Tensorflow missing cuDNN drivers

[ca-scribner]

Our GPU dockerfile stack is rooted on an NVIDIA image (nvidia/cuda:10.1-base-ubuntu18.04) which does not include the CUDA Deep Neural Network libraries which tensorflow requires. This is discussed in iot-salzburg/gpu-jupyter#25.

iot-salzburg/gpu-jupyter#25 resolves the issue by basing off a different NVIDIA image - using the current release of iot-salzburg/gpu-jupyter (or changing our NVIDIA base) would resolve this issue.

Current work here on #114 has a solution mocked up for this issue

[ca-scribner]

solution tested with:

import tensorflow as tf


print("Num GPUs Available: ", len(tf.config.experimental.list_physical_devices('GPU')))
tf.config.list_physical_devices('GPU')
# Should list at least one `PhysicalDevice(name='/physical_device:GPU:*', device_type='GPU')`


tf.debugging.set_log_device_placement(True)
# Create some tensors
a = tf.constant([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])
b = tf.constant([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]])
c = tf.matmul(a, b)
print(c)

# Should show something like (note the device used for work should be GPU):
# ```
# Executing op MatMul in device /job:localhost/replica:0/task:0/device:GPU:0
# tf.Tensor(
# [[22. 28.]
#  [49. 64.]], shape=(2, 2), dtype=float32)
#