[Bug 730783] bytereader: Improve the performance to find start code for H.264 and H.265

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=730783
  GStreamer | gstreamer (core) | git

--- Comment #12 from Sungho Bae <baver.bae at lge.com> 2014-05-30 09:24:04 UTC ---
(In reply to comment #10)
> So it's the classic example of "yes, this function is expensive, but instead of
> optimizing that function ... how about calling it less often ? :)"

First of all, thanks to you and Nicolas.

I entirely agree with your opinions.
if the problem is fixed, I expect that it is improved approximately twice.

but, I think the performance of scanning function is also important.
I do unit tests with this patch and got the result as follows:

1) target environments:
- CortexA9 dual 1.0Ghz
- Input Size: H.265 Bitstream ( 9MB, includes 1203 nalus )

2) Result: 
+--------------------+--------+-------+--------+--------+-------------+
| method             | avg.   | std.  | max.   | min.   | enhancement |
+--------------------+--------+-------+--------+--------+-------------+
| origin             | 40.072 | 1.552 | 44.250 | 37.376 | 100.00%     |
| w/o mem idx        | 37.162 | 1.284 | 40.141 | 34.937 | 107.83%     |
| w/o mem idx + neon | 27.890 | 1.284 | 30.891 | 26.201 | 143.67%     |
+--------------------+--------+-------+--------+--------+-------------+

According to the result,
The enhancement of this patch is expected to average 43.67% for scanning start
code.
It addresses that optimizing the scanning function as well as reducing the
number of calling it is important.

In addition, I plan to solve the bug which mentioned by you and Nicolas.
but, I think it should be processed by another thread because the improving
points are differed.

Thanks.

-- 
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.