Hi.I’m trying to run the Nvidia deep learning inference benchmarks from the GitHub repository (GitHub - NVIDIA-AI-IOT/jetson_benchmarks: Jetson Benchmark) on my Jetson Nano. I have followed the instructions: cloned the repository, installed the required packages, and downloaded the models. However, I am currently struggling with the section “Running All Benchmark Models at Once on Jetson Nano.”
I typed the following command in the console:
sudo python3 benchmark.py --all --csv_file_path ~/jetson_benchmarks/benchmark_csv/tx2-nano-benchmarks.csv
–model_dir ~/jetson_benchmarks/models
–jetson_devkit nano
–gpu_freq 921600000 --power_mode 0 --precision fp16
And I received this error message:
Setting Jetson nano in max performance mode
Traceback (most recent call last):
File “benchmark.py”, line 130, in
main()
File “benchmark.py”, line 28, in main
system_check.run_set_clocks_withDVFS()
File “/home/nano/jetson_benchmarks/utils/utilities.py”, line 39, in run_set_clocks_withDVFS
self.set_clocks_withDVFS(frequency=self.gpu_freq, device=‘gpu’)
File “/home/nano/jetson_benchmarks/utils/utilities.py”, line 76, in set_clocks_withDVFS
from_freq = self.read_internal_register(register=freq_register_, device=device)
File “/home/nano/jetson_benchmarks/utils/utilities.py”, line 102, in read_internal_register
reg_read = open(register, “r”)
FileNotFoundError: [Errno 2] No such file or directory: ‘/sys/devices/gpu.0/devfreq/57000000.gpunafll_gpc0/rate’
The directory /sys/devices/gpu.0/devfreq/57000000.gpu does exist, but there is no file called rate in it. The files available in that directory are as follows:
available_frequencies
available_governors
cur_freq
device
governor
max_freq
min_freq
polling_interval
power
subsystem
target_freq
trans_stat
uevent
What can I do to run the benchmark successfully? Any help would be greatly appreciated.