AWS GPU Spot Request

How to create an AWS Spot Request for a VM with an NVidia T4 GPU
0:00:00.000

1. Setting up AWS Spot Instance Request   (0:08:30.000)

▼
  • Initiated process to create a GPU-enabled VM on AWS
  • Accessed AWS console and navigated to spot instance request page
  • Began configuring spot instance request parameters
  • Selected Ubuntu server 22.04 as the Amazon Machine Image (AMI)
    • 0:00:00.000

  • 1.1. Initiating AWS Spot Instance Request   (0:02:57.960)

    • ▼
  • Accessed AWS console and logged in
  • Navigated to spot instance request page
  • Discussed previous attempt resulting in unintended fleet association
  • Began manual configuration of launch parameters
    • 0:02:57.960

  • 1.2. Selecting AMI and Creating Key Pair   (0:02:42.040)

    • ▼
  • Selected Ubuntu server 22.04 as the Amazon Machine Image (AMI)
  • Created a new key pair for SSH access
  • Downloaded and saved the key pair file for future use
  • Returned to spot fleet request page to continue configuration
    • 0:05:40.000

  • 1.3. Configuring Target Capacity and Maintenance Options   (0:02:50.000)

    • ▼
  • Set target capacity to 1 for a single instance
  • Discussed options for maintaining target capacity
  • Explored terminate, stop, and hibernate options
  • Chose 'stop' option as best for GPU instance development and testing
    • 0:08:30.000

    2. Configuring Instance Details and Security   (0:08:30.000)

    ▼
  • Selected network settings and availability zones
  • Chose appropriate GPU instance types (G5 and G6)
  • Configured security groups for the instance
  • Associated an Elastic IP address with the instance
    • 0:08:30.000

  • 2.1. Network Configuration and Instance Type Selection   (0:02:49.880)

    • ▼
  • Navigated to network settings
  • Encountered difficulties finding specific GPU instance types
  • Identified G5 and G6 instance types as suitable options
  • Selected instances based on spot price considerations
    • 0:11:19.880

  • 2.2. Finalizing Instance Selection   (0:02:50.120)

    • ▼
  • Refined instance pool selection to 12 options
  • Deleted unnecessary instance types
  • Prepared to launch the selected instance
  • Encountered issues with instance allocation
    • 0:14:10.000

  • 2.3. Associating Elastic IP   (0:02:50.000)

    • ▼
  • Associated an Elastic IP address with the instance
  • Verified the public IP address assignment
  • Attempted to ping the instance (unsuccessful)
  • Recognized need for security group configuration
    • 0:17:00.000

    3. Network Setup and Instance Type Selection   (0:08:30.000)

    ▼
  • Created and configured security groups
  • Troubleshooted network connectivity issues
  • Successfully established ping connection to instance
  • Prepared for SSH connection to the instance
    • 0:17:00.000

  • 3.1. Security Group Configuration   (0:02:50.000)

    • ▼
  • Created a new security group for the instance
  • Added necessary inbound rules to allow SSH and ping
  • Associated the security group with the instance
  • Verified security group assignment
    • 0:19:50.000

  • 3.2. Network Connectivity Troubleshooting   (0:02:50.000)

    • ▼
  • Reattempted to ping the instance
  • Successfully established ping connection
  • Prepared for SSH connection setup
  • Addressed known_hosts file configuration
    • 0:22:40.000

  • 3.3. SSH Key Configuration   (0:02:50.000)

    • ▼
  • Removed outdated entries from known_hosts file
  • Imported the new SSH key for the instance
  • Configured SSH profile for easy access
  • Prepared for initial SSH connection
    • 0:25:30.000

    4. SSH Connection and Instance Exploration   (0:06:03.240)

    ▼
  • Successfully established SSH connection to the instance
  • Explored instance specifications and resource usage
  • Configured remote development environment
  • Concluded setup process and prepared for handover
    • 0:25:30.000

  • 4.1. Initial SSH Connection and Resource Inspection   (0:02:01.080)

    • ▼
  • Successfully connected to the instance via SSH
  • Examined available memory and CPU resources
  • Investigated disk space usage and allocation
  • Noted discrepancies in reported vs. actual disk space
    • 0:27:31.080

  • 4.2. Remote Development Environment Setup   (0:02:01.080)

    • ▼
  • Configured initial directory for SSH sessions
  • Set up profile for quick access to the instance
  • Tested double-click login functionality
  • Prepared for handover to next phase of development
    • 0:29:32.160

  • 4.3. Conclusion and Handover   (0:02:01.080)

    • ▼
  • Confirmed successful setup of GPU-enabled AWS instance
  • Verified remote access and development environment
  • Prepared to transfer control to Bruce for next steps
  • Concluded the setup process and video recording