Thermal sensor data error in AGX Orin

Hi, I’m looking for the thermal data for the AGX Orin devkit, I found the linux thermal zones, but I notice that from zones 2/3/4 the temp value is -256000.
I found this post that seems to have a similar issue but no response for that issue.

I also try tegrastats, and I also see the same value for these three zones (GPU, CV1 and CV2).

03-14-2023 14:18:44 RAM 4892/30589MB (lfb 4515x4MB) SWAP 18/15294MB (cached 0MB) CPU [1%@729,0%@729,0%@729,0%@729,off,off,off,off,off,off,off,off] EMC_FREQ 0%@204 GR3D_FREQ 0%@0 GR3D2_FREQ 0%@0 VIC_FREQ 115 APE 174 CV0@-256C CPU@40.5C Tboard@29C SOC2@37.125C Tdiode@30.25C SOC0@37.531C CV1@-256C GPU@-256C tj@40.5C SOC1@37.781C CV2@-256C VDD_GPU_SOC 1991mW/1991mW VDD_CPU_CV 398mW/398mW VIN_SYS_5V0 2520mW/2520mW NC 0mW/0mW VDDQ_VDD2_1V8AO 301mW/301mW NC 0mW/0mW

I’m using Jetpack 5.1. Is this a known bug?

Hi ivanrodriguezferrandez,

The GPU, CV power rails might be turned off at idle by run time power management. The temperature cannot be read from GPU, CV thermal sensors when the power is off. Attempts to read a sensor with the power off will return error code -256000.

Ok now running something in the GPU yields some data for the temperature of the GPU.

03-15-2023 07:46:22 RAM 5585/30589MB (lfb 3999x4MB) SWAP 18/15294MB (cached 0MB) CPU [0%@1113,0%@1113,100%@1113,0%@1113,off,off,off,off,off,off,off,off] EMC_FREQ 1%@2133 GR3D_FREQ 99%@407 GR3D2_FREQ 99%@0 VIC_FREQ 115 APE 174 CV0@-256C CPU@40.812C Tboard@29C SOC2@37.218C Tdiode@30.25C SOC0@37.531C CV1@-256C GPU@37.25C tj@40.812C SOC1@37.718C CV2@-256C VDD_GPU_SOC 3980mW/2313mW VDD_CPU_CV 796mW/796mW VIN_SYS_5V0 3937mW/2822mW NC 0mW/0mW VDDQ_VDD2_1V8AO 1108mW/503mW NC 0mW/0mW

But why is that the case for only these three sensors? is some kind of power gating for that part of the chip when is not been in use?

That relates to the design of power management.

You could just open Jetson Power GUI to enable GPU usage.
For CV, you should run some computer vision tasks.