Can you run 2 or more Jetson TX2 Developer platforms in parallel to get high/efficient performance when running your model in either CUDA or Tensorflow? If so, are there any setup procedures to be able to do that? Thanks.
Sorry that I don’t catch your question well.
Do you mean you want to run two GPU tasks with different account on TX2 simultaneously?
If yes, it’s possible.
However, please noticed that the GPU resource is limited so the two developers will need to share the resources.
I suppose I mean being able to run two Jetson TX2 Developer kits in parallel to do a single task. Therefore, make your code run theoretically twice as fast with 2 GPUs running. Can you run two GPUs together like that?
We don’t have an API for this.
There are some related SDK for the desktop training tasks but not for the embedded system.
So you will need to implement it on your own.