[vlc] Re: measuring encoding delay

[Mark Moriarty]

I build it under Cygwin, no problem -- the Cygwin environment is "touchy" to
first set up, but fine once it's running.
Have to figure out the right hooks would be to insert time calls -- VLC,
under the hood, implements calls to microsecond-level clocks, just need to
find the right points to add the calls to the time functions, with a
counter. 
(the mtime structures are used throughout VLC to obtain time for PTS and the
like)

-----Original Message-----
From: vlc-bounce at videolan.org [mailto:vlc-bounce at videolan.org] On Behalf Of
Marcelo Manzato
Sent: Wednesday, April 05, 2006 1:52 PM
To: vlc at videolan.org
Subject: [vlc] Re: measuring encoding delay

Hi Mark, 

--- Mark Moriarty <mfmbusiness at earthlink.net> wrote:

> Interesting question.  By "encoding time" do you want to measure the 
> full time from when a frame is first sent to be encoded, to the point 
> where it has been encapsulated and is ready for transmission?

Exactly!!

> Might want to try a debug build, calls to the low-level time 
> functions, let it run for say 1000 frames, then take the average.

I'm using vlc under windows in order to capture and encoding the data. I'm
not being able to compile the source for win32, either by cygwin or
cross-compilation... nevertheless, I'm afraid unix time functions like
time() and clock() will not work under windows. Is there a way to debug the
available windows executable version of vlc without having to add time
functions into the source?

Thanks for your help

Regards
Marcelo Manzato


> -----Original Message-----
> From: vlc-bounce at videolan.org
> [mailto:vlc-bounce at videolan.org] On Behalf Of Marcelo Manzato
> Sent: Tuesday, April 04, 2006 3:46 PM
> To: vlc at videolan.org
> Subject: [vlc] Re: measuring encoding delay
> 
> Hi Simon, thanks for your answer!
> 
> --- Simon Hailstone <simon.hailstone at gmail.com>
> wrote:
> 
> > For a really simple experimental approach, and if
> millisecond accuracy
> > isn't a priority, how about pointing your webcam
> at a clock?  View the
> > resulting stream on another PC, and note the
> difference between
> > "wall-time" and the time displayed on the
> stream-receiving PC.
> 
> I have already thought about that, however, in this case I will be 
> measuring the transmission and the decoding time as well. I want a way 
> to measure just the encoding...
>  
> > Alternatively, start up vlc, begin
> > encoding/streaming, and use
> > Ethereal to observe the time that elapses between
> initialisation and
> > the generation of the first packet with video in
> it's payload...
> 
> It seems to be a good solution, however, I didn´t understand how I 
> will capture the initialisation time... Have you thought about that?
> 
> Regards
> Marcelo Manzato
> 
> 
> 
> 		
>
_______________________________________________________
> Abra sua conta no Yahoo! Mail: 1GB de espaço, alertas de e-mail no 
> celular e anti-spam realmente eficaz.
> http://br.info.mail.yahoo.com/
> 
> --
> This is the vlc mailing-list, see
> http://www.videolan.org/vlc/ To
> unsubscribe, please read
> http://www.videolan.org/support/lists.html
> 
> --
> This is the vlc mailing-list, see
> http://www.videolan.org/vlc/
> To unsubscribe, please read
> http://www.videolan.org/support/lists.html
> 
> 



		
_______________________________________________________
Abra sua conta no Yahoo! Mail: 1GB de espaço, alertas de e-mail no celular e
anti-spam realmente eficaz. 
http://br.info.mail.yahoo.com/

--
This is the vlc mailing-list, see http://www.videolan.org/vlc/ To
unsubscribe, please read http://www.videolan.org/support/lists.html

-- 
This is the vlc mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://www.videolan.org/support/lists.html