[vlc-devel] receiving HVDTS MPEG2 strams: UDP makes artifacts, RTP is OK.

Claudio Allocchio Claudio.Allocchio at garr.it
Tue Dec 15 09:20:29 CET 2009

On Tue, 15 Dec 2009, Marian Ä~NurkoviÄ~M wrote:

> Well, if it works fine with RTP, I wouldn't blame the buffers (unless there's
> some weird bug).

I was metioning them because we are developing another application for 
real time streaming with very low latency, which eats up even more 
bandiwidth (450Mbps), and we had a problem which looks very similar.
The problem was caused there by the large difference in the application 
buffers, and the network buffers used in some libraries. When we bypassed 
the library, and started to do direct socket calls, managing also the 
sockets bufers accordingly, the problem was over.

> More likely, this is yet another UDP-specific problem, like the
> absence of exact payload-type declaration. It means VLC must autodetect the
> payload, which is not always reliable. A common problem is that VLC selects
> wrong demux (PS instead of TS).

> To sum up: UDP is not the way to go, use RTP instead.

As RTP works perfectly (we tested it keeping up and running an HD MPEG2 
stream from USA to Europe for 48 hours), the UDP problem is not a drawback 
in using VLC and HDVTS together: "use RTP" solves it.

It was more an investigation to check if we can understand why it happens 
with UDP: note that it ONLY happens at high bit rates... UDP is ok at 
2Mbps streams.

Also in this case it might be enough for the moment to note the issue, and 
document the solution: "use RTP only when receiving high bit rates 
streaming (20Mbps and above), like for example when receiving HDVTS MPEG2 


best regards!

>    With kind regards,
>           M.
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> http://mailman.videolan.org/listinfo/vlc-devel

Claudio Allocchio             G   A   R   R          Claudio.Allocchio at garr.it
                         Senior Technical Officer
tel: +39 040 3758523      Italian Academic and       G=Claudio; S=Allocchio;
fax: +39 040 3758565        Research Network         P=garr; A=garr; C=it;

            PGP Key: http://www.cert.garr.it/PGP/keys.php3#ca

More information about the vlc-devel mailing list