Running your container

[jgato]

Hi there,

I dont know if you are giving too much support to this report, but... I am trying to create a Docker Image for the Jetson Nano with Tensorflow and GPU access. I have tried many docker build examples, but the combinations between CUDA and Tensorflow libraries is been crazy. So I reached your repo that seems to have a pre-compiled image but when I try, I have an error about

>>> import tensorlfow
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'tensorlfow'
>>> import tensorflow
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/pywrap_tensorflow.py", line 58, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
    _mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
  File "/usr/lib/python3.6/imp.py", line 243, in load_module
    return load_dynamic(name, filename, file)
  File "/usr/lib/python3.6/imp.py", line 343, in load_dynamic
    return _load(spec)
ImportError: libcublas.so.10.0: cannot open shared object file: No such file or directory

I guess is because the libculas version installed in your image is 10.2:

# ls -l /usr/lib/aarch64-linux-gnu/libcublas*
lrwxrwxrwx 1 root root       15 Aug  3 07:21 /usr/lib/aarch64-linux-gnu/libcublas.so -> libcublas.so.10
lrwxrwxrwx 1 root root       22 Aug  3 07:21 /usr/lib/aarch64-linux-gnu/libcublas.so.10 -> libcublas.so.10.2.2.89
-rw-r--r-- 1 root root 80530928 Oct 29  2019 /usr/lib/aarch64-linux-gnu/libcublas.so.10.2.2.89
lrwxrwxrwx 1 root root       17 Aug  3 07:21 /usr/lib/aarch64-linux-gnu/libcublasLt.so -> libcublasLt.so.10
lrwxrwxrwx 1 root root       24 Aug  3 07:21 /usr/lib/aarch64-linux-gnu/libcublasLt.so.10 -> libcublasLt.so.10.2.2.89
-rw-r--r-- 1 root root 33235064 Oct 29  2019 /usr/lib/aarch64-linux-gnu/libcublasLt.so.10.2.2.89

Reading about it, I tried to update inside your image to tensorflow-gpu 1.15 but then I have a seg fault:

/# python3
Python 3.6.8 (default, Jan 14 2019, 11:02:34) 
[GCC 8.0.1 20180414 (experimental) [trunk revision 259383]] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow
Segmentation fault (core dumped)

any idea/help?

[jgato]

I think I got it. The base docker image provided by nvidia, shares your local runtime environment into the container. Therefore, no matter what I do, I will have CUDA 10.2 inside the container so I need Tensorflow 2.2 installed. In your case, when you did this example, the conditions were different about the local runtime environment.