Hello,
I am looking for the Thermal Specifications of the DRIVE AGX. I am tracking the 6 thermo sensors on the board and want to know the limits of each of them.
I’ve found these values for the Jetson AGX but not for the DRIVE AGX.
Thank you very much.
Hi theobwroh,
DRIVE platforms are designed to comply with automotive grade requirements and processes for autonomous driving use cases. The platform and the software stack support thermal shutdown for the Xavier SOCs and the board in order to safeguard the components if they exceed the thresholds for the system.
The DRIVE AGX Platform will not throttle at certain levels as with the Jetson platform. Jetson is designed for the embedded market and doesn’t follow automotive grade requirements. It also supports dynamic voltage frequency scaling besides thermal throttling.
Is there a specific reason you are looking at individual thermal sensors? Perhaps if you explain what you are trying to achieve or debug we can better assist you.
The reason I am asking is, that we are integrating the DRIVE AGX into a closed but ventilated housing. This will increase the temperature of the hardware. I’ve tracked these six different temperatures under load without the housing and now I am doing it inside the closed housing and compare these two states now. I need to know what is an excaptable rise of the temperature. Where are the limits? What is the threshold which should never be exceeded?
Thank you for your answer
As long as you are complying with the operating environment requirements given in the Mechanical & Installation Guide. Then we do not expect you will hit temperature threshold limits.
Hi LukeNV,
I want to know the thermal shutdown thresholds.
The DRIVE AGX is installed in the car and the AV apps is running.
Before the thermal shutdown excutes, I want to safely stop the AV apps.
Thanks.
Hi atsutaka,
Are you seeing thermal shutdowns when complying with the mechanical & installation guide requirements? It may be a good idea to monitor for ambient temperature to determine if some action is needed.
Hi LukeNV,
We are complying with the mechanical & installation guide requirements, but are concerned about the unexpected rise of the temperatures at Xavier and/or dGPU.
For example, in case the fans suddenly do not work.
Hi atsutaka,
If you’d like to monitor temperature changes on the board, you can utilize “tegrastats”. There is some description of the parameters in this topic:
tj temperature should be the highest temperature reading on the board.
Hi LukeNV,
I want to safely stop the AV apps before the thermal shutdown excutes.
Therefore, I need to know the information below.
Could you give me the information on these Q1 and Q2?
Hi LukeNV, any updates on this?
Hello @atsutaka,
sorry for that it is taking long.
we are checking it internally to get your questions answered.
Hi @atsutaka,
to your first question:
please refer to the NvRmGpuDeviceGetTemperature API in the DriveOS documentation on how to get the thermal temperature of the dGPU:
https://docs.nvidia.com/drive/drive_os_5.1.6.1L/nvvib_docs/DRIVE_OS_Linux_SDK_Development_Guide/baggage/group__nvrm__gpu__group.html#gaa28c3eadd8c30bb6099020770321f033
I have attached a code sample for you to have a reference on how to get the dGPU temp.
thermTest.cpp (2.2 KB)
To compile it execute the following:
export GCC=$HOME/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/toolchains/gcc-linaro-7.3.1-2018.05-x86_64_aarch64-linux-gnu/bin/aarch64-linux-gnu-g++
export INCLUDES="-I $HOME/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/drive-t186ref-linux/include/ -I$HOME/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/toolchains/gcc-linaro-7.3.1-2018.05-x86_64_aarch64-linux-gnu/aarch64-linux-gnu/libc/usr/include/"
export LIBS="-lnvrm_gpu -lnvrm -lnvos -L$HOME/nvidia/nvidia_sdk/DRIVE_Software_10.0_Linux_OS_DDPX/DRIVEOS/drive-t186ref-linux/lib-target"
$GCC thermTest.cpp -DNV_SDK_BUILD $INCLUDES $LIBS -o thermTest
