Skip to main content

BCR Registry User Guide

Public Registry

Overview

The Public Registry in Bitdeer AI Container Registry (BCR) provides seamless access to high-quality, enterprise-grade container images, including those optimized for AI workflows. With frequent synchronization from NVIDIA NGC, users gain access to industry-leading resources designed for advanced AI development and deployment.

Key Features

  1. Access NVIDIA Enterprise-Grade Containers:
    • Includes containers for frameworks such as TensorFlow, PyTorch, and RAPIDS.
    • Enterprise-grade tools like NVIDIA NeMo for conversational AI and NVIDIA TensorRT for high-performance inference.
    • Pre-optimized containers designed for AI training, inference, and HPC workloads.
  2. Direct Execution via Bitdeer AI Container Service:
    • Users can directly execute public containers through the Bitdeer AI Container Service.
    • No manual pulling of images is required, ensuring secure and controlled usage.
  3. Centralized Management:
    • Manage and search for NVIDIA NGC images alongside other public containers from a single, unified dashboard.

Using the Public Registry

Steps to Browse and Execute Public Containers

  1. Log in to AI Studio:
    • Access the AI Studio Console.
  2. Navigate to Public Registry:
    • In the left-hand menu, select Containers & Registry > Public Registry.
  3. Browse NVIDIA NGC Containers:
    • Use the search bar to find NVIDIA containers by name or keyword (e.g., TensorFlow, PyTorch).
    • Apply filters to narrow results by framework, version, or use case.
  4. View Image Details:
    • Click on a container name to view detailed metadata, including:
      • Framework version.
      • Supported GPU types.
      • Pre-installed libraries and dependencies.
  5. Run the Container via Bitdeer AI Container Service:
    • From the container details page, click Run with Container Service.
    • Follow the prompts to configure and deploy the container directly on the Bitdeer AI platform.

Benefits of NVIDIA Enterprise-Grade Containers

  1. Optimized Performance:
    • Pre-configured for NVIDIA GPUs, ensuring maximum hardware utilization.
  2. Scalability:
    • Designed to scale across multiple GPUs for distributed training and high-performance inference.
  3. Security and Reliability:
    • Regular updates from NVIDIA ensure the latest patches, features, and optimizations.
  4. Enterprise-Ready:
    • Certified and tested by NVIDIA for production-grade applications.

Best Practices

  1. Stay Updated:
    • Regularly check for new versions of NVIDIA containers to benefit from the latest optimizations and features.
  2. Match Workloads to Containers:
    • Select containers that align with your AI workloads, such as NeMo for conversational AI or RAPIDS for data science workflows.
  3. Leverage Documentation:
    • Refer to NVIDIA NGC documentation for container-specific details and usage guidelines.