GPU server option
While CPUs handle a wide variety of tasks quickly, GPUs are better for applications that require parallel processing of large amounts of data, including mathematical computation. For your Jupyter notebook server, you can choose the type of server that is best for your work.
As part of logging into your Jupyter environment, choose either CPU or GPU for your notebook server and click Start:
A GPU server takes longer to launch than a CPU server.
You can select the option appropriate for your work in any session, and switch to the other whenever the need arises—without losing any work.
If you are running a GPU server, you might also choose to install GPU-related packages (TensorFlow, for example). For information and instructions, see:
Switching server types
To switch server types, click File > Hub Control Panel in your JupyterLab environment's top navigation bar. Click Stop My Server in the resulting window.
Once the server has stopped, a Start My Server button appears. Once you click that, you will again have the opportunity to select CPU or GPU.
GPU dashboards
If you are running a GPU server, your JupyterLab environment is able to display several GPU dashboards. The dashboards are an open source implementation through a JupyterLab extension called NVDashboard.
To access the dashboards, click the dashboards icon in the left navigation bar:
This icon does not appear in the navigation bar if you are running a CPU server.