[mythtv-users] How MANY backend threads!?!?!? ***** AUTO-REPLY *****

[Michael T. Dean]

  On 10/13/2010 06:20 PM, Brian Wood wrote:
> On Wednesday, October 13, 2010 04:11:52 pm Steve Smith wrote:
>> This list really has changed. 3 years ago when i first started using
>> myth people were glad to help each other out. Esoteric hardware was
>> seen as a challenge. Stupid remarks like that were not seen.
>> Yes I know my hardware is a bit long in the tooth. I'm not that daft.
>> So what? I'm trying to keep this old hardware going and save some
>> landfill. Quite frankly if a p3 is not fast enough for sd then myth is
>> getting too fat. Perhaps i should look to microsoft for something
>> leaner?
>>
>> Anyone got any sensible suggestions to debug and tweak the system?
> I'd settle for an answer to the original question, How many backend threads should we expect?
>
> I agree a fast P3 *should* be able to handle SD, without concurrent transcoding or commflagging, which should niceable to
> reduce system load.

The original question doesn't make much sense--which is why some people 
(such as me) assumed that Steve was seeing multiple mythbackend processes.

A thread doesn't necessarily take any CPU resources, so measuring the 
number of threads--or expecting the CPU usage be directly correlated 
with the number of threads to the point that reducing the number of 
threads would reduce CPU usage--isn't appropriate.

So, chances are a better question would get better responses.

For example, if the question were, "When recording SDTV at 720x480 to 
MPEG-4 in NUV using my framegrabber, I'm seeing approximately 30-40% CPU 
usage.  Does this seem normal?  Is there anything I can do to reduce the 
CPU usage?" then responses would likely be, "By switching to a hardware 
MPEG encoder, you could reduce CPU usage significantly," or "You could 
record to RTJPEG to trade increased disk usage for decreased processor 
usage."

If the question were, "I just upgraded my system from 0.21 to 0.23 on 
the same distro with the same kernel and same capture card drivers, but 
I'm seeing CPU usage around 30-40% versus the 10-20% I saw in 0.21 when 
recording SDTV at 720x480 to MPEG-4 in NUV using my framegrabber.  Is 
there anything that may have changed within MythTV that causes increases 
processor usage when software-encoding?" then someone would likely have 
said, "Please check your recording profiles, and--especially--the MPEG-4 
encoding options.  There were changes to the default options applied due 
to problems with certain settings."

The best answer to, "How MANY backend threads!?!?!?" is probably, 
"Exactly as many as you're seeing."

Mike