Hi,
This is how a startup looks like at my BI server.
Originally the video decoding of both cameras was done on the built-in GPU (8th gen. i7, HD 630), GPU 0. Then the GPU 0 was busy doing 3D operations according to the task manager, with at about 30 % duration.
CP AI was always set to GPU but I think the AI inference was also going on the HD 630 or the CPU because the nVidia GPU was showing 0 % in the task manager.
Now, I've set all camera's decoding to nVidia, GPU 1 NvDec. Since then the system memory usage nicely increased as well and nVidia card load displays larger than 0 % in task manager.
However can't recognize where the AI inference occurs. Now none of the GPUs do 3D operations in task manager. However the AI inference is at the same rate.
When YoloV5 6.2 starts, it doesn't tell that the computation will occur on the nVidia card. Previously, after one startup I've seen that. Then the GPU 1's 3D load went for a one-shot 100 % peak and CP AI stopped.
Haven't seen that note since then.
Did anybody have the same issue, or know a way to know/set which GPU is to be used for CP AI? There are many other tasks that can load GPUs, like video decoding, motion detection, etc. So it's pretty hidden I think.