How can I measure DLA performance by TensorRT?

I’m using Jetson AGX Xavier.

I wonder if there has any API to check the performance of DLA in Xavier, like Network MAC Utilization and HW MAC Utilization?

I see the “Network MAC Utilization” of ResNet50 running on DLA was provided officially in https://github.com/nvdla/hw/tree/nvdlav1/perf, how can i get these profiling data using TensorRT??