Alex Davies
Reputation: 41
Segmentation fault when importing torch-sparse (installing pytorch-geometric)
I am trying to install pytorch-geometric for a deep-learning project. Torch-sparse is throwing segmentation faults when I attempt to import it (see below). Initially I tried different versions of each required library, as I thought it might be a GPU issue, but I've since tried to simplify by installing cpu-only versions.
Python 3.9.12 (main, Apr 5 2022, 06:56:58) [GCC 7.5.0] :: Anaconda, Inc. on linux Type "help", "copyright", "credits" or "license" for more information. >>> import torch >>> import torch_scatter >>> import torch_cluster >>> import torch_sparse Segmentation fault (core dumped)
And the same issue, presumably due to torch_sparse, when importing pytorch_geometric:
Python 3.9.12 (main, Apr 5 2022, 06:56:58) [GCC 7.5.0] :: Anaconda, Inc. on linux Type "help", "copyright", "credits" or "license" for more information. >>> import torch_geometric Segmentation fault (core dumped)
I'm on an Ubuntu distribution:
Distributor ID: Ubuntu Description: Ubuntu 22.04.1 LTS Release: 22.04 Codename: jammy
Here's my (lightweight for DL) conda installs:
# Name Version Build Channel _libgcc_mutex 0.1 main _openmp_mutex 5.1 1_gnu blas 1.0 mkl brotlipy 0.7.0 py310h7f8727e_1002 bzip2 1.0.8 h7b6447c_0 ca-certificates 2022.07.19 h06a4308_0 certifi 2022.9.24 py310h06a4308_0 cffi 1.15.1 py310h74dc2b5_0 charset-normalizer 2.0.4 pyhd3eb1b0_0 cpuonly 2.0 0 pytorch cryptography 37.0.1 py310h9ce1e76_0 fftw 3.3.9 h27cfd23_1 idna 3.4 py310h06a4308_0 intel-openmp 2021.4.0 h06a4308_3561 jinja2 3.0.3 pyhd3eb1b0_0 joblib 1.1.1 py310h06a4308_0 ld_impl_linux-64 2.38 h1181459_1 libffi 3.3 he6710b0_2 libgcc-ng 11.2.0 h1234567_1 libgfortran-ng 11.2.0 h00389a5_1 libgfortran5 11.2.0 h1234567_1 libgomp 11.2.0 h1234567_1 libstdcxx-ng 11.2.0 h1234567_1 libuuid 1.0.3 h7f8727e_2 markupsafe 2.1.1 py310h7f8727e_0 mkl 2021.4.0 h06a4308_640 mkl-service 2.4.0 py310h7f8727e_0 mkl_fft 1.3.1 py310hd6ae3a3_0 mkl_random 1.2.2 py310h00e6091_0 ncurses 6.3 h5eee18b_3 numpy 1.23.3 py310hd5efca6_0 numpy-base 1.23.3 py310h8e6c178_0 openssl 1.1.1q h7f8727e_0 pip 22.2.2 py310h06a4308_0 pycparser 2.21 pyhd3eb1b0_0 pyg 2.1.0 py310_torch_1.12.0_cpu pyg pyopenssl 22.0.0 pyhd3eb1b0_0 pyparsing 3.0.9 py310h06a4308_0 pysocks 1.7.1 py310h06a4308_0 python 3.10.6 haa1d7c7_0 pytorch 1.12.1 py3.10_cpu_0 pytorch pytorch-cluster 1.6.0 py310_torch_1.12.0_cpu pyg pytorch-mutex 1.0 cpu pytorch pytorch-scatter 2.0.9 py310_torch_1.12.0_cpu pyg pytorch-sparse 0.6.15 py310_torch_1.12.0_cpu pyg readline 8.1.2 h7f8727e_1 requests 2.28.1 py310h06a4308_0 scikit-learn 1.1.2 py310h6a678d5_0 scipy 1.9.1 py310hd5efca6_0 setuptools 63.4.1 py310h06a4308_0 six 1.16.0 pyhd3eb1b0_1 sqlite 3.39.3 h5082296_0 threadpoolctl 2.2.0 pyh0d69192_0 tk 8.6.12 h1ccaba5_0 tqdm 4.64.1 py310h06a4308_0 typing_extensions 4.3.0 py310h06a4308_0 tzdata 2022e h04d1e81_0 urllib3 1.26.12 py310h06a4308_0 wheel 0.37.1 pyhd3eb1b0_0 xz 5.2.6 h5eee18b_0 zlib 1.2.13 h5eee18b_0
Any help would be greatly appreciated!
