[mythtv-users] Bad live TV on Ion

[Bill Williamson]

 Tue, Sep 29, 2009 at 7:58 PM, Bill Williamson <bill at bbqninja.com> wrote:
> On Mon, Sep 28, 2009 at 10:24 AM, Kevin Bailey
> <ke-myth at retriever.dyndns.org> wrote:
>> I didn't want to hijack the Ion performance thread for this issue
>> but I'm having trouble playing live TV on my Ion. It seems to play
>> at half speed. Maybe you could say it hesitates every second.
>> Recorded stuff plays fine.
>>
>> It plays that way even after the OSD disappears. SD does a little
>> better than HD.
>>
>> I've tried playing with the other options under playback settings
>> (use video as time base, opengl, etc.) including changing the
>> de-interlacer (mainly the 1x versions).
>>
>> I set-up real-time threads according to the wiki (and re-logged
>> in after adding myself to audio). The console says, "Using realtime
>> priority." The mythtvfrontend process has a priority of 20 and
>> nice of 0. Don't know if this is right.
>>
>> Then I assumed that maybe the data was getting buffered in /tmp
>> and indeed there's a pipe in /tmp called mythtv-media but
>> I don't know why it would be any different for live TV.
>> Since I'm using a USB drive, I thought maybe that was slowing it down.
>> So I tried doubling the amount of RAM (now making it dual channel)
>> and even explicitly mounting /tmp on tmpfs.
>>
>> My other system plays fine, although it has a different, discrete VDPAU video
>> card. The working system uses 180.60, while the Ion uses the new 190.39.
>> (190.25 was no better.) The Ion is pointed at ubuntu intrepid/avenard release testing.
>>
>> Anyone else seeing this ? Anyone have an idea for a setting to tweak ?
>> Or know if there's another buffer for live TV ?
>
>
>
> Have you tried pausing for 10 seconds IMMEDIATELY after starting
> livetv?  I have prebuffering pauses with HD content and a
> geforce9500GT running .21 JYA, but 10 seconds of pausing always fixes
> it.  The only live TV I watch is the occasional sports and thus I
> hadn't thought about it until you mentioned it...
>



Duh, to add, pausing fixed the problem.  I assumed it was a queue
starvation issue.