Describes how to create and deploy the GPU-enabled notebook servers.
Prerequisites
Sign in to HPE Ezmeral Unified Analytics Software.
About this task
Create GPU-enabled notebook servers in HPE Ezmeral Unified Analytics Software.
Procedure
-
Click Notebooks icon on the left navigation bar of
HPE Ezmeral Unified Analytics Software screen.
-
Click New Notebook Server. You will be navigated to the
Kubeflow Notebooks UI. You can choose JupyterLab as your notebook server within Kubeflow Notebooks.
-
Configure the notebook server with the following options:
- Select one of the following docker images:
- (Tensorflow CUDA image)
gcr.io/mapr-252711/kubeflow/notebooks/jupyter-tensorflow-cuda-full:<image-tag>
- (PyTorch CUDA image)
gcr.io/mapr-252711/kubeflow/notebooks/jupyter-pytorch-cuda-full:<image-tag>
- Set Requested memory in Gi to at least two to
three Gi.
- Set GPUs as follows:
- Number of GPUs: 1
NOTE
With
MIG configuration, only one GPU can be assigned per
application. To learn more on what happens when you assign
more than one GPU to the notebook server, see
GPU. For details regarding GPU, see
GPU Support.
- GPU Vendor: Nvidia
-
Click Launch.
Results
The new GPU-enabled notebook server is created.