Envirenet

• system: Ubuntu-18.04
• software: Open3D-0.13 + Open3D-ML-torch
• GPU: NVIDIA3070 * 2

Problem

hi, I have use Open3D-ML to train a semantic segment, which use RandLANet, and dataset semantic3D. in my sutation, I have train successful and predict well. but In my pc have two GPU, the train task only use one, how can i config to use two gpu for train.
by the way, I find this problem because I have another computer with only one gpu 3070, but in two PC ,the train cost time is simility. and use nvidia-smi, the only gpu 0 is in use.
thanks someone take a good solution.