How to use NVENV+GPU(GPC) for video encode

Hi,nvidia
Now we have two 4K@30fps cameras and ten 1080P@30fps cameras, so NVENC of jetson agx orin can 't encode those videos. We want to use NVENC+GPU to encode videos, can we do that? And if we can do that, how to do?
Best Regards

Hi,
The implemented software stacks are to use hardware NVENC engines for video encoding. There is no existing CUDA code for video encoding. Your proposal of using GPU for video encoding is feasible, but you would need to implement the CUDA code for the function.

Thanks, DaneLLL
Before NVENC was developed, NVIDIA used Gpus for encoding. Can we learn from the previous driver of GPU before NVENC?

Hi,
There is no existing code and you would need to implement it on your own.

Hi,
What is the coding capability of the GPU on Jetson agx orin? Can it complete the encoding of 2*4K@30fps?

Hi,

It is uncertain since there is no software stack to do profiling. As mentioned in previous comments, we don’t have implementation for H264 encoding on GPU. If you would like to enable it, you would need to implement the software stack and then you can do profiling.

Hi,
Thanks for your support. We wanted to quickly test the coding capabilities of the GPU. But the documentation that we have seen is basically ten years old. Could you give us some advice

Hi,
Since we use NVNEC engines for encoding on Jetson platforms, you would need to implement the CUDA code for encoding and profile the coding capability of GPU.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.