[Bug 702724] first-class miniobjects/API for mpeg-ts related SI (Service Information)

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=702724
  GStreamer | gst-plugins-bad | git

--- Comment #11 from Edward Hervey <bilboed at gmail.com> 2013-06-23 12:53:32 UTC ---
Silly me, I forgot to report the speedups :)

valgrind --tool=callgrind --trace-children=yes gst-launch-1.0
pushfile:///data/tf1-rip.ts ! <element> ! fakesink

With tsparse:
 * Overall (including registry loading) : 60% faster
 * PSI handling (Called 755 times): 9.5 times faster

With tsdemux:
 * Overall (including registry loading) : almost the same
 * PSI handling *Called 4 times): 60% faster

The numbers are therefore coherent with the expected speedup.

Notes:
 * tsdemux only detects and parses the PAT/PMT/TOT tables. That is expected
(and therefore gets very low gain from all of this).
 * the PSI handling was still parsing/converting all the same chains (via
g_convert_with_iconv) as before. It represents 13% of the overall PSI handling.
So it could be worth moving those to later ... but not that massive.
 * The SI pre-parsing/-detection code (mpegts_packetizer_push_section) costs as
much as the new optimized PSI handling. At a glance it could be made almost 2
times faster (by avoiding usage of lists for example).
 * The "PSI handling" code includes creation of the GstMessage (and
GstStructure to contain the GstMpegTSSection) so is comparing the same final
behaviour and codepath.

-- 
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.