tweaking gst_clock_id_compare_func

[Tim-Philipp Müller]

On Thu, 2012-11-22 at 00:21 -0800, srinidhi wrote:

Hi,

> I ran a gstreamer pipeline using gst-launch and captured profiling data
> using oprofile.
> In the profile report it is seen that gst_clock_id_compare_func is consuming
> a lot of cpu samples .
> Is there any way we can reduce the cpu usage by function
> gst_clock_id_compare_func by tweaking some gst clock parameters ?

What was the pipeline you benchmarked, and what elements were used in
it? Can we reproduce it somehow?

Did you benchmark an application or a gst-launch pipeline?

What version of GStreamer and plugins did you benchmark? On what system?

I think it's surprising to see that function feature in your profile, I
would've thought that in a normal usage scenario there are only a few
clock entries  in the list at any point, and _wait_async() wouldn't be
called much, if at all.

Do you have custom elements that use the clock API and add entries, or
does your application do that?

Cheers 
 -Tim