Unified memory concurrent access

I was under the impression that unified memory concurrent access would be supported on Xavier since it is compute capability 7.2:
https://docs.nvidia.com/cuda/cuda-for-tegra-appnote/index.html#memory-selection
https://devtalk.nvidia.com/default/topic/1042530/jetson-tx2/cpu-operation-is-very-slow-on-memory-allocated-by-cudamallochost-/post/5292788/#5292788

However, when I check the value of concurrentManagedAccess on Xavier using cudaGetDeviceProperties() it has value 0 and when I test concurrent access using test code, it fails. I’m using the latest JetPack 4.1.1.

Will unified memory concurrent access be enabled in a future JetPack for Xavier?

Thanks.

Hi,

We are checking this internally.
Will update information with you later.

Thanks.

Hi,

This feature is not supported on Xavier.
Thanks.

The documentation should probably be updated then with that information.

Thanks for the suggestion.
We will feedback this to our internal team.