[mythtv] Random blockiness that probably isn't CPU or PCI bus bandwidth-related

[Kyle Rose]

Daniel Thor Kristjansson <danielk at mrl.nyu.edu> writes:

> This doesn't really jibe with my experience, unless you are running out 
> of CPU, which shouldn't be the case for a P4 3.2Ghz machine.

Well, the problem is evident in the recording itself, so it may be
that the 1.533 athlon isn't enough.  But after the recording starts,
(which for some reason uses a huge amount of CPU), the thing never
uses more than 3% of the CPU.  As I said, I doubt it's interrupt
latency, or the driver would report buffer overruns, but something
somewhere is losing data only when both capture cards are being used
at once.

> Of course I 
> do compile with extra optimizations, even in debug mode. If it is not 
> the CPU, do the following:

> Make sure there are errors in the stream.
>  Yes - If there are is the signal level too low?
>    Yes - nothing we can do
>    No  - there may be a bug in the ringbuffer 
>  No - there may be bug in ffmpeg

ffmpeg isn't relevant to recording, so it isn't that.  Signal strength
is always >85%, and shouldn't fluctuate when more than one card is
being used; nonetheless, I'll check that because there could be some
kind of wacky interference going on.

Cheers,
Kyle