[mythtv-users] 0.21 buffering not as good as 0.20? (Stuttering...)

[Mark Knecht]

Hi,
   Just an observation here. This is not an apples to apples
comparison as I've changed backend servers in the process of doing
this but since going to 0.21 I'm seeing far more small delays when
viewing material across my network than I ever say using 0.20. Is
anyone else experiencing this sort of issue? Are there any
optimization instructions/ideas that I should pay attention to WRT
this issue?

   Observations:

1) While watching material - none of it 'transcoded', it's all just
the output of the PVR's MPEG encoder - I see and hear little delays
that are probably anywhere from 0.1 to 0.5 seconds long. Stuttering...

2) When I skip a commercial I get a fairly standard start-up that goes
something like this:
- Material starts
- Immediately get a 1 second delay
- Material runs again
- Get a 1/2 second delay
- Material runs
- Get a 1/4 second delay
- Material runs fine from here...

   The new server doesn't seem taxed at all. It's a low-end Semphron
machine using less than 2% CPU when a single frontend is attached and
operating. The hard drive gives approximately 50MB/S using hdparm -tT
as a test.

   The existing frontends have run for 2 years and haven't changed
except for the Myth-0.21 updates. CPU usage is certainly higher on my
0.21 frontend. A 3GHz AMD64 Gentoo box is using 15-20% CPU to run a SD
program. I believe that with 0.20 this was more like 10-12%. Still,
it's not that bad.

   There have been no changes to the network and in fact all the
frontends are dual boot and can play the NetFlix streaming service for
hours without stuttering. However NetFlix's video quality is lower so
I expect that is probably a far lower data rate. Additionally they do
a LOT of local buffering before they start playing.

   Any comments or ideas about how I might improve this?

Thanks,
Mark