Creating a New Node Group

1. Log in and Navigate
   Log in to the GPU Cluster Manager. In the left navigation bar, locate and click Nodes and Node Groups. You will see a list of Node Groups.

   

2. Open the “New Node Group” Dialog
   In the upper-right corner of the Node Groups page, click + New Node Group. A dialog box appears, allowing you to configure the new Node Group.

   

3. Enter Basic Information

   • Name: Assign a descriptive name for your Node Group (e.g., ng-engineering).
   • Department: Select or type in the department name (e.g., Engineering).

4. Select Nodes
   In the lower portion of the dialog, a list of available nodes is displayed along with the following information for each node:

   • Name: The node’s hostname or label.
   • Status: The readiness status of the node.
   • GPU Vendor: The name of the GPU Vendor (NVDIA, AMD)
   • GPU Model: The GPU Model within the node.
   • GPU Driver Version: The version of the driver this GPU is currently running.
   • CUDA Version: The version of CUDA that the GPU is running.
   • GPU Memory Capacity: The amount of memory assigned to the GPU.
   • Number of GPUs: Number of GPUs assigned to the node.
   • CPU Memory Capacity: The amount of memory assigned to the CPU.
   • CPU Cores: The number of cores the CPU assigned to this node has.

   Check the box next to the node(s) you want to include in this Node Group.

5. Create the Node Group
   After confirming your node selections, click Create. You will be returned to the Node Groups page, which now displays the newly created Node Group, including information such as the number of nodes, GPUs, and CPU cores.

6. Verify Node Group Creation

   • Confirm that the new Node Group is listed under Node Groups.
   • Check that the status shows the expected number of nodes, GPUs, and memory.

Once the Node Group is created, it can be used to schedule workloads that require its specific GPU resources or configurations.