Remove the Fan and run deep learning


I have a Robocar which is currently driven by a Jetson TX2 devkit, as shown in the picture below

As you can see, I removed the heatsink from TX2 to reduce the weight so the car can run faster in race. I am replacing the TX2 with an Xavier NX devkit for its smaller formfactor and multiple USB ports. I would like to remove the fan on the Xavier NX devkit also. During race, I run deep learning YOLOv3 (input image 640x360) on GPU and particle filter to process 2D Lidar data (1000 points 40 frame per second) on CPU. I also ran some ROS nodes. The car speed can reach 20mph but a single race run last for only 5 seconds. Between races, I do not need to run DL or heavy computation, just run some Linux command line operations

My questions is if it is safe to remove the fan?

hello AutoCar,

this should be okay,

but, please bear in mind there’re thermal throttling, software clock throttling and hardware thermal shutdown.

please also check developer guide for reference, Power Management for Jetson Xavier NX and Jetson AGX Xavier Series Devices.
you may also check device tree for thermal zones.

Thank you for the information.