bpa wrote: 
> I don't know what conhost is - it looks like something new since Vista
> and some sort of MS shim layer but since Groovesalad plays OK I don't
> think it is cause of problem.
> 
> Since no CPU blips when using R4 - that seem to indictae that write of
> data tfrom LMS to socketwrapper's input Named Pipe are not happening.  -
> the BBCIplayer plugin has its own write routine so it looks like there
> is some subtle difference between the LMS version and BBCiPlayer but it
> may be due to  Windows Server/VM specific behaviour and handling of IO
> sockets.

psketch, did you say you get a very brief bit of audio from the player
when the stream starts (couple of seconds maybe?), before the
rebuffering occurs?

If so, that means something has got through transcoding. LMS is writing
data to -somewhere-, because of the log lines I mentioned earlier
'Slim::Player::Pipeline::sysread (310) Wrote 32694 bytes to pipeline
writer'. My guess is that the flood of data written to socketwrapper
immediately after the first bit has got through to the player (about
400KB / 30 seconds or so worth of data is written in a very short space
of time - in some ways your server may be -too- good!) somehow
overwhelms the transcoding processes, fills up a socket/pipe buffer or
keeps one of the transcoding processes busy for too long (probably not
the latter if you don't see an initial CPU blip).

If there is some console debug output from socket wrapper, that sounds
like a sensible next step to working out what is going on.


------------------------------------------------------------------------
utgg's Profile: http://forums.slimdevices.com/member.php?userid=40900
View this thread: http://forums.slimdevices.com/showthread.php?t=53229

_______________________________________________
plugins mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/plugins

Reply via email to