Bug 19113

Summary:	Performance opportunities in liboil
Product:	liboil	Reporter:	Brian Cameron <brian.cameron>
Component:	unknown	Assignee:	David Schleef <ds>
Status:	NEW ---	QA Contact:	David Schleef <ds>
Severity:	enhancement
Priority:	medium	CC:	che
Version:	HEAD
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:

Description Brian Cameron 2008-12-16 11:55:52 UTC

I run totem and play an ogg video, and notice the following performance report using the Sun Studio compiler performance tool:

This is running a 43.49 second test:

- 8.666 seconds - resample_linear_argb_ref in liboil
- 6.545 seconds - yuv420p_to_argb32 in gstffmpegcolorspace
- 6.044 seconds - merge_linear_argb_ref in liboil

So almost 40% of the time seems spent in the two liboil function that are 2 of the top 3 CPU consumers.

It would be good to provide hardware accelerated versions of these two functions I think.

Comment 1 Brian Cameron 2008-12-16 11:58:39 UTC

For reference, I ran the report viewing this video:

http://ia310809.us.archive.org/3/items/whycopyrightdoc/whycopyrightdoc.ogg

Comment 2 Che Kristo 2008-12-17 02:12:27 UTC

I ran a 172.541 second test using sunstudio collect, my top 2 were:

merge_linear_argb_ref (/usr/lib/liboil-0.3.so.0.3.0) with 56.039 seconds
resample_linear_argb_ref (/usr/lib/liboil-0.3.so.0.3.0) with 39.708 seconds

Comment 3 Brian Cameron 2008-12-17 07:37:50 UTC

Note I filed a bug with GStreamer about the performance issue in the gstffmpegcolorspace function.

http://bugzilla.gnome.org/show_bug.cgi?id=564777

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.