Setting_up_TensorFlow_Conda_Environment/Install TensorFlow on Win Server 2019 Conda Env.md

In John-R-Wallace-NOAA/FishNIRS: NIRS on fish structures.

 
# To start, you first need to see the Nvidia graphics card as a GPU on your Windows Server 2019 client, 
#   here are instructions for Windows Server 2016:
#     https://community.esri.com/t5/implementing-arcgis-questions/enabling-gpu-rendering-on-windows-server-2016/td-p/658522

# For a picture, see the Task Manager screenshot here, but don't follow this site past the screenshot:
#     https://towardsdatascience.com/setting-up-tensorflow-gpu-with-cuda-and-anaconda-onwindows-2ee9c39b5c44

# Install Anaconda from here:
#     https://www.anaconda.com/products/distribution

# After the Anaconda installation follow the steps below, but first here is an informational reference:
#     https://conda-forge.org/blog/posts/2021-11-03-tensorflow-gpu/


# In the first two sections of code below, the only real difference is that for the CPU version "tensorflow" is installed, 
#    and for the gpu version "tensorflow-gpu" is installed.

# ---------- Tensorflow CPU only, no GPU nor TensorRT. (Using the conda-forge package manager.) --------------

# *** The CPU only version does work in R under Windows Server 2019.  *** 
# See the other markdown doc in this folder for a GPU version that works in R for Windows 10 & 11.

conda create -y -p tf_cpu_only python=3.8

conda env list

conda activate tf_cpu_only

python --version

conda install -y -c conda-forge cudatoolkit=11.2 cudnn=8.1.0

# If versions and builds are not listed below, something went wrong
conda list cudatoolkit
conda list cudnn

# Check CUDA installation 
nvcc --version
conda list cuda

conda install tensorflow -c conda-forge

# Find version number of tensorflow (2.10.1)
conda list tensorflow

# Verfiy TensorFlow using a single line submission to python approach
python -c "import tensorflow as tf;print('\n\n\n====================== \n GPU Devices: ',tf.config.list_physical_devices('GPU'), '\n======================')"
python -c "import tensorflow as tf;print('\n\n\n====================== \n', tf.reduce_sum(tf.random.normal([1000, 1000])), '\n======================' )"


# Verfiy TensorFlow using an interactive approach
python
>>> 
import tensorflow as tf
print(tf.__version__)
print(tf)

tf.config.list_physical_devices('CPU')
tf.config.list_physical_devices('GPU')
len(tf.config.list_physical_devices('GPU'))

a = tf.constant(7)
b = tf.constant(10)
print(tf.add(a,b))

print(tf.reduce_sum(tf.random.normal([1000, 1000])))

quit()
>>>

conda deactivate


# ----------- Tensorflow GPU works in Python, but not in R. (Using the conda-forge package manager.) ---------------------

conda create -y -p tf_gpu_py python=3.8

conda env list

conda activate tf_gpu_py

python --version

conda install -y -c conda-forge cudatoolkit=11.2 cudnn=8.1.0

conda list cudatoolkit
conda list cudnn

conda install tensorflow-gpu -c conda-forge


# Verfiy TensorFlow - single line approach
python -c "import tensorflow as tf;print('\n\n\n====================== \n GPU Devices: ',tf.config.list_physical_devices('GPU'), '\n======================')"
python -c "import tensorflow as tf;print('\n\n\n====================== \n', tf.reduce_sum(tf.random.normal([1000, 1000])), '\n======================' )"


# Verfiy TensorFlow - interactive approach
python
>>> 
import tensorflow as tf
print(tf.__version__)
print(tf)

tf.config.list_physical_devices('CPU')
tf.config.list_physical_devices('GPU')
len(tf.config.list_physical_devices('GPU'))

a = tf.constant(7)
b = tf.constant(10)
print(tf.add(a,b))

tf.reduce_sum(tf.random.normal([1000, 1000]))

quit()
>>>

conda deactivate



# ------------- Mamba Package Manager install - only CPU worked for me ------------------

# Main reference for Mamba:
     https://github.com/mamba-org/mamba

# For code see the second Comment 3 by prerakmody here (I'm not sure why there are two Comment 3's.): 
     https://stackoverflow.com/questions/54271094/conda-install-c-conda-forge-tensorflow-just-stuck-in-solving-environment

conda create -y --name tf_mamba python=3.8

conda env list

conda activate tf_mamba


# Mamba install includes cudatoolkit and cudnn, so there are less steps and a nice interface, but 'tensorflow-gpu' failed  
#       for me under Windows Server 2019 and Windows 10.
conda install -y -c conda-forge mamba
#  mamba install -y -c conda-forge tensorflow-gpu   
mamba install -y -c conda-forge tensorflow


# Verfiy TensorFlow - single line approach
python -c "import tensorflow as tf;print('\n\n\n====================== \n GPU Devices: ',tf.config.list_physical_devices('GPU'), '\n======================')"
python -c "import tensorflow as tf;print('\n\n\n====================== \n', tf.reduce_sum(tf.random.normal([1000, 1000])), '\n======================' )"


# Verfiy TensorFlow - interactive approach
python
>>> 
import tensorflow as tf
print(tf.__version__)
print(tf)

tf.config.list_physical_devices('CPU')
tf.config.list_physical_devices('GPU')
len(tf.config.list_physical_devices('GPU'))

a = tf.constant(7)
b = tf.constant(10)
print(tf.add(a,b))

quit()
>>>

conda deactivate

John-R-Wallace-NOAA/FishNIRS documentation built on April 12, 2025, 12:59 a.m.