I would like to know how to use docker with Jetson devices. I have two jetson which are nano and xavier nx. I normally use Anaconda software and install python packages in virtual env. But I think it is better to use docker image.
Normally, Jetson come with Cuda package install only, if my remember is correct. So need to find suitable CuDnn and other Nvidia and Python package then install those into Jetson. So It need to find suitable compatibility version. So, I think, using Docker image better option.
I have basic questions regrading this,
Can I use docker image directly when have Jetson OS image and Docker installed environment? Or Do I need to install some other packages like CuDnn or something?
When use Docker, Can I access both GPU and CPU because some python package use CPU and other use GPU?
is Nvidia have single Docker image with all python packages like Pytouch, Tensorflow, Keras, OpenCV, Panda and so on?
Other than adding volume and GPU CLI parameters, what other parameters need to add when run Docker image?