[mythtv-users] Raspberry Pi suitability for MythFrontend

Raymond Wagner raymond at wagnerrp.com
Mon Feb 20 02:41:56 UTC 2012


On 2/19/2012 20:58, Mark Hutchinson wrote:
>
>
> On Sun, Feb 19, 2012 at 5:17 PM, Nick Rout <nick.rout at gmail.com 
> <mailto:nick.rout at gmail.com>> wrote:
>
>     On Mon, Feb 20, 2012 at 2:06 PM, Mark Hutchinson
>     <markhsa at gmail.com <mailto:markhsa at gmail.com>> wrote:
>     > I hope this gets developed.  The amount of 100+ watt frontends
>     replaced by a
>     > 3 watt machine would save serious $ overall. Standardized
>     hardware like the
>     > pi could be just the ticket. One optimized download for myth
>     frontend on
>     > known hardware. Massive advantages.
>     > I would view this as a critical move forward.
>
>     As I think Raymond was explaining, it is not just playback that is
>     needed. Myth themes use up a heap of ram. Ram is cheap on x86/amd64
>     machines, not so on Raspberry Pi etc.
>
>     I think therefore that R pi is unlikely to make a good frontend if you
>     want to run mythfrontend.
>
>     It may work well as a uPnP or xbmc based frontend, in the same way as
>     a WD Live or Popcornhour and other STB's can be used as a front end.
>
>
> OK.  So if XBMC can do it, my can't Myth?  Are the themes that much 
> more intensive and such?
> Not being sarcastic at all, I just really don't know.  I am not a 
> coder.  I am just really interested in saving power and optimized 
> software for standardized hardware is always a good match I would 
> think?  Just a possible opportunity.
> I know it would take off huge if people could get a $25 or so fronted 
> for myth.  Maybe this is not the device, but you will see major 
> acceptance of this device I suspect for many projects.
>
> dd an image onto a flash drive for RPi and off to the races.  It would 
> be huge.

I don't know if XBMC can do it well.  The things I've seen with XBMC 
have said to use the 256MB version, meaning it too wouldn't run all that 
well on 128MB.  The basic issue is that MythTV was never designed to run 
on embedded hardware.  We use PC hardware, where 512MB of memory was 
typical a decade ago, so development has been rather fast and loose with 
memory usage.  MythTV could be tightened up to run a graphical theme at 
1920x1080 with only 256MB of memory, it's just never had to before, and 
will require someone with significant knowledge of the code base 
spending a long time optimizing code.

If this 3W device would be replacing an existing 100W+ frontend, then 
all you have to blame for that is poor purchasing decisions in your 
past.  The Core2 Duo and Athlon II X2 lines topped out at 65W TDP, and 
ran considerably lower at idle.  Modern chipsets don't take much, nor 
does a modest video card, and a dedicated frontend needs no hard drive.  
It's not difficult to get one idling under 40W, and less than double 
that under load.  i3 systems with Intel graphics can be done for well 
under 30W.  You could even go for a Mac Mini, or something with similar 
laptop hardware, that ran under 10W.

Now sure, 30W is still an order of magnitude higher consumption than the 
Pi, but then realize that utility rates in Northern America put power 
cost at somewhere around $1/W/yr.  That's only $30/yr to run that 
comparatively high end i3.  Now consider the fact that it is a dedicated 
frontend.  It does nothing but playback, and has no need to be powered 
up except for playback.  Turn it on for the few hours a day you might be 
watching TV, and then put it in standby or powered off.  Even when you 
account for the fact that those hours it is on will be under higher 
load, and consuming more power, you're still talking something well 
under $10/yr.  There's a point somewhere at which that law of 
diminishing returns kicks in, and that power consumption cost becomes 
irrelevant to the argument.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.mythtv.org/pipermail/mythtv-users/attachments/20120219/a2d4e570/attachment.html 


More information about the mythtv-users mailing list