AWS introduces new PyTorch DLAMI for AWS Graviton-based instances such as Amazon EC2 G5g Instances

We are excited to announce the ARM64 PyTorch GPU Deep Learning Amazon Machine Images (DLAMI) optimized for the AWS Graviton-based Amazon EC2 G5g instances. The AWS ARM64 GPU DLAMIs are designed to provide high performance and cost efficiency for deep learning workloads, and is pre-configured with PyTorch and required dependencies such as NVIDIA driver, CUDA, CuDNN, NCCL, Docker, NVIDIA Docker etc. G5g instances are powered with AWS Graviton2 processor and NVIDIA T4G Tensor Core GPUs. These instances are designed as a cost-effective platform for machine learning inference and ideal for deploying deep learning applications that need access to NVIDIA GPUs and their associated AI libraries.

With the AWS ARM64 GPU DLAMIs and Amazon EC2 G5g instances, customers can now take advantage of the price and performance benefits of AWS Graviton to deploy GPU-accelerated deep learning models at a significantly lower cost when compared with non-Graviton-based instances with GPU acceleration.

To learn more, visit the ARM64 DLAMI developer guide. For a complete list of frameworks and versions supported, visit our DLAMI documentation.

Click here for ARM64 DLAMI Release Notes