Upvotes: 4
Views: 2598
Answers (1)
Alex Davies
Reputation: 41
I've found a combination of packages that works for me - hopefully someone else will have this issue at some point and be able to reproduce the steps from me talking to myself here. The full process for getting stuff working was:
- Fresh conda environment with forced Python=3.9 (
conda create -n ENVNAME python=3.9) - Activate that environment
- Install basic python packages (
conda install numpy pandas matplotlib scikit-learn) - Check CUDA version if working with a GPU (
nvidia-smiin terminal prints these details for NVIDIA cards) - Install Pytorch using their suggested conda command (
conda install pytorch torchvision torchaudio cudatoolkit=CUDA_VERSION -c pytorch -c conda-forge). This had to go through the env solving process on my machine. - Install pytorch geometric (or just torch sparse if that's all you need) with
conda install pyg -c pyg. Again this had a solving process. - Check that
torch_sparseimports without fault
Here's the conda list for this working combination of packages:
# Name Version Build Channel _libgcc_mutex 0.1 main _openmp_mutex 5.1 1_gnu blas 1.0 mkl bottleneck 1.3.5 py39h7deecbd_0 brotli 1.0.9 h5eee18b_7 brotli-bin 1.0.9 h5eee18b_7 brotlipy 0.7.0 py39hb9d737c_1004 conda-forge bzip2 1.0.8 h7f98852_4 conda-forge ca-certificates 2022.9.24 ha878542_0 conda-forge certifi 2022.9.24 py39h06a4308_0 cffi 1.14.6 py39he32792d_0 conda-forge charset-normalizer 2.1.1 pyhd8ed1ab_0 conda-forge cryptography 37.0.2 py39hd97740a_0 conda-forge cudatoolkit 11.6.0 hecad31d_10 conda-forge cycler 0.11.0 pyhd3eb1b0_0 dbus 1.13.18 hb2f20db_0 expat 2.4.9 h6a678d5_0 ffmpeg 4.3 hf484d3e_0 pytorch fftw 3.3.9 h27cfd23_1 fontconfig 2.13.1 h6c09931_0 fonttools 4.25.0 pyhd3eb1b0_0 freetype 2.11.0 h70c0345_0 giflib 5.2.1 h7b6447c_0 glib 2.69.1 h4ff587b_1 gmp 6.2.1 h58526e2_0 conda-forge gnutls 3.6.13 h85f3911_1 conda-forge gst-plugins-base 1.14.0 h8213a91_2 gstreamer 1.14.0 h28cd5cc_2 icu 58.2 he6710b0_3 idna 3.4 pyhd8ed1ab_0 conda-forge intel-openmp 2021.4.0 h06a4308_3561 jinja2 3.0.3 pyhd3eb1b0_0 joblib 1.1.1 py39h06a4308_0 jpeg 9e h7f8727e_0 kiwisolver 1.4.2 py39h295c915_0 krb5 1.19.2 hac12032_0 lame 3.100 h7f98852_1001 conda-forge lcms2 2.12 h3be6417_0 ld_impl_linux-64 2.38 h1181459_1 lerc 3.0 h295c915_0 libbrotlicommon 1.0.9 h5eee18b_7 libbrotlidec 1.0.9 h5eee18b_7 libbrotlienc 1.0.9 h5eee18b_7 libclang 10.0.1 default_hb85057a_2 libdeflate 1.8 h7f8727e_5 libedit 3.1.20210910 h7f8727e_0 libevent 2.1.12 h8f2d780_0 libffi 3.3 he6710b0_2 libgcc-ng 11.2.0 h1234567_1 libgfortran-ng 11.2.0 h00389a5_1 libgfortran5 11.2.0 h1234567_1 libgomp 11.2.0 h1234567_1 libiconv 1.17 h166bdaf_0 conda-forge libllvm10 10.0.1 hbcb73fb_5 libpng 1.6.37 hbc83047_0 libpq 12.9 h16c4e8d_3 libstdcxx-ng 11.2.0 h1234567_1 libtiff 4.4.0 hecacb30_0 