Choosing Your Server Options

When you log in to the CDSS DataHub (JupyterHub), you will see the Server Options page. This is where you decide how your computing environment will be set up.

You will have two main choices:

CPU instance¶

• Your server will not have a GPU, but it will typically launch faster.

• Best suited for:

  • Data cleaning and preprocessing

  • Exploratory data analysis

  • Running small to medium computations

• You can set the CPU RAM limit (default: 10G).

GPU instance¶

• Your server will request and wait for a GPU. Launch time may be longer due to limited availability.

• Best suited for:

  • Training machine learning and deep learning models

  • Large-scale parallel computations

  • Tasks requiring GPU acceleration (e.g., TensorFlow, PyTorch)

• You can also set the CPU RAM limit (default: 10G).

👉 Tip:
Start with a CPU instance unless your project specifically requires GPU resources.
Once you’ve made your choice, click the Start button to launch your server.