I’m trying to gather data on the layer network times of different custom models using imagenet-console from jetson inference.
The results that I’m getting fluctuate a lot each time I run the application. When I used the script provided here: https://stackoverflow.com/questions/45779938/permision-denied-when-change-gpu-clock-rate-in-tx1, the layer network times of one of my models became much faster and more consistent. However, one of my other models is still jumping around a lot. How can I fix this?