undefined symbol error with PyTorch 1.12 and Torchvision 0.13

Thanks for reporting the unexpected results and we appreciate it a lot.

Describe the Issue

With CUDA 11.3, PyTorch 1.12 and Torchvision 0.13, using MMCV 1.5.3, I get the error:

ImportError: /opt/conda/lib/python3.8/site-packages/mmcv/_ext.cpython-38-x86_64-linux-gnu.so: undefined symbol: _ZNK3c1010TensorImpl36is_contiguous_nondefault_policy_implENS_12MemoryFormatE

Reproduction

1. What command, code, or script did you run?

MMDetection v2.25.0 train.py

1. Did you make any modifications on the code? Did you understand what you have modified?

No

Environment

1. Please run python -c "from mmcv.utils import collect_env; print(collect_env())" to collect necessary environment information and paste it here.

Can’t run this - same error as above

1. You may add addition that may be helpful for locating the problem, such as
   • How you installed PyTorch [e.g., pip, conda, source]: pip install torch==1.12.0+cu113 torchvision==0.13.0+cu113 --extra-index-url https://download.pytorch.org/whl/cu113
   • Other environment variables that may be related (such as $PATH, $LD_LIBRARY_PATH, $PYTHONPATH, etc.)
   • PATH: /opt/amazon/openmpi/bin:/opt/conda/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
   • LD_LIBRARY_PATH: /opt/conda/lib/python3.8/site-packages/smdistributed/dataparallel/lib:/opt/amazon/openmpi/lib/:/opt/amazon/efa/lib/:/opt/conda/lib:/usr/local/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64

Error traceback

root@1a67d2e00d91:/# python -c “from mmcv.utils import collect_env; print(collect_env())” Traceback (most recent call last): File “<string>”, line 1, in <module> File “/opt/conda/lib/python3.8/site-packages/mmcv/utils/env.py”, line 112, in collect_env from mmcv.ops import get_compiler_version, get_compiling_cuda_version File “/opt/conda/lib/python3.8/site-packages/mmcv/ops/init.py”, line 2, in <module> from .active_rotated_filter import active_rotated_filter File “/opt/conda/lib/python3.8/site-packages/mmcv/ops/active_rotated_filter.py”, line 10, in <module> ext_module = ext_loader.load_ext( File “/opt/conda/lib/python3.8/site-packages/mmcv/utils/ext_loader.py”, line 13, in load_ext ext = importlib.import_module(‘mmcv.’ + name) File “/opt/conda/lib/python3.8/importlib/init.py”, line 127, in import_module return _bootstrap._gcd_import(name[level:], package, level) ImportError: /opt/conda/lib/python3.8/site-packages/mmcv/_ext.cpython-38-x86_64-linux-gnu.so: undefined symbol: _ZNK3c1010TensorImpl36is_contiguous_nondefault_policy_implENS_12MemoryFormatE

[zhouzaida]

Hi @austinmw , we have just uploaded the pre-built packages for torch1.12. Enjoy the latest PyTorch and MMCV with the following commands.

pip install -U openmim
mim install -U mmcv-full

[zhouzaida]

Hi @austinmw , we have not provided the pre-built packages for torch1.12 so you need to install mmcv-full from source (https://mmcv.readthedocs.io/en/latest/get_started/build.html#build-on-linux-or-macos). Another way is to install torch1.11 we had provided packages for it. BTW, we will provide pre-built packages for torch1.12 in the next few days.