1
votes

An import tensorflow statement triggers the following error:

>>> import tensorflow as tf
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcublas.dylib locally
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcudnn.dylib locally
I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcufft.dylib locally
"import tensorflow" terminated by signal SIGSEGV (Address boundary error)
1

1 Answers

2
votes

By default, CUDA creates libcuda.dylib, but TensorFlow tries to load libcuda.1.dylib. To fix the problem, create a symbolic link between libcuda.dylib and libcuda.1.dylib. For example, if CUDA is installed in /usr/local/cuda, issue the following ln command:

ln -sf /usr/local/cuda/lib/libcuda.dylib /usr/local/cuda/lib/libcuda.1.dylib