Question about GPU Utilization in nvidia-smi output

Accelerated Computing CUDA CUDA Programming and Performance
1

hello,

I recently reviewed the information on GPU utilization provided in the NVIDIA documentation at Useful nvidia-smi Queries | NVIDIA.

It mentions that GPU utilization is the ratio of active time within a sampling period, with the sampling time varying from 1/6s to 1s depending on the product.

I would like to inquire where I can find detailed information about the precise sampling period for GPU utilization for each specific product.

Thanks in advance.