zero-copy pipeline test cases

[Sepehr Laal]

I finally found some time to test this today thoroughly. I recompiled the
entire sdk with Cerbero. I updated the binaries and re-linked my
application with it. The strange thing that happened is that, performance
is much worse than before. Not sure if I am doing something wrong here.
Here's my pipeline:

playbin uri="..." video-sink="glupload ! appsink
caps=video/x-raw(memory:GLMemory),width=3840,height=2160,format=BGRA"

I receive OpenGL textures in appsink perfectly fine but cpu usage is oddly
high. The device in question is a Samsung Galaxy S6 (Verizon).

Another strange problem is that, this pipeline also consumes lots of memory
(twice as much as a regular playbin with only appsink as its video-sink) on
Windows (with latest 1.6 64bit binaries).

This made me question the pipeline I am testing with. Am I missing the
right caps?

Other parameters set on my appsink are:

drop=yes
async=yes
qos=yes
sync=yes
max-lateness=GST_SECOND
max-buffers=10

Thanks and as always any pointers and tips are much appreciated!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/gstreamer-android/attachments/20151028/b8f198a5/attachment.html>