How to measure GPUDirect RDMA performance?

Accelerated Computing CUDA CUDA Programming and Performance
1

Is there any performance monitoring tool for GPUDirect RDMA?
I am searching nv_peer_mem/nvidia_peermem documents for monitoring method.
But I cannot found it.

2

I’m not aware of what perf monitoring tools may exist. But you can measure it with networking tools such as those in the perftest package.

3

In my understanding, nv_peer_mem/nvidia-peermem is GPUDirect (not only GPUDirect RDMA) function.

Is threre any tool to monitor gpu read/write performance via nv_peer_mem/nvidia_peermem?