[mythtv-users] Requesting some sample kill a watt meter numbers

Greg Oliver oliver.greg at gmail.com
Tue Jan 18 20:45:02 UTC 2011


On Tue, Jan 18, 2011 at 2:01 PM, Michael T. Dean
<mtdean at thirdcontact.com> wrote:
>  On 01/18/2011 10:48 AM, Joe Hickey wrote:
>>
>> I think the answer to my question is: if I want to substantially
>> reduce that number I have to look toward the atoms,
>
> No.  If you want to substantially reduce the power consumption, you need to
> shut down the system when not in use.
>
>>  and give up on the
>> idea of having commflag jobs complete in a anywhere near real time (or
>> possibly just have it fire up this box when needed).  Something to
>> think about.... I've been looking for a project lately.
>
> Note that a MythTV developer (who isn't currently able to send an e-mail to
> this list--and I'll leave out names to protect the innocent) mentioned in
> IRC, early this morning, that his Mac Mini idles at lower than the 20W that
> many are quoting for their Atom-based systems.  (Note, also, that you can
> easily find the conversation in the MythTV IRC channel logs if you're so
> inclined, or pop into IRC and I'm sure several of us can help you find it or
> give you similar thoughts on it.)
>
> AIUI, the Atom is purposely designed as a low-cost, low-transitor-count
> processor, so most of the power-saving goodies that allow real processors to
> idle at low power usage were removed.  Instead, it's pretty much a processor
> with a small range of power draw (meaning not far off max power draw) and
> marketing only gets away with calling it "low power" because they chopped
> off the headroom--you /can't/ draw high power (but even when your at load,
> you're not accomplishing as much as, say a system with a nice Core 2 Duo
> that may hit higher power draw, but for significantly smaller times--and may
> idle at equivalent or lower power draw...).
>
> Mike
>
> In the interest of full disclosure, I am a confirmed Atom hater.

While I share your same opinion on the Atoms, I have to jump in and
also say that all of the low power capable cpus that throttle down to
the same power usage all exhibit the same issue - and that is they
throttle more than the cpu, and starve other components.  This hampers
high bitrate playback, and renders them useless as well..  Of course,
all of this can be avoided by "upping the power", which makes them
less useful as a mythtv frontend than an atom..  Of course, if you are
a tv viewer who thinks a studder every hour or two is acceptable, then
please, do not read on..

Of course, I am sure the same people who still argue that an ION (v1)
system can handle the normal playback profile will chime in  :)  They
all either have small TVs, or a non-discerning eye  :)

The point is, there is no current low power system that handles all
corner cases, and all viewing requirements.  I truly believe that the
folks that have been doing arm are waaaay further ahead than the x86
architecture will be for a while, and look forward to android pushing
that arch more toward real-time media.  Even if mythtv will not run a
frontend on it, a replacement device that can watch the recordings
with a (as close as can be) replacement gui would be ideal..  The pci
architecture became too complex for simple power management..  The
allowable differing clock speeds for almost every bus in the system
means that without direct access to it all (kernel only), some parts
just will not be capable of keeping up..  I know I'm comparing apples
to footballs here, but this is a specialized area..

I guess my main point is, if you value your tv viewing to be optimal,
either prepare to run your system at full throttle while you watch and
power it of while you do not, or settle for hiccups..  I have
obviously not been doing this near as long as a lot of people, but my
opinion is that when I watch TV, I expect to be able to watch it
without jerks and studders..  I have not been able to accomplish that
without running every system (atom/ions, dual core atom/ions,
phenom2x4's/gt200s, core2quads/gt220s)..

Just my $.02


More information about the mythtv-users mailing list