[mythtv-users] Power Saving Frontend Ideas
rogerheflin at gmail.com
Sun Apr 13 23:55:56 UTC 2008
Rick Hudson wrote:
> > plus puts more stress on the components.
> This one sounds like the filament lamp myth! Those always die on startup
> because it's the most stressful time for them but they do so after their life
> expectency regardless of how many times they have been started in the interim.
> Electronics components have a specified mean /time/ before failure (MTBF) and
> the number of current surges is not a factor provided they operate within
> their spec.
Hard disk will burn out if you turn them on/off often enough. MTBF is a
fictional number that really means nothing, it is calculated usually based on
rated mtbf of the underlying components, and this rating is based on an
assumption of how many startup surges that there are, but the calculation is not
really worth much since the operational data disagrees massively with the
calculated value, note that on hard disks the official MTBF provided by the
various manufactures is as least 10x the MTBF calculated from actual failure
rates, which means that their MTBF is not worth much. Read Google's report on
hard disk failures, and I had experience with running ~2000 hard disks for about
2-3 years, and saw very similar things to what Google's reports has.
And the same is somewhat true of power supplies, the initial in-rush does put
more wear on the power supply that running, usually the power supplies do blow
on start up (if they are going to blow).
And I completely agree with the claim of it taking more power to startup than
run for a long time as being total crap, at worst as your say it might take a
few seconds to use the same amount of power as startup on almost anything
normal people care about.
On a separate note:
On my backend I put in a 80plus power supply that changed the power usage from:
135W idle -> 120 W idle (-15 W)
190W heavy cpu -> 165 W heavy cpu (-25 W)
This is for a backend with 5 hard disks in it, and I run recording on it all
sorts of times and have a number of transcoding jobs on it, so it is not unused
that much. My primary frontend is a MediaMVP box (SD only) that uses a trival
amount of power (<15 watts).
So a decent amount of savings from just a simple change. Given my power cost
of <$.06/kwh the 80plus thing does not pay for itself, but for someone paying
the higher rates it would easily pay for itself.
More information about the mythtv-users