Segmentation fault
bio-rat opened this issue · 9 comments
Hi everyone, I just did a fresh install of DiffDock on my local machine and I encountered this error after trying to run the example inference:
python -m inference --protein_ligand_csv data/protein_ligand_example_csv.csv --out_dir results/user_predictions_small --inference_steps 20 --samples_per_complex 40 --batch_size 10 --actual_steps 18 --no_final_step_noise
the only output:
Segmentation fault
Hi everyone, I just did a fresh install of DiffDock on my local machine and I encountered this error after trying to run the example inference:
python -m inference --protein_ligand_csv data/protein_ligand_example_csv.csv --out_dir results/user_predictions_small --inference_steps 20 --samples_per_complex 40 --batch_size 10 --actual_steps 18 --no_final_step_noisethe only output:
Segmentation fault
I have encountered the same error did you found any solution for it.
I have encountered the same error did you found any solution for it.
The same phenomenon is occurring to me.
I thought it was something to do with the versions of cuda, pytorch, and torch-geometric, so I tried everything, but could not resolve it successfully.
Have you found the solution? Please share it. Thank you very much.
I didn't remember what I did exactly but I followed this #180 (comment)
I fixed it by using:
conda install pytorch==1.13.0 pytorch-cuda=11.7 -c pytorch -c nvidia
pip install torch-scatter==2.0.9 torch-sparse==0.6.15 torch-cluster==1.6.0 torch-spline-conv==1.2.1 torch-geometric==2.0.4 -f https://data.pyg.org/whl/torch-1.13.0+cu117.html
Then, it threw a new error:
OSError: libcudart.so.11.0: cannot open shared object file: No such file or directory
I found the solution here.
Setting cudatoolkit path can fix this,
CUDNN_PATH=$(dirname $(python -c "import nvidia.cudnn;print(nvidia.cudnn.file)"))
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CONDA_PREFIX/lib/:$CUDNN_PATH/lib
export PYTORCH_CUDA_ALLOC_CONF=max_split_size_mb:32