How to get average execution time of CUDA kernel using NSight Systems or NSight Compute...

Can I skip ahead to profile a specific invocation of a specific kernel?...