I'm new to Docker on Windows and noticed that the desktop interface doesn't seem to expose every container option. In particular, I couldn't find an obvious way to enable NVIDIA GPU access or publish a container port through the graphical interface. After installing the NVIDIA Container Toolkit, I was able to launch the container from WSL with: `docker run --gpus all -p 8188:8188 image_name`. Is there a way to configure these options through the Windows application, or is using the command line the expected approach?
3 Answers
Those options are supported, but the documentation is easy to miss. The command-line flags are the standard way to expose GPUs and publish ports: `--gpus all` grants access to the available NVIDIA GPUs, while `-p 8188:8188` maps the host port to the container port. The graphical interface doesn’t expose every Docker run option.
A lot of people avoid the desktop application and run Docker directly inside WSL with the command-line tools and Compose. It gives you more control and keeps the configuration in files, which is often easier to reproduce.
So the alternative is essentially to skip the desktop application and use the CLI inside WSL?
That’s one option. The desktop application is convenient for seeing what’s running, but the direct Linux-based setup usually exposes more functionality and uses fewer resources.
For a long-running server, Linux is generally the smoother Docker environment. Windows can work, especially through WSL, but GPU passthrough and networking tend to involve more layers and extra setup.
That explains it. I eventually found the GPU setup through a video, but the desktop interface made it seem like the option didn’t exist.