[mythtv] CC benchmarks and procedure list ? Was: how would one speed up the channel changing in mythtv ?

[Ed W]

Mark Buechler wrote:
> Hi
>
> On Tue, Oct 21, 2008 at 1:18 AM, Matt Hakim <thepigs at gmail.com 
> <mailto:thepigs at gmail.com>> wrote:
>
>     > Thank you for this detailed answer.  I'll go through it in the
>     next day
>     > or so and comment.
>
>     I agree about the channel change time -- about 5 sec. Then again, I
>     find the whole UI quite slow in 0.21 fixes. I can see the widgets
>     redraw themselves.
>     Your options are: 1. fix it. 2. write your own linux based PVR that's
>     optimised for watching live TV. 3. Buy a PVR.
>     There's no point asking the same question over and over again...
>     _______________________________________________
>     mythtv-dev mailing list
>     mythtv-dev at mythtv.org <mailto:mythtv-dev at mythtv.org>
>     http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
>
>
> Several months ago I submitted a patch which made the various buffers 
> in MythTV configurable with a number (IO multiplier) from 2 to 6 per 
> storage group. Depending on the speed of your system & storage, 
> setting this to a 2 or 3 really speeds up channel changing, but can 
> break very low bandwidth streams like DVB radio if set to 2 for LiveTV.
>
> I believe the ticket is still open but likely the patch won't apply 
> cleanly to trunk.

I agree - I think that most of this delay is fixable just by tuning the 
current myth buffers, but they are largely currently optimised to work 
on maximum hardware rather than optimised for speed.  I also managed to 
get my channel change speeds down to about 50% of normal with some mild 
tweaks, but I have let the patches rot now. 

One of the things that I spotted last time I worked on this is that the 
various semaphores slow things down needlessly.  This might be changed 
now, but for example my system was waiting 1/4 second for the audio to 
pause before moving on with the next step of the channel change... 
Pointless wait state and so you can likely drop out a fair bit of time 
by fixing that kind of thing also.

I think that this is a GOOD question, but it needs to be done 
objectively and without these 500 messages which don't have any code or 
objective questions about how/where to profile.

Good luck

Ed W
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-dev/attachments/20081102/e6b84d9d/attachment.htm