we’re testing some technical issues related to hailo-8 temperature issue.
with some system, hailo-8 temp goes over 100’C, we should make thermal solution for this system.
Now we’re just monitoring Hailo-8 temperature after starting stress test in Hailo-integration-tool.
we want to know how many load hailo-8 taking during stress test !!
if possible, i want to know percentage of hailo-8 load.
basically, CPU load is very low, under 10~20% under hailo-8 stress test.
we run hailo monitor but failed. we guess that it’s because stress test under working. right?
any method for us to monitor hailo-8 load under stress test ?
thanks
we’tried based on your advice but not work yet.
we tried this before…
we tested under same hailo condition like docker and others.
we run monitor and run integration…but not response any more…
But we can see monitor with general hailo example like tappas sample normally.
we think that it’s only problem with Hailo-integration-tool…
we tried 4.17, 4.19 version, same result.
Have a look at the log files that are created while the tests are running. They are hidden files. You can view them live using VS Code. They will give you current power, FPS, temperature and the models used.
You can then test the individual networks using HailoRT CLI.
hailortcli run 1.hef
This should give you something like 41 FPS. That is 100% for that model.
You can find the models in /var/hailo_accelerator_integration_tool/
The bigger the number the higher the power usage of that model.