We have a Xavier sitting in the room running some light code, but the heatsink gets super super hot(70 + C) and the fan doesn’t turn on, unless we manually turn it on from software, even leading to it shutting itself off from overheating. I have checked the fan cable and it’s securely in there.
Has anyone else run into this issue and found a fix besides a script that checks the CPU temp and turns on/off the fan vs custom heat levels
There should be some thermal node that can be simulated as high degree. I am still checking how to clarify this issue.
Could you also dump below value when you said “temperature is high”?
root@jetson-0422418042099:/sys/kernel/debug/bpmp/debug# cd soctherm/
root@jetson-0422418042099:/sys/kernel/debug/bpmp/debug/soctherm# ls
bias_en group_AUX group_CPU group_GPU group_PLLX oc resume sensor_params sensor_params_raw sensors suspend temp_sw_override throttlectl
root@jetson-0422418042099:/sys/kernel/debug/bpmp/debug/soctherm# cat group_*/temp
Also, below method should enable fan,right?
sudo -s
echo 255 > /sys/kernel/debug/tegra_fan/target_pwm
I’ve resolved it by running jetson_clocks.sh which comes with it. Seems to turn on the fan at top RPM, and as I am currently working on it, I’m not inclined to let it get back towards the 85-90 degree C it was at before. Thank you though :)
Xaoc000
I don’t think you should just call this as a fix. If your fan really cannot start to run under normal case (w/o jetson_clock.sh), it may be an issue.
The fan would start to run when the weighted average temperature of CPU,GPU,AUX thermzones (3:3:4) > 50C.
Thus, please share the degree you observed of these 3 during the normal test.
cat /sys/kernel/debug/bpmp/debug/soctherm/group_AUX/temp
cat /sys/kernel/debug/bpmp/debug/soctherm/group_CPU/temp
cat /sys/kernel/debug/bpmp/debug/soctherm/group_GPU/temp
I’m also having this exact same issue.
My system keeps freezing and crashing and I don’t see the fan ever turning on.
root@jetson-0423218011160:/home/nvidia# cat /sys/kernel/debug/bpmp/debug/soctherm/group_AUX/temp
44500
root@jetson-0423218011160:/home/nvidia# cat /sys/kernel/debug/bpmp/debug/soctherm/group_CPU/temp
45000
root@jetson-0423218011160:/home/nvidia# cat /sys/kernel/debug/bpmp/debug/soctherm/group_GPU/temp
44500
@whoishoa, the temperatures you show are 44.5°C or 45°C, so the fan is not expected to turn on (as long as it keeps below 50°C, as said by @WayneWWW). Maybe your freeze and crash is related to something else.
We notice some issue for fan. Please follow up the steps in this thread.
https://devtalk.nvidia.com/default/topic/1048433/jetson-agx-xavier/pcie-and-fan-issue/
The policy of fan running > 50C is for normal case. If you have problem as above thread, please let us know.