• How to install Cuda on Windows 11 and Anaconda?
  • Virtual Environment in Anaconda
• Run Jupyter Notebooks in PyCharm
• pip commands
• Delete Kernel

---

How to install Cuda on Windows 11 and Anaconda?

1. Cuda & CuDNN files Download

2. Python 3 & pip with Anaconda(latest installed)

3. Microsoft Visual Studio 2022 C++ redistributables with dot net 4.5 or above installed

4. Make sure you have already Anaconda installed with JupyterNotebook & everything is working fine.

5. Download the CUDA toolkit from here:- https://developer.nvidia.com/cuda-downloads#

6. choose your options like OS and arch of OS based on your system configurations. Then download the file and install it in your C//: DRIVE, while installation uses EXPRESS(recommended) mode.

7. After installing CUDA, you must download CuDNN (https://developer.nvidia.com/rdp/cudnn-archive), an extension that supports CUDA files for running Deep Learning applications.

• NOTE: MAKE SURE THE CUDNN VERSION AND CUDA TOOLKIT VERSIONS ARE SAME

8. Once you have downloaded CuDNN Local Installer for Windows (Zip), extract files to the CUDA installation directory in C:// drive
9. now one by copy & replace these folders to the CUDA installation directory

10. you have finally completed your CUDA Toolkit installation with CuDNN.
11. Now it’s time to set Path variables → open Edit environment variables in the Windows search bar, Click on Environment variables, click on “path” under user variables, then add the paths of “bin” and “lineup.”

• Right-click on Start, then click on Run. Type: sysdm.cpl and click on OK. Click on the "Advanced" tab, then click on the "Environment Variables..."

12. Finally, you’re done with the installation part of CUDA & CuDNN; just verify it’s properly installed, Goto Power-Shell, and enter the command “Nvidia-semi.”

13. nvcc --version
14. nvidia-smi

---

Virtual Environment in Anaconda

In Anaconda cmd.exe prompt:

1. conda create -n "name of your virtual environment" python=3.6.8
2. conda activate "name of your virtual environment"
3. conda install -c anaconda tensorflow-gpu keras-gpu
4. conda install pytorch torchvision torchaudio pytorch-cuda=12.4 -c pytorch -c nvidia # batavajoh be version cuda

-https://pytorch.org/get-started/locally/

5.pip install ipykernel

6. python -m ipykernel install --user --name --display-name "Python (GPU)" # then launch jypiternotebook

7. pip install --upgrade jupyterhub
8. pip install --upgrade --user nbconvert

---

How to delete the old kernel

1. jupyter kernelspec list
2. jupyter kernelspec uninstall your_env_name

or

1. jupyter kernelspec list
2. cd /path/to/kernelspecs/kernel_name
3. rm -r /path/to/kernelspecs/kernel_name
4. jupyter kernelspec remove kernel_name

---

Run Jupyter Notebooks in PyCharm

---

pip commands

1. pip install package_name
2. pip uninstall package_name
3. pip uninstall package_name
4. pip show package_name
5. pip freeze requirements.txt
6. pip cache purge
7. pip list
8. python -m site --user-site # path to site package
9. python -m venv # python3.8 -m venv env
10. source env/bin/activate in linux or env/Scripts/activate.bat //In CMD windows

---

How to Add a Python 3 Kernel to Jupyter IPython

step 0: Prerequisites

1. Python 3.x (preferably the latest version)
2. Jupyter Notebook
3. virtualenv (for creating virtual environments)

Step 1: Create a Python 3 Virtual Environment

1. virtualenv -p python3.9 name

Step 2: Activate the Virtual Environment

1. source name/bin/activate

Step 3: Install the IPython Kernel Package

1. pip install ipykernel

Step 4: Register the Kernel with Jupyter

1. python -m ipykernel install --user --name=my-python3-kernel