[mythtv-users] Backend hardware advice - Power usage - RESULTS

dave johnson DJ4904 at hotmail.com
Sat Aug 5 09:27:15 UTC 2006


brad, you didn't follow my point... "in use" meant, "in use at typical idle workload"

for instance, the raid system in my storage box is pretty much going at a decent enough clip that there is a measurable difference between "idle" (no disk io) and "idle workload" where there is cron jobs being fired off, zoe (http://zoe.nu) checking email and indexing it, and a few other systems doing IO on their iSCSI disks just as part of their "idle" work.

Soo..... if you read carefully, the laptop in question (the 266MHz) if running completely "idle workload" is at around 20-25% cpu.  I don't really care about the performance of the system since it's just a domain controller and a whole bunch of other services running, but it certainly never runs at the 3-5% cpu usage that it would if i had booted it up without any services loaded.

bottom line is that i'm trying to measure "real world" usage, not artificially "guessed" usage.

regardless, i hadn't though the laptop used that much power so if the Goede system works out with plenty of spare cpu overhead, i think i'll load VMWare Server 1.0 on it and try running the Windows domain as a virtual machine and see where that puts me.

-=dave

----- Original Message ----- 
From: "Brad Templeton" <brad+myth at templetons.com>
To: "Discussion about mythtv" <mythtv-users at mythtv.org>
Sent: Saturday, August 05, 2006 12:15 AM
Subject: Re: [mythtv-users] Backend hardware advice - Power usage - RESULTS


On Fri, Aug 04, 2006 at 11:19:30PM -0700, dave johnson wrote:
> after reading your response on the laptop, i realized an oversight on my part... i hadn't waited for the flourescent backlight to shut off on the laptops.  i went back and measured them and they were each about only about 5w lower, at 20w and 29w respectively.  remember, these measurements weren't "at idle" (what would be the use ?) but "in use" and these babies are running Windows 2003 and XP respectively.  Hell, the W2k one runs at approx 20% cpu usage just doing nothing at all really (266Mhz doesn't give a lot of processing overhead for today's OSs).  although, the Tecra i don't quite get as it runs a lower cpu usage and isn't doing much.  perhaps it's just that these old laptops were always power mongers and only recently since the wave of super-efficient power manager chips became standard fare did they get efficient.   i know that my old apple 3400 lasted about an hour on a full charge while my iBook lasts about 3, and on a battery 1/2 the size.


No, actually idle is one of the more interesting power measurements if we're
actually aiming to calculate the power cost.  Most machine sit idle most of the
time and do intense work for a small portion.  Myth machines only do intense
work when playing video, recording (if they have software encoder), commercial
flagging and transcoding.   (Some people of course may be doing all these things
all day.)

Myth machines, and linux servers, tend to be on 24/7.  Because 1 watt 24/7 costs $1 or
more per year, the idle power, plus a bit more, is a good measure of cost.

Today's CPUs will burn 70 watts or more in full speed, but they all know to
step themselves down when idle.  But those older slower chips do take less
power.

_______________________________________________
mythtv-users mailing list
mythtv-users at mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users



More information about the mythtv-users mailing list