Google Cloud Deployment

Requirements

The requirements for running NVIDIA Isaac Sim on Google Cloud are:

• A Google Cloud account with Compute Engine access that is able to create a Virtual Machine with GPU support.

• A GCP virtual machine with the following recommended specifications:

  T4

  • GPU: nvidia-tesla-t4

  • Machine type: n1-standard-8 or better

  • Image: Ubuntu 22.04 LTS

  L4

  • GPU: nvidia-l4

  • Machine type: g2-standard-4 or better

  • Image: Ubuntu 22.04 LTS

Setup

To launch the GCP virtual machine, use the following steps:

1. Search for GPU Zones with the NVIDIA T4 or L4 GPU model.

2. Create a Default VPC network.

3. Setup SSH connection to VM instances using a browser.

4. Follow the steps in Launch Cloud Shell to start Cloud Shell session on GCP.

5. Run the following command in the Cloud Shell session session to create a VM. Replace <project_name> and <instance_name>. The zone is set to us-central1-a, in this example, but can be replaced with the zones from step 1.

   T4

   $ gcloud compute \
   --project "<project_name>" \
   instances create "<instance_name>" \
   --zone "us-central1-a" \
   --machine-type "n1-standard-8" \
   --subnet "default" \
   --metadata="install-nvidia-driver=True" \
   --maintenance-policy "TERMINATE" \
   --accelerator type=nvidia-tesla-t4,count=1 \
   --image "ubuntu-2204-jammy-v20230919" \
   --image-project "ubuntu-os-cloud" \
   --boot-disk-size "100" \
   --boot-disk-type "pd-ssd"
   

   L4

   $ gcloud compute \
   --project "<project_name>" \
   instances create "<instance_name>" \
   --zone "us-central1-a" \
   --machine-type "g2-standard-4" \
   --subnet "default" \
   --metadata="install-nvidia-driver=True" \
   --maintenance-policy "TERMINATE" \
   --accelerator type=nvidia-l4,count=1 \
   --image "ubuntu-2204-jammy-v20230919" \
   --image-project "ubuntu-os-cloud" \
   --boot-disk-size "100" \
   --boot-disk-type "pd-ssd"
   

6. Follow the steps in Connect to Linux VMs using Google tools to connect to the VM.

7. Follow the steps in Install NVIDIA driver.

   $ curl https://raw.githubusercontent.com/GoogleCloudPlatform/compute-gpu-installation/main/linux/install_gpu_driver.py --output install_gpu_driver.py
   $ sudo python3 install_gpu_driver.py
   

8. See Container Installation to install the Docker and NVIDIA Container Toolkit.

9. Proceed to Container Deployment.