High cpu utilization with raw-video(UYVY) network streaming pipeline

[Amit Pandya]

Hi Nicolas,

Yes, omxh264dec buffers are marked as NO_SHARE.
Though it'll not cause a copy in the payloader plugin, as if refer the
pipeline b) uses "nvvidconv" plugin between "h264 decoder and payloader" for
hw accelerated format conversion(NV12 -> UYVY) using nv hw engines.

nvvidconv maintains output bufferpool to provide converted UYVY format
buffers to payloader  & GST_MEMORY_FLAG_READONLY is used for wrapped memory
of output gst-buffers.   

When "rtpvrawpay" uses mentioned buffers utilize high cpu load.(decode +
format conversion + payload + fakesink). Here only sw copy involved is for
RTP packets creation by payload plugin.  

As mentioned with pipeline a) which is accelerated optimized path (decode +
format conversion + fakesink) consumes very low cpu load, as expected
because no sw copy is involved.






--
View this message in context: http://gstreamer-devel.966125.n4.nabble.com/High-cpu-utilization-with-raw-video-UYVY-network-streaming-pipeline-tp4669784p4669835.html
Sent from the GStreamer-devel mailing list archive at Nabble.com.