[Bug 702380] [basertppayload] Provide a GstBufferPool for output rtp buffers

GStreamer (bugzilla.gnome.org) bugzilla at gnome.org
Mon Jun 17 02:55:27 PDT 2013


https://bugzilla.gnome.org/show_bug.cgi?id=702380
  GStreamer | gst-plugins-base | unspecified

--- Comment #7 from Wim Taymans <wim.taymans at gmail.com> 2013-06-17 09:55:22 UTC ---
(In reply to comment #6)
> (In reply to comment #5)
> > Ok, when making sure we never completely empty the pool (and need to read the
> > socket) we can get somewhat faster (but it jitters, sometimes slower):
> > 
> > *** total 0:00:00.960529176 - average 0:00:00.000000480  - Done creating
> > 2000000 buffers
> > *** total 0:00:00.671104842 - average 0:00:00.000000335  - Done creating
> > 2000000 buffers
> 
> That looks more realistic than the first one (because you were mostly creating
> buffers all the time instead of re-using, so it obviously adds more overhead).

No, it was writing and reading from the socket to signal pool full/empty, only
one buffer was allocated.

> 
> It does make it 45% faster, which seems on par with the overhead I was seeing
> when profiling rtpmp2tpay for example.

Here's another typical run, now I added (2nd line) simply doing
gst_buffer_new() without memory:

*** total 0:00:01.250591428 - average 0:00:00.000000625  - Done creating
2000000 buffers
*** total 0:00:00.397951348 - average 0:00:00.000000198  - Done creating
2000000 buffers
*** total 0:00:01.017563303 - average 0:00:00.000000508  - Done creating
2000000 buffers

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.


More information about the gstreamer-bugs mailing list