Skip to main content

Model Deployment Overview

The Model Deployment page provides an intuitive interface for users to deploy machine learning models by selecting the appropriate number of GPUs and GPU configurations. The page includes details on hardware specifications, pricing breakdowns, and resource management to ensure smooth deployment of models for various machine learning tasks.

Key Features:​

  1. Number of GPUs:

    • Users can select the number of GPUs required for deployment. The available options include:
      • 1X: Single GPU
      • 2X: Double GPU
      • 4X: Quad GPU
      • 8X: Eight GPUs
    • The number of GPUs selected affects the speed and efficiency of model inference or training. More GPUs generally offer higher processing power and speed but come with an increased cost.

  2. Price Breakdown:

    • Users can view a detailed price breakdown for GPU usage, disk space, and total costs based on the selected configuration.
    • Example: In the case of using 2X GPUs, the total cost includes:
      • GPU On-Demand rate (e.g., 0.28 USD/hr)
      • Disk usage cost (e.g., 16 GB disk usage at 0.00 USD/hr)
      • Total daily and monthly costs are provided to help users estimate the deployment expenses efficiently.
    • Users can directly proceed with the Rent Now button to initiate GPU usage for the deployment.

  3. Detailed GPU Specifications:

    • The platform provides comprehensive details for each selected GPU:
      • GPU Name: Shows the model of the GPU (e.g., 1x A40).
      • GPU RAM: Displays the available GPU RAM, crucial for large models.
      • CPU Information: Number of cores, CPU name, and architecture.
      • Memory Bandwidth: Shows the available memory bandwidth, which affects data transfer speeds.
      • Max CUDA: Specifies the maximum number of CUDA cores for parallel processing.
      • Ports and Disk Space: Information about available ports and disk space for storage.
      • Reliability and Internet Speeds: Highlights the system's reliability percentage and download/upload internet speeds.

  4. Plan Selection and Deployment:

    • After reviewing the GPU details, users can click the Plan button to finalize the GPU and deployment settings.
    • Once ready, click the Deploy button on the top right of the page to initiate the deployment process.
    • The deployment page also provides real-time feedback and cost estimation based on GPU resource selection.

How to Use:​

  1. Step 1: Choose the number of GPUs from the options (1X, 2X, 4X, or 8X).
  2. Step 2: View the detailed price breakdown and specifications for the selected GPU.
  3. Step 3: Click on the Plan button to confirm the GPU setup.
  4. Step 4: Click Deploy to begin the model deployment process.

The Model Deployment page offers a streamlined experience for deploying machine learning models, allowing users to manage both hardware resources and costs effectively.