MPS support of Jetson Xavier

I am currently using AWS Tesla V100 GPU for my project.

The Tesla V100 GPU follows volta architecture which supports hardware-based Multi-Process Service (MPS).
I was wondering if Jetson Xavier supports the hardware-based MPS under volta architecture.
In case Jetson TX2, it doesn’t support MPS even though the GPU is based on Pascal architecture.

Can anyone from Nvidia confirm if Jetson Xavier supports the hardware-based MPS and nvidia-cuda-mps-control daemon described in the following document?
4.1.1. nvidia-cuda-mps-control


Hi ibaek, Jetson doesn’t support the MPS features or nvidia-smi based tools.

Thank you for the confirmation.

since MPS is not supported on Xavier, is there any performance optimization methods that can be used for the multi-process running scenario on Xavier?