Found Jetson Benchmark Python script been created 2 years ago, do anyone can tell me is there I can find latest version of Jetson Benchmark script? Here is the old url - NVIDIA-AI-IOT/jetson_benchmarks: Jetson Benchmark (github.com)
I am moving this topic to the Jetson Category - has a better chance of getting an answer.
Supposed the script should still work.
After understanding the script and the trtexec workload, but found there is no way to user to provide the images themselves to do inferencing. Also want to know how the trtexec load the images and do the inferencing? Because trtexec command line didn’t seem like to provide parameter to load the images path (data images), is it using Dummy images? or formula?
Trtexec is a tool for benchmarking tool to profile DNN models and not to perform inference and get output. It can use random inputs for given model and share profiling results.
From the script, found there is formula to calculate the throughput (FPS), would like to know why it’s using 1000? It’s datasets (input images)?
would like to know why it’s using 1000?
It is due to milli seconds to seconds conversion of latency timings.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.