Closed vinayakabhatkale closed 5 months ago
Sorry for the trouble, the problem was due to nvcc fatal : Value 'sm_35' is not defined for option 'gpu-architecture'. Checked the allowed value using nvcc --help and made a change in setup.py for NVCC_FLAGS and re-built it. Worked well. Thanks.
Hello, thank you so much for your work, its been really helpful. I am using the latest version of the repo as a submodule and built my docker container. I have installed CUDA and my GPU is accessible. My ubuntu version: 22.04 LTS CUDA Version: 12.2 Nvidia Driver Version: 535.171.04 echo $CPATH result: /usr/local/cuda/include echo Pythonpath: torch points kernel is added
from deepviewaggregation.torch_points3d.trainer import Trainer File "/home/developer/deepviewaggregation/torch_points3d/trainer.py", line 13, in
from torch_points3d.datasets.dataset_factory import instantiate_dataset
File "/home/developer/deepviewaggregation/torch_points3d/datasets/init.py", line 2, in
from .segmentation import
File "/home/developer/deepviewaggregation/torch_points3d/datasets/segmentation/init.py", line 3, in
from .shapenet import ShapeNet, ShapeNetDataset
File "/home/developer/deepviewaggregation/torch_points3d/datasets/segmentation/shapenet.py", line 13, in
from torch_points3d.core.data_transform.grid_transform import SaveOriginalPosId
File "/home/developer/deepviewaggregation/torch_points3d/core/data_transform/init.py", line 5, in
from .transforms import
File "/home/developer/deepviewaggregation/torch_points3d/core/data_transform/transforms.py", line 16, in
from torch_points_kernels.points_cpu import ball_query
File "/home/developer/torch-points-kernels/torch_points_kernels/init.py", line 1, in
from .torchpoints import *
File "/home/developer/torch-points-kernels/torch_points_kernels/torchpoints.py", line 7, in
import torch_points_kernels.points_cpu as tpcpu
ModuleNotFoundError: No module named 'torch_points_kernels.points_cpu'
Please help.