immediately-above paragraph you mention a fast computer and a
> slow computer. But in the paragraphs that proceeded it, I only see
> mention of fast-running KDE4.
>
> If you read the previous paragraphs carefully, you're stating that both
> installers ("FreeBSD Druid" and the official, both) produced fast-
> running installations of KDE4.
>
> Which is it?
>
>
>

"Fast computer" means "FreeBSD executed KDE4 fast"  ( having Intel
processor E2220 ) ,
"Slow computers" means "FreeBSD executed KDE4 slow"  ( having Intel
processor Q6600 ) ,

The problem is not generated by disk layouts , because sysinstall and
bsdinstall
are NOT factors on the problem : Only physical placement of memory chips in
the slots :

Rearranging memory chips in the slots  or using a different chip set with
the same ( brand , model , speed ) is causing the problem or it is curing
the problem .



> >
> > Starting times after first boot ( to eliminate initialization effects )
> are
> > the following
> > ( All timings are from "root" ) :
> >
>
> Is this on the box that was installed with FreeBSD Druid or box
> installed with official 9.0-RELEASE media?
>
>
> > >From "startx" ( which contains "exec ... kde4 ..." )
> > to   appearance of KDE menu symbol at the bottom left corner :
> >
> >
> > Fast computer : 8 GB : 0+ ( < 1 ) minute ( 4 x 2 GB )
> > Slow computer : 4 GB : 2+ ( < 3 ) minutes ( 2 x 2 GB ) ( 2 x ! GB chips
> > removed ) ,
> >                 6 GB : 8+ ( < 9 ) minutes ( 2 x ( 2 , 1 ) GB ) .
> >                 ( Memory chip installation conforms to main board manual
> . )
> >                 ( The clock does not have second counter . )
> >
> > Fast Computer
> >   CPU : Intel Pentium Dual CPU E2220 _at_ 2.40 GHz ( 2397.65-MHz K-8class
> CPU )
> >   ACPI APIC Table : < INTEL DG965WH >
> >
> > Slow Computer
> >   CPU : Intel Core 2 QUAD CPU Q6600 _at_ 2.40 GHz ( 2397.65-MHz K-8class
> CPU )
> >   ACPI APIC Table : < INTEL DG965WH >
> >
> > ( The main boards are the same ) .
> > ( All of the memory chips are the same : Kingston HyperX 800 MHz )
> >
>
> In your metrics above, which machine (fast vs slow) was installed with
> which installer?
>
>
>
> > I could not understand the reason(s) of the differences .
> >
>
> Disk partitions? (I'm guessing).
>


As I explained above , disk layouts do not have any effect .


>
> There's really no difference between what the official 9.0-RELEASE
> installer installs and what the FreeBSD Druid installs. They were both
> compiled from RELENG_9_0_RELEASE within a day from each other. The
> likelihood that the binaries would differ with respect to code they were
> compiled from is highly unlikely.
>
> NOTE: The installed kernel should even have the same MD5 (can't say the
> same for the rest of the system, because I did have to do a fresh
> release(7) process -- though differing MD5s will not matter, the code
> was built within 24 hours from the official release build).
> --
> Devin
>
> _____________
> The information contained in this message is proprietary and/or
> confidential. If you are not the intended recipient, please: (i) delete the
> message and all copies; (ii) do not disclose, distribute or use the message
> in any manner; and (iii) notify the sender immediately. In addition, please
> be aware that any message addressed to our domain is subject to archiving
> and review by persons other than the intended recipient. Thank you.
>



Up to now I did not recognize the effect of memory chips , I have used the
same computer for all
of my FreeBSD installations .

When I have installed FreeBSD Druid into another computer , I have reached
to the conclusion of
memory chip placement effect .


I am NOT an electronics engineer but there is a concept "Impedance
mismatch" :

http://en.wikipedia.org/wiki/Impedance_matching
( I could not find a proper explanation of "Impedance mismatch" , but it is
  opposite of "Impedance matching" . When impedance mismatches a part of a
circuit can not
 trigger other part of the circuit . ) .


Perhaps such an effect is causing different detection of some value and
this is triggering
setting a variable for memory management to a very negatively effective
value .

This is only a pure guess without any proper knowledge .



Thank you very much .

Mehmet Erol Sanliturk