https://github.com/devslem/docker-cuda-ubuntu-zsh
Docker for CUDA Available Ubuntu with ZSH Shell
https://github.com/devslem/docker-cuda-ubuntu-zsh
Last synced: about 1 year ago
JSON representation
Docker for CUDA Available Ubuntu with ZSH Shell
- Host: GitHub
- URL: https://github.com/devslem/docker-cuda-ubuntu-zsh
- Owner: DevSlem
- License: mit
- Created: 2025-03-21T09:25:26.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-04-25T08:50:12.000Z (about 1 year ago)
- Last Synced: 2025-05-07T12:54:51.093Z (about 1 year ago)
- Language: Shell
- Homepage:
- Size: 38.1 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Docker for CUDA Available Ubuntu with ZSH Shell
This repository provides Docker files for CUDA availalbe Ubuntu OS with ZSH shell.
For each version of Ubuntu, there are two Dockerfiles with different dependencies:
1. `cuda12.8.1-ubuntu22.04-zsh/Dockerfile`
2. `cuda11.3.1-ubuntu20.04-zsh/Dockerfile`
These Dockerfiles provide the following environments:
* **Deep learning**:
* CUDA
* CUDNN
* Anaconda
* uv
* **ZSH shell**:
* Oh My Zsh
* Powerlevel10k theme
* Autosuggestions plugin
* Syntax-highlighting plugin
* **Essential packages**:
* build-essential
* wget
* vim
* curl
* git
* tmux
* openssh-server
* Github CLI
* colorls (only Ubuntu 22.04)
* and so on...
Build a Docker image by:
```bash
docker build -t cuda12.8.1-ubuntu22.04-zsh:v1 ./cuda12.8.1-ubuntu22.04-zsh
```
Run the container by:
```bash
docker run --rm -it --gpus all --name ai-dev-env cuda12.8.1-ubuntu22.04-zsh:v1
```
Also, this repository provides `docker-compose-template.yml` for handling multiple containers. You can easily create and configure a `docker-compose.yml` file with ``, `` and `` by entering the following command:
```bash
bash config.sh
```
Then, Start the container by:
```bash
export SSH_PUBLIC_KEY=""
docker compose up --build -d
```
`nvidia-smi` doesn't show whether processes using GPU are running in Docker containers or not. To check it, you can use the `nvidia-docker-smi` shell script provided in this repository. Install nvidia-docker-smi shell script:
```bash
chmod +x nvidia-docker-smi.sh
sudo mv nvidia-docker-smi.sh /usr/local/bin/nvidia-docker-smi
```
Run the command:
```bash
nvidia-docker-smi
```
Example output:
```bash
GPU Container ID Container Name PID Process GPU Memory Usage (MiB)
=====================================================================================
1 abcdefg12345 testuser.ai-dev-env-base 884311 python 280
2 hijklmn67890 otheruser.ai-dev-env-base 866294 python 6316
```