HOST memory out of use and killed #284
Description
I ran pointcloud semantic segmantation rask recently, and the program was killed. Instead of the error of G RAM of of use, I found the extra high usage of host memory. It even took all swap space on my device. Why the train_semseg.py uses so much host memory, and does anyone know how to solve this?
My device equiped with intel i9-13900h and Nvidia RTX 4060 8GB.
