Data Center / Cloud

AWS Brings NVIDIA A10G Tensor Core GPUs to the Cloud with New EC2 G5 Instances

A round conference room with a sphere in the middle.

Today, AWS announced the general availability of the new Amazon EC2 G5 instances, powered by NVIDIA A10G Tensor Core GPUs. These instances are designed for the most demanding graphics-intensive applications, as well as machine learning inference and training simple to moderately complex machine learning models on the AWS cloud.

The new EC2 G5 instances feature up to eight NVIDIA A10G Tensor Core GPUs that are optimized for advanced visual computing workloads. With support for NVIDIA RTX technology and more RT (ray tracing) cores than any other NVIDIA GPU instance on AWS, it offers up to 3X better graphics performance. Based on NVIDIA Ampere Architecture, G5 instances offer up to 3X higher performance for machine learning inference and 3.3X higher performance for machine learning training, compared to the previous generation Amazon EC2 G4dn instances.

Customers can use the G5 instances to accelerate a broad range of graphics applications like interactive video rendering, video editing, computer-aided design, photorealistic simulations, 3D visualization, and gaming. G5 instances also deliver the best user experience for real-time AI inference performance at scale for use-cases like content and product recommendations, voice assistants, chatbots, and visual search.

Getting the most out of EC2 G5 instances using NVIDIA optimized software

To unlock the breakthrough graphics performance on the new G5 instances, creative and technical professionals can use the NVIDIA RTX Virtual Workstation (vWS) software, available from the AWS Marketplace. Only available from NVIDIA, these NVIDIA RTX vWS advancements include hundreds of certified professional ISV applications, support for all of the leading rendering apps, and optimization with all major gaming content. 

NVIDIA RTX technology delivers exceptional features like ray tracing and AI-denoising.  Creative professionals can achieve photorealistic quality with accurate shadows, reflections, and refractions—creating amazing content faster than ever before. 

NVIDIA RTX vWS also supports Deep Learning Super Sampling (DLSS). This gives designers, engineers, and artists the power of AI for producing the highest visual quality, from anywhere. They can also take advantage of technologies like NVIDIA Iray and NVIDIA OptiX for superior rendering capabilities.

Developers on AWS can use state-of-the-art pretrained AI models, GPU-optimized deep learning frameworks, SDKs, and end-to-end application frameworks from the NGC Catalog on AWS Marketplace soon. In particular, developers can take advantage of NVIDIA TensorRT and NVIDIA Triton Inference Server to optimize inference performance and serve ML models at scale using G5 instances. 

Developers have multiple options to take advantage of NVIDIA-optimized software on AWS. Whether you provision and manage the G5 instances yourself or leverage them in AWS managed services like Amazon Elastic Kubernetes service (EKS) or Amazon Elastic Container Service (ECS).

Learn more about the EC2 G5 instances and get started. >>

Discuss (0)