Hi, I have multi live streams to be processed at same time. It is known that the capacity of one GPU cannot support realtime processing for all the streams. From my understanding, the FPS for all of them should decrease in roughly the same percentage to keep holding the streams in a cheap way. Just wanna double-check it is working like this (as my wish).
Also, there are params like live-source in streammux and sync in sink, which seems to be related. In my case, what should I set for those params?
Besides, if I want to decrease the computation cost for each stream. For example, I can do inference for every 10 frames. There is option to set in drop-frame-interval in source, and to set interval in gie. From my understanding, setting drop-frame-interval will be fine if I do not need the redundant frames at all. But if this is the right way to chase for the optimal efficiency? BTW, in my test, both options will cause unbalanced fps for streams.
• Hardware Platform (Jetson / GPU) T4
• DeepStream Version 5.1
• JetPack Version (valid for Jetson only)
• TensorRT Version w/ offical deepstream docker
• NVIDIA GPU Driver Version (valid for GPU only) 450.80.02
• Issue Type( questions, new requirements, bugs) question