[mythtv] Effect of Multi-rec on channel change times....?

Simon Kenyon simon at koala.ie
Thu Jul 12 14:23:18 UTC 2007

Mark Buechler wrote:
> The problem here is not the channel changing time of the backend but 
> more the buffering time of the frontend. The backend actually changes 
> channels quite fast, in my opinion, considering what it's doing. 
> Myth's frontend at times seems to have trouble estimating bitrate and 
> tends to buffer too much and then read too much, resulting in more 
> prebuffering. On most HD content, however, it does quite well on since 
> the bitrate is much higher. Basically, I think Myth/ffmpeg is 
> estimating bitrate too high is most circumstances.
> But I could be (and probably am) talking out my ass.
> - Mark.
> On 7/12/07, *Simon Kenyon* <simon at koala.ie <mailto:simon at koala.ie>> 
> wrote:
>     Daniel Kristjansson wrote:
>     > STB firmware doesn't have to work with the variety of hardware
>     > that MythTV does; when I set a timing threshold for something
>     > in MythTV I find a value that works on all my machines
>     > then double or triple it (or if all it will cause is a
>     > temporary glitch and it is critical that it is low, I add
>     > 30% and try to get as many people as possible to test it).
>     >
>     >
>     would it be possible to (i.e. will you)?:
>     - get these various timeouts from the database
>     - use the current numbers as the defaults
>     - add some instrumentation to the code to record the various
>     events that
>     fire
>     - optionally provide a UI to allow users to tweak these numbers
>     this way, users with faster hardware could improve switching time.

please accept that i was not complaining in any way about the switching 
speed. it is good enough for me.
what i was reactign to was daniel's statement that he chose timeouts 
that were multiples of that which was really required on his hardware 
and this got me thinking that "one size does not fit all". with the 
various timeouts configurable (be the frontend or backend) then there 
was the potential to tune the system to the hardware.
thats all.

More information about the mythtv-dev mailing list