With respect to use CUDA for video encoding and decoding


When creating an NVR using jetson nano, the video stream is a basic function and you need to be able to do other ai computing, right?
If you use CUDA for jpeg, h264 encoding and decoding in a real-time video stream, is there a lack of gpu resources for other ai computing? Or is it only a small fraction of the performance of jetson nano using CUDA for encoding and decoding?

Thank you.

How to delete this article?

Encoding/decoding is not done on GPU. There are individual hardware encoding/decoding components in Jetson Nano. Please check Tegra X1 SoC Technical Reference Manual:

1 Like