Showreel Performance and suggestions
Answered
Hello :)
Recently we have had a few clients wanting to display up-to 32 cameras on 4K displays around their home and workplace. We have been doing this through Showreels on the NxWitness client but performance is fairly low, even on a high end CPU. NxWitness crashes at times and the CPU sits at 93%. We have it in a cold server room so cooling isn't an issue but no matter how big of a CPU we put in, we cannot get the CPU down. GPU isn't used by NxWitness client.
Now I am aware that NxWitness does not currently support GPU encoding so this is likely the culprit.
Does anyone know of another solution we could use to display the 32 RTSP streams of the camera in a grid? I have been trying "RTSP Camera Viewer", but this is unreliable and the images freeze at times. I have also used a Raspberry Pi image for displaying RTSP feeds in a grid but only up-to 4 feeds. This solution was good as the feeds would restart if they crashed, but I was unable to do more than 6 reliably.
Would we just need to use a bigger CPU or is there other software people are using to display mosaics this large? I see them at malls and corporate buildings so this must be possible.
Any help or suggestions are appreciated :)
-
Hi Joel O'Connor,
Displaying 32 devices on a single screen should not be an issue; below an example of 32 devices on my laptop (i7 CPU + 32 GB RAM) displaying without any issues.
But due to the following;
Does anyone know of another solution we could use to display the 32 RTSP streams of the camera in a grid?
Could it be that these are a single RTSP stream without a low res secondary stream? If that is the case, the client has to display 32 High Res stream which gives a much higher load to the client, and it might crash indeed.
Below an example, with the same content when I force the HiRes stream for all cameras on the layout, the CPU instantly jumps to 100%
So assure there is a secondary stream available, so the system can benefit from the use of the adaptive scaling feature, this will reduce the load on the hardware a lot.
0
Please sign in to leave a comment.
Comments
1 comment