Install problem on Ubuntu 20.04 LTS + GPU

TensorFlow General Discussion
, ,
1

I am having trouble installing TensorFlow on Ubuntu 20.04 LTS with GPU:

$ export LD_LIBRARY_PATH=/usr/local/lib/python3.8/dist-packages/numpy.libs/
$ source /opt/tensorflow/bin/activate
$ python3 -c 'import tensorflow as tf; print(tf.__version__)'
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/lib/python3.8/site-packages/tensorflow/__init__.py", line 41, in <module>
    from tensorflow.python.tools import module_util as _module_util
  File "/usr/lib/python3.8/site-packages/tensorflow/python/__init__.py", line 41, in <module>
    from tensorflow.python.eager import context
  File "/usr/lib/python3.8/site-packages/tensorflow/python/eager/context.py", line 30, in <module>
    import numpy as np
  File "/home/daldrich/.local/lib/python3.8/site-packages/numpy/__init__.py", line 163, in <module>
    from . import ma
ImportError: cannot import name 'ma' from partially initialized module 'numpy' (most likely due to a circular import) (/home/<snip>/.local/lib/python3.8/site-packages/numpy/__init__.py)

Can anyone help please?

3

Can you try to use the recommended virtual environment as in:

4

@Bhack Thanks for your reply.
I followed the instructions you referred to and saw:

(venv) $ python -c "import tensorflow as tf;print(tf.reduce_sum(tf.random.normal([1000, 1000])))"
<snip>
ImportError: libopenblasp-r0-2d23e62b.3.17.so: cannot open shared object file: No such file or directory

I then set:

$ export LD_LIBRARY_PATH=/usr/local/lib/python3.8/dist-packages/numpy.libs/
and now see the same error I reported above:

(venv) daldrich@ubuntusim11:/data/daldrich$ python -c "import tensorflow as tf;print(tf.reduce_sum(tf.random.normal([1000, 1000])))"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/usr/lib/python3.8/site-packages/tensorflow/__init__.py", line 41, in <module>
    from tensorflow.python.tools import module_util as _module_util
  File "/usr/lib/python3.8/site-packages/tensorflow/python/__init__.py", line 41, in <module>
    from tensorflow.python.eager import context
  File "/usr/lib/python3.8/site-packages/tensorflow/python/eager/context.py", line 30, in <module>
    import numpy as np
  File "/home/daldrich/.local/lib/python3.8/site-packages/numpy/__init__.py", line 163, in <module>
    from . import ma
ImportError: cannot import name 'ma' from partially initialized module 'numpy' (most likely due to a circular import) (/home/daldrich/.local/lib/python3.8/site-packages/numpy/__init__.py)
5

I cannot reproduce your error with a fresh install on Ubuntu 20.04:

docker run --rm -it ubuntu/ubuntu:20.04 /bin/bash

apt-get update && apt-get -y install python3 python3.8-venv

python3 -m venv --system-site-packages ./venv
source ./venv/bin/activate  # sh, bash, or zsh
pip install --upgrade pip
pip install --upgrade tensorflow
python -c "import tensorflow as tf;print(tf.reduce_sum(tf.random.normal([1000, 1000])))"

tf.Tensor(97.31439, shape=(), dtype=float32)
6

Thanks again.

I tried again and set:

export LD_LIBRARY_PATH=/usr/lib/cuda/lib64

and now see:

$ python -c "import tensorflow as tf;print(tf.reduce_sum(tf.random.normal([1000, 1000])))"
2021-11-17 18:25:24.770863: I tensorflow/core/platform/cpu_feature_guard.cc:151] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-11-17 18:25:25.271273: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1525] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 30985 MB memory:  -> device: 0, name: Tesla V100S-PCIE-32GB, pci bus id: 0000:d8:00.0, compute capability: 7.0
tf.Tensor(1303.674, shape=(), dtype=float32)

$ python3 -c 'import tensorflow as tf; print(tf.__version__)'
2.7.0

So I think that is successful?

7

It seems yes. Try to print the available GPU device: