Image acceleration helps speed up container image downloads and instance startup times.

Overview

When working with large container images, pulling from remote registries can be time-consuming. Image acceleration provides a caching layer to reduce download times.

Features

• Fast Pull: Cached images are pulled from local mirrors
• Global Cache: Multi-region caching for worldwide access
• Automatic: No configuration required for supported images

Supported Images

Common deep learning frameworks and tools are pre-cached:

• PyTorch
• TensorFlow
• CUDA runtime images
• Popular ML libraries

Usage

No additional action is needed. When you launch an instance with a supported image, the system automatically uses the accelerated cache if available.