[mythtv-users] MythTV backend running on NVIDIA Ka-el, 35W replaced by

[mythtv]

On Tue, 08 Mar 2011 21:01:47 -0500, Raymond Wagner <raymond at wagnerrp.com>
wrote:
> On 3/8/2011 20:28, mythtv wrote:
>> The primary reason I want to reduce my power footprint right now is
that
>> my combined BE/FE sits below my TV and the fan is noticeably loud.  
> 
> Can you just move this machine to another room?  
 
I have concrete slab on grade so I'd have to run cables through walls
which would mean cutting a hole every 24" = very low WAF and lots of work. 
Unfortunately there's no easy way to move my cable tv line or ethernet so
I'm stuck with my BE/FE near the tv.

>> Viewing content vs idle doesn't make any noticeable difference in
power.
> 
> That E6600 should be dropping multiplier, and dropping the voltage to 
> match.  Unless you're using VDPAU for all your decoding and idling all 
> the time anyway, there should be a noticeable difference between 
> playback and idle.  

It runs at 1.6GHz usually and yes I use VDPAU for all decoding so CPU
usage is supposed to be light.  :) 

>> After my RAID array had a failure I replaced a 3.5" drive with a 2.5"
one
>> and noticed about 5W less power.  So I moved everything except my 1.5TB
>> drive to 2.5" disks as time and budget allowed.
> 
> A 2.5" hard drive may use less than half the power of a 3.5", however 
> three 1TB 2.5" drives will consume more than one 3TB 3.5" drive.  Make 
> sure you have your system set up to allow idle spindown.  An idling 3.5"

> hard drive will consume 4-6W each, but ones that are parked and spun 
> down will be well under 1W.

Good point however my system has one small drive for the OS, one big one
for mythtv media, and two drives in raid1 to store my pictures (instead of
having a separate NAS).  Given my usage reducing the number of drives has
reliability drawbacks that I don't want to make.

>> I'm seriously thinking about experimenting with the PicoPSU 130W as a
>> way to further improve my power efficiency without a large expense.
> 
> The PicoPSU itself is going to be 96-98% efficient, but you still need 
> to get DC in the first place.  Most external switched mode supplies are 
> going to be around 70% efficient, and you can get some real crummy ones 
> closer to 60%.  The linear rectifiers in wall warts are even worse than 
> that.  Unless you're looking to do something fancy, like run DC power 
> throughout your home, look up a computer directly to a UPS, or power it 
> straight off a solar cell, the only thing a PicoPSU is going to allow 
> you to do is move the heat and bulk of a normal ATX PSU outside the
case.

Exactly!  The heat moves outside the case to a passively cooled "brick"
letting my box be cooler inside.  1st order end to end efficiency doesn't
improve much over my 80+ bronze but component temps inside my case should
be lower and that might result in lower leakage for the CPU/GPU.  Whether
that results in lower or higher overall efficiency is an open question.