# Installation

To install `gamdpy` you need a computer 
with a **Nvidia GPU**, Python3, and the following software installed (Windows users: see below):

1. the [CUDA toolkit](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html), and
2. the [numba](https://numba.readthedocs.io/) python package with CUDA GPU support (install `cudatoolkit`).

Ensure that this is working before proceeding with the installation (see below):

```sh
python3 -c "import numba.cuda; print(numba.cuda.get_current_device().name)"
```

should execute without errors.

## Linux

### Install with the Python Package Index (pip)

Get the latest stable version with (recommended):

```sh
pip install gamdpy
```

Alternatively, the latest developers version (on GitHub) can be installed:

```sh
pip install git+https://github.com/ThomasBechSchroeder/gamdpy.git
```

### From source (for developers)

If you want to inspect or modify the source code, the package can installed by cloning the source code 
from GitHub to a local directory (change `[some_directory]` to the desired path):

```sh
cd [some_directory]
git clone https://github.com/ThomasBechSchroeder/gamdpy.git  # Clone latest developers version
cd gamdpy
python3 -m venv venv  # Create virtual enviroment
. venv/bin/activate   # ... and activate
pip install -e .      # Install gamdpy 
```

Update to a latest version by executing

```sh
git pull
```

in the `gamdpy` directory.

## Windows (using WSL)

The following show how to install `gamdpy` 
on windows using Windows Subsystem For Linux (WSL).

### Install WSL
Open PowerShell or Windows Command Prompt in administrator mode by right-clicking and selecting "Run as administrator", enter the command

```sh
wsl --install
```

press enter and then restart your machine. 
The default installation is Ubuntu, for others check: <https://learn.microsoft.com/en-us/windows/wsl/install>

### Install python and pip on WSL

- open Windows Command Prompt
- in the tab bar click on "v" and select ubuntu
```sh 
sudo apt-get update
sudo apt-get install python3
sudo apt-get install pip
```

### Install miniconda 

See <https://docs.anaconda.com/miniconda/>

```sh
mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh
~/miniconda3/bin/conda init bash
```

### Install cuda

```sh
miniconda3/condabin/conda install cudatoolkit
sudo apt install nvidia-cuda-toolkit
```

- modify `.bashrc` adding: `export LD_LIBRARY_PATH="/usr/lib/wsl/lib/"` from <https://github.com/numba/numba/issues/7104>


### Install gamdpy

For the latest stable version,

```sh
pip install gamdpy
```

and for the latest developer version

```sh
pip install git+https://github.com/ThomasBechSchroeder/gamdpy.git
```

## Windows (using Anaconda)

### Install Anaconda

Install Anaconda: <https://docs.anaconda.com/anaconda/install/windows/>

### Install gamdpy 

Finally, we install `gamdpy` (and `pip`) using Powershell Prompt in Anaconda

- open Anaconda Powershell as admin (from search)

```sh
conda update -n base -c defaults conda
conda install anaconda::pip
conda install anaconda::git
conda config --set channel_priority flexible
conda install cudatoolkit
```

Then type

```sh
pip install gamdpy
```

for the stable release, or

```sh
pip install git+https://github.com/ThomasBechSchroeder/gamdpy.git
```

for the latest developers version.

## Known issues

### LinkerError: libcudadevrt.a not found
A workaround to fix the error `numba.cuda.cudadrv.driver.LinkerError: libcudadevrt.a not found` 
is to make a symbolic link to the missing file. 
This can be done by running the something like the below in the terminal:

```bash
ln -s /usr/lib/x86_64-linux-gnu/libcudadevrt.a .
```

in the folder of the script. Note that the path to `libcudadevrt.a` to the file may vary depending on the system.