Specify additional steps to utilize GPU for Linux users

site/en/install/pip.md

@@ -193,7 +193,55 @@ The following NVIDIA® software are only required for GPU support.
     nvidia-smi
     ```
 
-    ### 3. Install TensorFlow
+    ### 3. Install Miniconda
+
+    You can skip this section if you have already installed `Miniconda` (referred as *option #1* in the next steps) or you prefer to use Python’s built-in `venv` module (referred as *option #2* in the next steps) instead.
+
+    [Miniconda](https://docs.conda.io/en/latest/miniconda.html){:.external}
+    is the recommended approach for installing TensorFlow with GPU support.
+    It creates a separate environment to avoid changing any installed
+    software in your system. This is also the easiest way to install the
+    required software especially for the GPU setup.
+
+    Follow the instuctions of the conda user guide to install miniconda
+    [Miniconda Installation Guide](https://conda.io/projects/conda/en/latest/user-guide/install/linux.html){:.external}.
+
+    ### 4. Create a virtual environment
+
+    * ***Option #1: Miniconda*** 
+
+    Create a new conda environment named `tf` with the following command.
+
+    ```bash
+    conda create --name tf python=3.11
+    ```
+    You can activate and deactivate it with the following commands.
+
+    ```bash
+    conda activate tf
+    conda deactivate
+    ```
+
+    * ***Option #2: venv***
+
+    The [venv](https://docs.python.org/3/library/venv.html){:.external} module supports creating lightweight “virtual environments”, each with their own independent set of Python packages installed in their site directories. 
+
+    Navigate to your desired virtual environments directory and create a new venv environment named `tf` with the following command.
+
+    ```bash
+    python3 -m venv tf 
+    ```
+
+    You can activate and deactivate it with the following commands.
+
+    ```bash
+    source tf/bin/activate
+    deactivate
+    ```
+
+    Make sure that the virtual environment is activated for the rest of the installation.
+
+    ### 5. Install TensorFlow
 
     TensorFlow requires a recent version of pip, so upgrade your pip
     installation to be sure you're running the latest version.
@@ -211,7 +259,106 @@ The following NVIDIA® software are only required for GPU support.
     pip install tensorflow
     ```
 
-    ### 4. Verify the installation
+    Note: Do not install TensorFlow with `conda`. It may not have the latest stable version. `pip` is recommended since TensorFlow is only officially released to PyPI.
+
+    ### 6. Set environment variables
+
+    You can skip this section if you only run TensorFlow on the CPU.
+
+    * ***Option #1: Miniconda*** 
+
+    Locate the directory for the conda environment in your terminal window by running in the terminal:
+    `echo $CONDA_PREFIX`
+
+    Enter that directory and create these subdirectories and files:
+
+    ```bash
+    cd $CONDA_PREFIX
+    mkdir -p ./etc/conda/activate.d
+    mkdir -p ./etc/conda/deactivate.d
+    touch ./etc/conda/activate.d/env_vars.sh
+    touch ./etc/conda/deactivate.d/env_vars.sh
+    ```
+    Edit `./etc/conda/activate.d/env_vars.sh` as follows:
+
+    ```bash
+    #!/bin/sh
+
+    # Store original LD_LIBRARY_PATH 
+    export ORIGINAL_LD_LIBRARY_PATH="${LD_LIBRARY_PATH}" 
+
+    # Get the CUDNN directory 
+    CUDNN_DIR=$(dirname $(dirname $(python -c "import nvidia.cudnn; print(nvidia.cudnn.__file__)")))
+
+    # Set LD_LIBRARY_PATH to include CUDNN directory
+    export LD_LIBRARY_PATH=$(find ${CUDNN_DIR}/*/lib/ -type d -printf "%p:")${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
+
+    # Get the ptxas directory  
+    PTXAS_DIR=$(dirname $(dirname $(python -c "import nvidia.cuda_nvcc; print(nvidia.cuda_nvcc.__file__)")))
+
+    # Set PATH to include the directory containing ptxas
+    export PATH=$(find ${PTXAS_DIR}/*/bin/ -type d -printf "%p:")${PATH:+:${PATH}}
+    ```
+    Edit `./etc/conda/deactivate.d/env_vars.sh` as follows:
+
+    ```bash
+    #!/bin/sh
+
+    # Restore original LD_LIBRARY_PATH
+    export LD_LIBRARY_PATH="${ORIGINAL_LD_LIBRARY_PATH}"
+
+    # Unset environment variables
+    unset CUDNN_DIR
+    unset PTXAS_DIR
+    ```
+    * ***Option #2: venv***
+
+    Locate the directory for the venv environment in your terminal window by running in the terminal:
+    `echo $VIRTUAL_ENV`
+
+    Enter that directory and add the following lines at the end of the activate script `./bin/activate` as follows:
+
+    ```bash
+    # Store original LD_LIBRARY_PATH
+    export ORIGINAL_LD_LIBRARY_PATH=$LD_LIBRARY_PATH 
+
+    # Get the CUDNN directory
+    CUDNN_DIR=$(dirname $(dirname $(python -c "import nvidia.cudnn; print(nvidia.cudnn.__file__)")))
+
+    # Set LD_LIBRARY_PATH to include CUDNN directory
+    export LD_LIBRARY_PATH=$(find ${CUDNN_DIR}/*/lib/ -type d -printf "%p:")${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
+
+    # Get the ptxas directory 
+    PTXAS_DIR=$(dirname $(dirname $(python -c "import nvidia.cuda_nvcc; print(nvidia.cuda_nvcc.__file__)")))
+
+    # Set PATH to include the directory containing ptxas
+    export PATH=$(find ${PTXAS_DIR}/*/bin/ -type d -printf "%p:")${PATH:+:${PATH}}
+    ```
+
+    Add the following lines at the end of `deactivate` block in the activate script to ensure that the necessary NVIDIA environment variables are set only while the virtual environment is active:
+
+    ```bash
+    deactivate () {
+        # ...
+        # Unset the added path to PATH if within the virtual environment
+        if [ -n "$VIRTUAL_ENV" ]; then
+            # Remove the path from PATH
+            PATH=$(echo $PATH | sed -e "s|${PTXAS_DIR}/*/bin/:||g")
+        fi
+
+        # Restore original LD_LIBRARY_PATH
+        if [ -n "$ORIGINAL_LD_LIBRARY_PATH" ]; then
+            export LD_LIBRARY_PATH=$ORIGINAL_LD_LIBRARY_PATH
+            unset ORIGINAL_LD_LIBRARY_PATH
+        fi
+
+        # Unset environment variables
+        unset CUDNN_DIR
+        unset PTXAS_DIR    
+    }
+    ```
+
+    ### 7. Verify the installation
 
     Verify the CPU setup: