Ubuntu 14.04 cuda7.5 cudnn 5.1 Anaconda2.7 install problem

[filmo]

I'm recently updated my cudnn to 5.1 and thought to recompile cudarray. In hindsight this might not have been needed, but now I seem to have broken my install. (It was working before the recompile.)

I've got CUDA 7.5 installed correctly in /usr/local/cuda and my environmental variables are correctly set as per the README file. (I've compiled the NVIDIA examples to make sure those work, and they do.)

I did

make
make install
python setup.py install

Results here: https://gist.github.com/filmo/0821af03f01da8592e09921953b4ae5a Seems like it correctly compiled to me and installed correctly in my anaconda lib/python2.7/site-packages folder, but perhaps not ??? (I see the cudarray-0.1.dev0-py2.7-linux-x86_64.egg in that folder.)

However when I now try to import inside of Python I'm seeing the following:

philglau (master *) cudarray $ python
Python 2.7.12 |Anaconda 2.4.1 (64-bit)| (default, Jul  2 2016, 17:42:40) 
[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://anaconda.org
>>> import cudarray as ca
CUDArray: CUDA back-end not available, using NumPy.
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "cudarray/__init__.py", line 40, in <module>
    from .numpy_backend import *
  File "cudarray/numpy_backend/__init__.py", line 2, in <module>
    from .nnet import *
  File "cudarray/numpy_backend/nnet/__init__.py", line 3, in <module>
    from .conv_bc01 import *
ImportError: No module named conv_bc01

If I list the folder referenced in the error message, I see conv_bc01

philglau nnet $ ls
activations.py   conv_bc01.py   conv_bc01.so  conv.pyc     __init__.pyc    lrnorm_bc01.pyc  pool_bc01.py   pool_bc01.so  pool.pyc    special.pyc
activations.pyc  conv_bc01.pyc  conv.py       __init__.py  lrnorm_bc01.py  lrnorm_bc01.so   pool_bc01.pyc  pool.py       special.py
philglau nnet $ 

Any pointers on where I might have gone wrong?

[andersbll]

Try make clean and python setup.py clean before building.