KALDodocker is a Docker container designed for computational simulations, integrating LAMMPS with Python support and KALDo software. Tailored for CPU environments, it also provides instructions for GPU usage adaptation.
To configure the Docker environment for CPU-based computations, rename the CPU-specific Dockerfile:
```bash mv Dockerfile-cpu Dockerfile ```
Build the Docker image with the following command:
```bash docker build -t kaldo . ```
Execute the following to run KALDo in an interactive shell environment:
```bash
docker run -it --rm -u
This grants access to a full Ubuntu environment within the container.
For easy file transfer, mount your current directory inside the container using the -v option:
```bash
docker run -it --rm -u
This command mounts your present working directory to /root inside the container.
To enable GPU support, follow these instructions, ensuring you have nvidia-cuda-toolkit and compatible drivers installed.
Switch to the GPU-specific Dockerfile:
```bash mv Dockerfile-gpu Dockerfile ```
Build the GPU-enabled Docker image with:
```bash docker build -t kaldo-gpu . ```
For container execution with GPU support:
```bash
docker run -it --rm -u
Use the --gpus all flag to enable all GPUs or --gpus=0,1 for specific GPUs.
Set up KALDo for testing or contributions with the Docker-dev environment.
To run development tests with pytest, execute:
```bash mv Dockerfile-dev Dockerfile docker build --tag kaldo-dev . docker run --rm --gpus all kaldo-dev pytest ```
Post-execution, verify KALDo's GPU support and additional features. The container installs the main branch of KALDo, suitable for a test environment.
- For detailed Docker and GPU support information, consult the NVIDIA Container Toolkit documentation and the TensorFlow Docker installation guide.
- The development version (
devel) includes the full KALDo package, intended for debugging or full access to KALDo's features.