Hi!
I managed to install CUDA and Theano to the point where Theano would use my GPU, but cuDNN failed. After adding paths to .theanorc, Theano would break with this error:
(base) jan@jan-Blade:~$ python
Python 3.7.3 (default, Mar 27 2019, 22:11:17)
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type “help”, “copyright”, “credits” or “license” for more information.
import theano
Traceback (most recent call last):
File “”, line 1, in
File “/home/jan/miniconda3/lib/python3.7/site-packages/theano/init.py”, line 156, in
import theano.gpuarray
File “/home/jan/miniconda3/lib/python3.7/site-packages/theano/gpuarray/init.py”, line 33, in
[…]
File “/home/jan/miniconda3/lib/python3.7/site-packages/theano/gof/cmodule.py”, line 317, in dlimport
rval = import(module_name, {}, {}, [module_name])
ImportError: libcudnn.so.7: cannot open shared object file: No such file or directory
libcudnn.so.7 is in the lib64 directory. My .theanorc looks like that:
[global]
floatX = float32
device = cuda0
[gpuarray]
preallocate = 0.9
[dnn]
library_path = /usr/local/cuda-10.1/lib64
include_path = /usr/local/cuda-10.1/include
The NVIDIA-SMI shows:
±----------------------------------------------------------------------------+
| NVIDIA-SMI 430.26 Driver Version: 430.26 CUDA Version: 10.2 |
|-------------------------------±---------------------±---------------------+
I could not find an useful tips elsewhere, I would heavily appreciate your help!
Thx