Hi

Currently I try to optimize my vMix system for less CPU usage.

What I found so far:
- Playing h264 files: changing video renderer from VMR9 to EVR results in a CPU usage of 2-3%
- Streaming: Enabling hardware encoding for FFMPEG results in a CPU usage of 2-3%
- Recording: Using FFMPEG with codec "MP4 NVENC AAC" results in a CPU usage of 2-3%

I think these values are quite good!

Now I'm looking for some tweaks to optimize the CPU usage of incoming RTSP streams. Every RTSP stream uses about 8%-15% of CPU (depends on bitrate). Because I feed in my camera signals through RTSP (4 to 5 h264 streams), I have a quite high CPU usage only for this task. So I guess there is only CPU decoding in place and no GPU decoding.

I did some googling about this topic and found an interesting overview sheet from Nvidia GPU conference:
http://on-demand.gputech...4_video_surveillance.pdf

They describe how they did Nvidia GPU decoding of RTSP streams and the results are quite impressive. I don't know if something like this is already implemented, not possible in vMix or planned for the future. Would be cool to get some feedback from the developers of vMix about this :-)

Regards,
Patrick