[question] About --num-gpus.

[leeyunhome]

Hello,

Thank you for your good writing.

$ python train.py --config-file configs/lp_faster_rcnn_R_50_FPN_3x.yaml --num-gpus 2 I don't think num-gpus means the number of CUDA cores, does it mean the number of physical graphics cards?

I don't know what exactly it means.

I use jetson xavier nx and the specs are as follows:

GPU	384-core NVIDIA Volta™ GPU with 48 Tensor Cores

In case of xavier nx, how should I set num-gpus?

Starting training from iteration 0 /home/hodu/.local/lib/python3.6/site-packages/detectron2/modeling/roi_heads/fast_rcnn.py:217: UserWarning: This overload of nonzero is deprecated: nonzero() Consider using one of the following signatures instead: nonzero(*, bool as_tuple) (Triggered internally at ../torch/csrc/utils/python_arg_parser.cpp:882.) num_fg = fg_inds.nonzero().numel() Killed hodu@hodu-desktop:~/coding/GitHub/detectron2-licenseplates$ python3 train.py --config-file configs/lp_faster_rcnn_R_50_FPN_3x.yaml --num-gpus 1

As above, I started with num-gpus given 1, and it ended with the message'Killed'.

How do I get it to work? I would appreciate it if you could give me directions.

Thank you.