cuda 113,
python 3.8,
pytorch 1.11.0,
spconv-plus 2.1.21
import spconv.pytorch
Traceback (most recent call last):
File "", line 1, in
File "/home/suwei/anaconda3/envs/max_voxelnext/lib/python3.8/site-packages/spconv/pytorch/init.py", line 6, in
from spconv.pytorch.core import SparseConvTensor
File "/home/suwei/anaconda3/envs/max_voxelnext/lib/python3.8/site-packages/spconv/pytorch/core.py", line 21, in
from spconv.tools import CUDAKernelTimer
File "/home/suwei/anaconda3/envs/max_voxelnext/lib/python3.8/site-packages/spconv/tools.py", line 16, in
from spconv.cppconstants import CPU_ONLY_BUILD
File "/home/suwei/anaconda3/envs/max_voxelnext/lib/python3.8/site-packages/spconv/cppconstants.py", line 15, in
import spconv.core_cc as _ext
ImportError: arg(): could not convert default argument 'timer: tv::CUDAKernelTimer' in method '<class 'spconv.core_cc.cumm.gemm.main.GemmParams'>.init' into a Python object (type not registered yet?)
I solved this bug by upgrading gcc from 7 to 9，But there is a new bug
Traceback (most recent call last):
File "train.py", line 229, in
main()
File "train.py", line 175, in main
train_model(
File "/home/suwei/suwei_ws/mm/VoxelNeXt/tools/train_utils/train_utils.py", line 173, in train_model
accumulated_iter = train_one_epoch(
File "/home/suwei/suwei_ws/mm/VoxelNeXt/tools/train_utils/train_utils.py", line 56, in train_one_epoch
loss, tb_dict, disp_dict = model_func(model, batch)
File "../pcdet/models/init.py", line 42, in model_func
ret_dict, tb_dict, disp_dict = model(batch_dict)
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "../pcdet/models/detectors/voxelnext.py", line 13, in forward
batch_dict = cur_module(batch_dict)
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "../pcdet/models/backbones_3d/spconv_backbone_voxelnext.py", line 185, in forward
x = self.conv_input(input_sp_tensor)
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/spconv/pytorch/modules.py", line 137, in forward
input = module(input)
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
return forward_call(*input, **kwargs)
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/spconv/pytorch/conv.py", line 487, in forward
out_features = Fsp.implicit_gemm(
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/torch/cuda/amp/autocast_mode.py", line 118, in decorate_fwd
return fwd(*args, **kwargs)
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/spconv/pytorch/functional.py", line 205, in forward
raise e
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/spconv/pytorch/functional.py", line 190, in forward
out, mask_out, mask_width = ops.implicit_gemm(features, filters,
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/spconv/pytorch/ops.py", line 1861, in implicit_gemm
tune_res, _ = CONV.tune_and_cache(
File "/home/suwei/anaconda3/envs/mmvoxel/lib/python3.8/site-packages/spconv/algo.py", line 618, in tune_and_cache
inp = inp.clone()
ValueError: /io/include/tensorview/tensor.h(171)
don't compiled with cuda
You have to specify your Cuda version before compiling cumm and spconv-plus.
export CUMM_CUDA_VERSION="your cuda version"
export CUMM_DISABLE_JIT="1"
export SPCONV_DISABLE_JIT="1"
Now you can rebuild and install cumm and spconv-plus from source. It should work fine.
dvlab-research
