Create a GPU Node
Creating a GPU node uses the same MyAccount create-node flow as a regular node. The only difference is the order in which you make the selection:
- On a regular node, you pick the OS first, then the plan screen lets you configure version, volume, encryption, and the rest.
- On a GPU node, you pick the GPU card first, then the plan screen lets you configure the OS and version, volume, encryption, and the rest.
For every step that is not GPU-card selection — billing, node name, count, backup, VPC, public IP, security group, SSH keys, volume, advanced settings, launch — follow Create a Node. This page only documents the GPU delta.
GPU availability is region-gated. If the GPU tab or the card you want is not visible, switch regions or contact cloud-platform@e2enetworks.com.
Step 1: Open the GPU Tab and Pick a Card
- Log in to MyAccount.
- Confirm the location selector at the top of the portal is set to a region that offers GPU plans.
- Go to Compute > Nodes and open the create-node flow.
- Switch from the CPU tab to the GPU tab.
- Select the GPU card tile (for example, Nvidia-H100, Nvidia-A100, Nvidia-L40S, Nvidia-A40, Nvidia-A30, Nvidia-L4, Nvidia-V100, Nvidia-T4).
For a comparison of cards before you pick, see Choose a GPU Card.
Step 2: Configure OS, Plan, and Everything Else
After you select the GPU card, the portal opens the plan screen. From here, the configuration surface is the same as a regular node, with these GPU-specific notes:
| Field | What's the same as a regular node | GPU-specific note |
|---|---|---|
| OS and version | Same dropdown layout; same supported families (Ubuntu, Rocky, Windows Server). | Choose the OS and version on this screen rather than on the first screen. Use the edit icon near the selected card to change the GPU. |
| Plan | Same plan dropdown. | Plan format is <vCPU> – <RAM> – <GPU Memory> – <CUDA (Version)> – <Disk Space> – <IOPS (R/W)>. The number of GPU cards is fixed for the plan. |
| Billing | Same on-demand and committed options. | GPU plan SKUs use the GPU-, GDC-, or GDC3- prefixes. |
| Volume | Same attach-at-launch flow when the option is shown. | Most GPU cards do not support attaching a volume during creation. The Volume section is hidden for those plans — create the node first, then attach the volume from the node details page. |
| Encryption | Same advanced-settings toggle. | Same passphrase rules; encryption at rest cannot be disabled later. |
| All other steps | Node name, count, backup, VPC, public IP, security group, SSH, IPv6, BitNinja, start script — all identical. | See Create a Node. |
Click Launch when the Summary panel looks right. GPU nodes generally take longer to reach the Running state than CPU nodes because the driver and toolkit initialize on first boot.
After Launch: Verify the GPU
Once the GPU node is Running, SSH in and confirm the GPU is visible to the driver:
nvidia-smi
You should see the card, driver version, CUDA version, memory, and utilization.
Related Resources
| Resource | Use it for |
|---|---|
| Create a Node | Full step-by-step reference for every non-GPU field on the flow. |
| Choose a GPU Card | Compare cards before launch. |
| Connect to a Linux GPU node | SSH access and nvidia-smi verification. |
| Manage GPU Nodes | GPU-specific differences for monitoring, actions, and lifecycle. |
| TIR AI/ML Platform | Managed notebooks, endpoints, and training jobs. |