Skip to content

Google Cloud - Attach GPU to VM Instance

Step-01: Introduction

  • Create a VM with GPU attached to it

Step-02: Request for GPU Quota

  • Go to IAM & Admin -> Quotas & System Limits
  • Go to Compute Engine API -> Search for GPUS_ALL_REGIONS
  • Select **GPU (all regions) -> EDIT QUOTA
  • New Value: 1
  • Click on DONE
  • Click on SUBMIT REQUEST
  • Verify email about approval request

Step-03: Create a VM Instance with attached GPUs

  • Go to Compute Engine -> VM Instances -> CREATE INSTANCE
  • Name: gpu-vm1
  • Region: us-central1
  • Zone: us-central1-b
  • Machine Configuration: GPU
  • GPU Type: NVIDIA T4
  • Number of GPUs: 1
  • BOOT DISK: SWITCH IMAGE
  • Note: The selected image requires you to install an NVIDIA CUDA stack manually. To skip manual setup, click "Switch Image" below to use a GPU-optimized Debian OS image with CUDA support at no additional cost.
  • REST ALL LEAVE TO DEFAULTS
  • click on CREATE

Step-04: Review VM Details

  • Go to Compute Engine -> VM Instances -> gpu-vm1 -> DETAILS TAB
  • Review Machine configuration for GPU
  • Review Boot disk, we will find deeplearning Image

Step-05: Clean-Up

  • Go to Compute Engine -> VM Instances -> gpu-vm1 -> DELETE
🎉 New Course
Ultimate DevOps Real-World Project Implementation on AWS
$15.99 $84.99 81% OFF
DEVOPS2026FEB
Enroll Now on Udemy
🎉 Offer