How to profile Nvidia Bluefield-2 system?

I want to profile the Nvidia Bluefield-2 SmartNIC. I want to record :

  • events, throughput, latency, accelerator use (like RegEx module, Sha-2 module) of an application running on
    Bluefield;

  • packets I/O, PCIe throughput, DRAM activity, hardware unit (like core) utilization;

  • tracing the DOCA libraries that an application on Bluefield use.

I am not sure how to start and if there is any API to access these data. Any pointers are welcome.