libuuid 1.0.3 h7f8727e_2 libwebp 1.2.4 h11a3e52_0 libwebp-base 1.2.4 h5eee18b_0 libxcb 1.15 h7f8727e_0 libxkbcommon 1.0.1 hfa300c1_0 libxml2 2.9.14 h74e7548_0 libxslt 1.1.35 h4e12654_0 lz4-c 1.9.3 h295c915_1 markupsafe 2.1.1 py39h7f8727e_0 matplotlib 3.5.2 py39h06a4308_0 matplotlib-base 3.5.2 py39hf590b9c_0 mkl 2021.4.0 h06a4308_640 mkl-service 2.4.0 py39h7f8727e_0 mkl_fft 1.3.1 py39hd3c417c_0 mkl_random 1.2.2 py39h51133e4_0 munkres 1.1.4 py_0 ncurses 6.3 h5eee18b_3 nettle 3.6 he412f7d_0 conda-forge nspr 4.33 h295c915_0 nss 3.74 h0370c37_0 numexpr 2.8.3 py39h807cd23_0 numpy 1.23.3 py39h14f4228_0 numpy-base 1.23.3 py39h31eccc5_0 openh264 2.1.1 h780b84a_0 conda-forge openssl 1.1.1q h7f8727e_0 packaging 21.3 pyhd3eb1b0_0 pandas 1.4.4 py39h6a678d5_0 pcre 8.45 h295c915_0 pillow 9.2.0 py39hace64e9_1 pip 22.2.2 py39h06a4308_0 ply 3.11 py39h06a4308_0 pycparser 2.21 pyhd8ed1ab_0 conda-forge pyg 2.1.0 py39_torch_1.12.0_cu116 pyg pyopenssl 22.0.0 pyhd8ed1ab_1 conda-forge pyparsing 3.0.9 py39h06a4308_0 pyqt 5.15.7 py39h6a678d5_1 pyqt5-sip 12.11.0 py39h6a678d5_1 pysocks 1.7.1 pyha2e5f31_6 conda-forge python 3.9.13 haa1d7c7_2 python-dateutil 2.8.2 pyhd3eb1b0_0 python_abi 3.9 2_cp39 conda-forge pytorch 1.12.1 py3.9_cuda11.6_cudnn8.3.2_0 pytorch pytorch-cluster 1.6.0 py39_torch_1.12.0_cu116 pyg pytorch-mutex 1.0 cuda pytorch pytorch-scatter 2.0.9 py39_torch_1.12.0_cu116 pyg pytorch-sparse 0.6.15 py39_torch_1.12.0_cu116 pyg pytz 2022.1 py39h06a4308_0 qt-main 5.15.2 h327a75a_7 qt-webengine 5.15.9 hd2b0992_4 qtwebkit 5.212 h4eab89a_4 readline 8.2 h5eee18b_0 requests 2.28.1 pyhd8ed1ab_1 conda-forge scikit-learn 1.1.2 py39h6a678d5_0 scipy 1.9.1 py39h14f4228_0 setuptools 63.4.1 py39h06a4308_0 sip 6.6.2 py39h6a678d5_0 six 1.16.0 pyhd3eb1b0_1 sqlite 3.39.3 h5082296_0 threadpoolctl 2.2.0 pyh0d69192_0 tk 8.6.12 h1ccaba5_0 toml 0.10.2 pyhd3eb1b0_0 torchaudio 0.12.1 py39_cu116 pytorch torchvision 0.13.1 py39_cu116 pytorch tornado 6.2 py39h5eee18b_0 tqdm 4.64.1 py39h06a4308_0 typing_extensions 4.4.0 pyha770c72_0 conda-forge tzdata 2022e h04d1e81_0 urllib3 1.26.11 pyhd8ed1ab_0 conda-forge wheel 0.37.1 pyhd3eb1b0_0 xz 5.2.6 h5eee18b_0 zlib 1.2.13 h5eee18b_0 zstd 1.5.2 ha4553b6_0
Upvotes: 0
Related Questions
- PyTorch torch_sparse installation without CUDA
- Installation problem with PyTorch's Geometric. "torch-scatter" produces an error with exit status 1
- Pytorch AssertionError: Torch not compiled with CUDA enabled
- "[WinError 127] The specified procedure could not be found" for torch_geometric
- Incompletable PyTorch with any CUDA version (module 'torch' has no attribute 'cuda')
- Installing PyTorch via Conda
- Failed to Build Torch-Scatter in Pytorch Geometry
- AssertionError in torch_geometric.nn.GATConv
- Importing pytorch geometric results in an error message
- Import error when importing torch