Re: [gentoo-user] Linux becomes expensive ;)

[Dan Farrell]

On Sun, 3 Jun 2007 19:37:19 +0200
Florian Philipp <f.philipp@××××××.de> wrote:

> Am Sonntag 03 Juni 2007 19:06 schrieb Ryan Sims:
> > On 6/3/07, Florian Philipp <f.philipp@××××××.de> wrote:
> > > Am Sonntag 03 Juni 2007 18:03 schrieb Dan Farrell:
> > > > On Sun, 3 Jun 2007 13:16:33 +0200
> > > >
> > > > Florian Philipp <f.philipp@××××××.de> wrote:
> > > > > Am Samstag 02 Juni 2007 20:03 schrieb Jeff Horelick:
> > > > > > Florian,
> > > > > >
> > > > > > That's not that big of a difference...Also, Gentoo/Linux
> > > > > > does not have powersaving for every device like Windows
> > > > > > XP...it's writing to the hard drive more often and it
> > > > > > doesn't spin as much down when it's not in use to help
> > > > > > performance. Also, if i was you, i'd be worried about your
> > > > > > system using that LITTLE energy especially since you have a
> > > > > > pretty hefty CPU, video card, motherboard, 2 hardrives and
> > > > > > al the rest of your components.
> > > > > >
> > > > > > On 6/2/07, Florian Philipp <f.philipp@××××××.de> wrote:
> > > > > > > Hi guys!
> > > > > > >
> > > > > > > I've just tested the energy consumption of my PC.
> > > > > > > Aparently Gentoo consumes a
> > > > > > > quiet a bit more than Windows XP: 213 W compared to 188 W
> > > > > > >
> > > > > > > PowerNow is activated and works on both cores (tested).
> > > > > > > The same hardware is
> > > > > > > plugged in and works. I'll attach the output of lspci,
> > > > > > > lsmod and cpuinfo as
> > > > > > > well as my world-file just in case it's related to some
> > > > > > > software.
> > > > > > >
> > > > > > > Is there anything I've forgotten? Where does my energy go?
> > > > > > >
> > > > > > > A short overview of my hardware:
> > > > > > >
> > > > > > > AMD Athlon64 X2 4200+ EE
> > > > > > > Asus M2N32-SLI Deluxe (WLAN should be deactivated)
> > > > > > > 2048 MB DDR2 Corsair
> > > > > > > SoundBlaster Audigy 2 ZS
> > > > > > > ATI Radeon 1950 Pro (fglrx)
> > > > > > > 2 SATA2 HDDs
> > > > > > > 1 SATA1 DVD-RAM
> > > > > > > Floppy
> > > > > > > USB mouse, keyboard and printer
> > > > > > > TFT screen (connected via DVI)
> > > > >
> > > > > Well, I've forgotten to mention that I didn't substract all
> > > > > peripheral devices. My new calculations (idle, nothing but
> > > > > the big black box under my desk): Linux 137W, Win 114W (20%
> > > > > or 18EUR / 20$ p.a.).
> > > > >
> > > > > It seems I can't disable my onboard WLAN completely and while
> > > > > Win deactivates it because I don't provide drivers, Linux
> > > > > gives it some power although no software is accessing it.
> > > > >
> > > > > By the way: Maximum output while testing with 3DMark 2006:
> > > > > 219W. I wonder why I had to buy a 400W power supply...
> > > >
> > > > Maybe you can power off the wlan with a wireless-utils program,
> > > > or maybe by unloading the kernel module?
> > > >
> > > > Have you set up power management, powersave frequency
> > > > governors?  Have you set up your disk(s) to idle quickly?
> > >
> > > There is no kernel module. I'll play around with modules, configs
> > > and tools later. It's not urgent, it was more like a mystery that
> > > I wanted to solve.
> > >
> > > Yes, powermanagement (aka "PowerNow!") is activated. No, my disks
> > > do not spin down and should not because of the attrition (I hope
> > > that's the right word) that comes with spinning up.
> >
> > [somewhat OT]:
> > Please read this: http://labs.google.com/papers/disk_failures.pdf
> > The damage done to hard drives in spinup/spindown is in the same
> > category of juju as ricer cflags and cloud seeding.  Drive activity
> > and such is *not* an indicator of failure, while there may be some
> > mechanical stress on the disk, but it's not going to cause your
> > drive to fail noticeably earlier.  Spin them down, save the power,
> > and don't listen to fearmongers.[/OT]
> >
> > --
> > Ryan W Sims
>
> Thanks!
>
> I've known that this report exists but have newer actually seen it
> myself. I'm still a bit reluctant because I don't suspect that HDDs
> in Google's server farm spind down as often as mine would.
> Well, I'll just close my eyes and hope for the best when I hear my
> darlings shutting down. ;)
In my experience, a drive is quite a lot more likely to last a long
time when you _do_ spin it down regularly.  The only drive I ever
killed before its time, was set to _not_ spin down accidentally, and
was in a tiny slimline case, and by the time i got back from work and
realized something was wrong, the outside surface of the drive was hot
enough to cook eggs on (or so i'd guess).  Now I make sure my drives
are set to spin down after a few minutes.  Don't think this is gonna
save much for power though.  I actually thought that's what you were
referring to with 'attrition;' that is, it takes just as much power to
spin up the drive as to keep it spinning for a few extra minutes.  

Thanks for the report, I found it very interesting.  
-- 
gentoo-user@g.o mailing list