🐛 [Bug] undefined symbol: _ZNK3c104Type14isSubtypeOfExtERKSt10shared_ptrIS0_EPSo

[ikergarcia1996]

Bug Description

When I install TensorRT and Torch-TensorRT using pip I encounter the error:

import torch_tensorrt

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/iker/envs/tensorrt/lib/python3.9/site-packages/torch_tensorrt/__init__.py", line 11, in <module>
    from torch_tensorrt._compile import *
  File "/iker/envs/tensorrt/lib/python3.9/site-packages/torch_tensorrt/_compile.py", line 2, in <module>
    from torch_tensorrt import _enums
  File "/iker/envs/tensorrt/lib/python3.9/site-packages/torch_tensorrt/_enums.py", line 1, in <module>
    from torch_tensorrt._C import dtype, DeviceType, EngineCapability, TensorFormat
ImportError: /iker/envs/tensorrt/lib/python3.9/site-packages/torch_tensorrt/lib/libtorchtrt.so: undefined symbol: _ZNK3c104Type14isSubtypeOfExtERKSt10shared_ptrIS0_EPSo

To Reproduce

Steps to reproduce the behavior:

Python 3.9.7, A30 GPU with 470.82.01 driver version. Cuda version: cuda_11.3.r11.3. Cudnn: 8.2

I run the following command in a clean Python environment:

pip3 install --upgrade setuptools pip
pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
pip3 install --upgrade nvidia-pyindex
pip3 install --upgrade nvidia-tensorrt
pip3 install --upgrade torch-tensorrt -f https://github.com/NVIDIA/Torch-TensorRT/releases

export LD_LIBRARY_PATH="/iker/envs/tensorrt/lib/python3.9/site-packages/tensorrt/"

I can import tensorRT successfully

python3 -c "import tensorrt; print(tensorrt.__version__)"
8.4.0.6

Environment

aiohttp==3.8.1
aiosignal==1.2.0
async-timeout==4.0.2
attrs==21.4.0
certifi==2021.10.8
charset-normalizer==2.0.12
click==8.1.2
datasets==2.1.0
dill==0.3.4
filelock==3.6.0
frozenlist==1.3.0
fsspec==2022.3.0
huggingface-hub==0.5.1
idna==3.3
joblib==1.1.0
multidict==6.0.2
multiprocess==0.70.12.2
numpy==1.22.3
nvidia-cublas-cu11==2022.3.24
nvidia-cublas-cu116==11.9.2.110
nvidia-cuda-runtime-cu11==2021.12.20
nvidia-cuda-runtime-cu116==11.6.55
nvidia-cudnn-cu11==2022.4.2
nvidia-cudnn-cu115==8.3.3.40
nvidia-pyindex==1.0.9
nvidia-tensorrt==8.4.0.6
packaging==21.3
pandas==1.4.2
Pillow==9.1.0
pyarrow==7.0.0
pyparsing==3.0.8
python-dateutil==2.8.2
pytz==2022.1
PyYAML==6.0
regex==2022.4.24
requests==2.27.1
responses==0.18.0
sacremoses==0.0.49
six==1.16.0
tokenizers==0.12.1
torch==1.11.0+cu113
torch-tensorrt==1.0.0
torchaudio==0.11.0+cu113
torchvision==0.12.0+cu113
tqdm==4.64.0
transformers==4.18.0
typing_extensions==4.2.0
urllib3==1.26.9
xxhash==3.0.0
yarl==1.7.2

[narendasan]

This is probably because you are using PyTorch 1.11 with Torch-TensorRT 1.0.0 which targets PyTorch 1.10. Our 1.1.0 release is coming soon. In the meantime master supports 1.11

[ikergarcia1996]

Problem solved by installing PyTorch 1.10, thank you!!