Introducing three new NVIDIA GPU-based Amazon EC2 cases

Amazon Elastic Compute Cloud (Amazon EC2) accelerated computing portfolio affords the broadest alternative of accelerators to energy your synthetic intelligence (AI), machine studying (ML), graphics, and excessive efficiency computing (HPC) workloads. We’re excited to announce the enlargement of this portfolio with three new cases that includes the newest NVIDIA GPUs: Amazon EC2 P5e cases powered by NVIDIA H200 GPUs, Amazon EC2 G6 cases that includes NVIDIA L4 GPUs, and Amazon EC2 G6e cases powered by NVIDIA L40S GPUs. All three cases can be obtainable in 2024, and we sit up for seeing what you are able to do with them.

AWS and NVIDIA have collaborated for over 13 years and have pioneered large-scale, extremely performant, and cost-effective GPU-based options for builders and enterprise throughout the spectrum. We have now mixed NVIDIA’s highly effective GPUs with differentiated AWS applied sciences equivalent to AWS Nitro System, 3,200 Gbps of Elastic Fabric Adapter (EFA) v2 networking, a whole lot of GB/s of information throughput with Amazon FSx for Lustre, and exascale computing with Amazon EC2 UltraClusters to ship essentially the most performant infrastructure for AI/ML, graphics, and HPC. Coupled with different managed companies equivalent to Amazon Bedrock, Amazon SageMaker, and Amazon Elastic Kubernetes Service (Amazon EKS), these cases present builders with the business’s finest platform for constructing and deploying generative AI, HPC, and graphics purposes.

Excessive-performance and cost-effective GPU-based cases for AI, HPC, and graphics workloads

To energy the event, coaching, and inference of the most important giant language fashions (LLMs), EC2 P5e cases will characteristic NVIDIA’s newest H200 GPUs, which provide 141 GBs of HBM3e GPU reminiscence, which is 1.7 instances bigger and 1.4 instances sooner than H100 GPUs. This increase in GPU reminiscence together with as much as 3200 Gbps of EFA networking enabled by AWS Nitro System will allow you to proceed to construct, prepare, and deploy your cutting-edge fashions on AWS.

EC2 G6e cases, that includes NVIDIA L40S GPUs, are constructed to offer builders with a broadly obtainable possibility for coaching and inference of publicly obtainable LLMs, in addition to help the rising adoption of Small Language Fashions (SLM). They’re additionally optimum for digital twin purposes that use NVIDIA Omniverse for describing and simulating throughout 3D instruments and purposes, and for creating digital worlds and superior workflows for industrial digitalization.

EC2 G6 cases, that includes NVIDIA L4 GPUs, will ship a lower-cost, energy-efficient resolution for deploying ML fashions for pure language processing, language translation, video and picture evaluation, speech recognition, and personalization in addition to graphics workloads, equivalent to creating and rendering real-time, cinematic-quality graphics and recreation streaming.